The Research of Complex Product Design Process Model under the Concept of Self-Recovery

Abstract

The working environment of contemporary mechanical products is becoming more complex, and the working conditions are becoming more extreme. This has led to a significant increase in the frequency of problems in mechanical products. In order to reduce the frequency of human repair after problems, the application of the self-recovery concept has become a hot research topic in the area of smart design. However, the current application of the self-recovery concept is mostly limited to the structural and parametric levels, with less research at the functional level, which may lead to a waste of resources within products. To solve this problem, this research combines the functional-level product research method with the self-recovery concept and establishes a design process model of complex products under functional self-recovery. This model extends the application scope of the self-recovery concept and improves the efficiency of resource utilization in the product. The design process model has six steps. First, according to the user requirements and the existing product, the initial function solving is carried out, and the initial function model of the product is established. Next, the main functions of the product are determined based on the initial function model of the product. Then, according to the determined main functions of the product, combined with the parameters marked in the function structure, the self-diagnosis function is designed. After that, the LT matrix and effect library are used to design the self-regulation function corresponding to the main functions, and the parameters are used to screen the self-regulation function design scheme. Finally, according to the design scheme of the self-diagnosis function and self-regulation function, the functional period oriented to self-recovery is constructed to ensure the realization of the main functions of the product. The effectiveness of the design process model is proved through the design process of an intelligent photovoltaic power generation system at the end of the paper.

1. Introduction

As modern mechanical products become larger and more automated, their complexity continues to increase. At the same time, the complex working environment and extreme working conditions lead to a sharp increase in the possibility of product problems. To reduce human involvement in maintenance, Gao, a member of the Chinese Academy of Engineering, has expanded the traditional five-block theory of electromechanical products [1] by considering the bionic design to transfer the self-recovery functions of humans and animals to machines, which can be used to achieve self-recovery of product problems and maintain long-term and safe operation of mechanical products.

The current application of the self-recovery concept is mostly focused on the structural and parametric levels, while there is less research at the functional level. Axiomatic design theory [2] points out that structural elements are mapped from functional elements. Through the design matrix, a designer can judge whether the design is an ideal design and improve the design. Analyzing the product at the functional level is an effective way to reduce the design process complexity of the product and reduce the workload at the structural and parametric levels. At the same time, there are a large number of problem analysis tools at the functional level, such as function models [3], function structures [4], resource analysis [5], and conflict area analysis [6], and a large number of problem-solving tools, such as effect libraries [7], standard solutions and trimming [8].

Currently, functional-level product research methods are not well integrated with the self-recovery concept, which may lead to the wastage of functional-level resources within the product, resulting in increased difficulty in subsequent structural and parametric design.

To relieve the limitations of the current research on the self-recovery concept, this paper extends the application of the self-recovery concept to the functional level, introduces the function classification criteria to the complex product design process under the self-recovery concept, and proposes the criteria for judging the self-recovery priority of product functions. In addition, effect library software is introduced into the self-regulation function design process for predicting solutions to product problems and screening solutions to product problems by marking parameters in the function structure. After extending the application of the self-recovery concept to the functional level, the resources within the product can be maximized and the ideas of complex product design under the self-recovery concept can be broadened. Finally, a design process model of complex products under the concept of self-recovery is established to provide designers with more perspectives to realize product self-recovery.

2. Relevant Research

2.1. Research on the Concept of Self-Recovery

With the development of the self-recovery concept, Yang et al. [9] proposed a hybrid energy structure composed of solar cells and self-recovery nano-generators to solve the problem of reduced output capacity due to the potential mechanical damage of solar cells. Pan et al. [10] used a liquid-transfer active balancing device to build a targeting self-recovery regulation system which realizes imbalance fault self-recovery at the structural level by injecting compressed air into the targeting chamber to change the mass distribution of the balancing disc and rotor. Wang et al. [11], based on the reliability principle and self-recovery concept, proposed a multiple therapy target fault self-recovery regulation method for an electro-hydraulic control system, where the occurrence of displacement faults of the catalytic cracking unit in the refinery was reduced at the structural level by designing an adjustable actuator. Wu et al. [12] reasoned the situation change in a smelting furnace according to the fluctuation rate and duration of the raw material granule size in the smelting furnace by adjusting the current parameters between electrodes. Yao et al. [13] proposed a control method for rotor multi-frequency cycle vibration in rotor-bearing systems, which suppresses rotor multi-frequency vibration through parameter changes to achieve rotor self-seeking optimal control of vibration. Chen et al. [14] identified unbalanced vibrations with the help of the axial trajectory method and used a multi-directional vibration vector as the self-balancing control signal, where the unbalanced vibration was offset by parameter adjustment.

At present, the self-recovery concept is researched and applied extensively at the structural and parametric levels. Gao et al. [15] summarized the current research on the concept of self-recovery and formally proposed the theory of artificial self-recovery. Artificial self-recovery theory includes different technical fields, which are divided into two categories: parameter adjustment and structural adaptation, meaning that the system will be guarded against deviating from normal operation by regulating parameters, or the product will be self-adapted to variable operating conditions by changing the product structure, stiffness, and damping distribution.

At the structural and parametric levels, the research on the self-recovery concept has formed a relatively well-developed theoretical system. However, in the design stage, there are fewer research achievements in applying the self-recovery concept at the functional level.

2.2. Research of Complex Products at the Functional Level

Mechanical products designed under the self-recovery concept usually need to operate under extreme working conditions, and there will be non-linear, coupling, openness, time-varying, and other complex product features between various physical processes in the product. At the same time, in axiomatic design, Suh [2] proposed that product analysis from the functional perspective is an effective method for complex product design; therefore, the product analysis method and solution method at the functional level can be applied to the complex product design process under the self-recovery concept.

Li et al. [16] combined physical laws with function decomposition, subdivided product requirements through function units, and used physical laws to assist function unit solutions, thus improving the analysis speed and design accuracy of complex products. Liu et al. [17] quantified the similarity between each function unit and established the relationship between user requirements and each function unit through function decomposition and functional analogy, which accelerated the generation of conceptual solutions for ideal product principal solutions. Using function decomposition to decompose the total function of complex products into function units is the key to analyzing products from the functional perspective and reducing the difficulty of product design.

Wang et al. [18] organically combined the function units obtained by function decomposition with flows to establish the product function structure. The changes in product parameters were expressed in the function structure by flow attributes, and a model for analyzing and solving multi-flow problems of complex products was constructed. Pu et al. [19] analyzed the parameter changes in the product through the changes in flows in the function structure and established the process model of implicit conflict identification of complex products in combination with failure analysis to solve the problem of the strong latency of implicit conflict in complex products. By marking the flow in the function structure, the process of parameter change from raw materials to goods can be clearly expressed, and the functions that have problems in the products can be analyzed by detecting parameter changes.

Wang et al. [7] transformed the dimension of parameters in the product into the length and time dimension (LT parameter), and the genetic idea of the length and time dimension matrix (LT matrix) was applied to the product problem-solving process, combined with an effect library to produce a disruptive and innovative design solution for complex products. Cao et al. [20] built an effect chain reasoning process guided by idealized physical quantity resources through the connection between the LT dimension and physical effects, where the generation of high-level innovative solutions was accelerated by using an LT matrix. Using the connection between the LT dimension and effect and combining the genetic ideas in the LT matrix with effect library software for solving product problems will speed up the solution of complex products and improve the efficiency of complex product design.

As can be seen, scholars all over the world have conducted a lot of research at the functional level and put forward relatively perfect problem analyses and solutions. By combining the self-recovery concept with product analysis methods and solution methods at the functional level and extending the application of the self-recovery concept to the functional level, intelligent product design will be carried out from more angles and the design ideas of complex products will be broadened.

3. The Principle of Functional Self-Recovery

Functional self-recovery is a supplement to the current application of the self-recovery concept, from the functional perspective to maintain the operation status of the product as a starting point. Based on the principle of avoiding the introduction of new components as much as possible, the design goal is to realize the self-diagnosis function and self-regulation function for the main functions of the product through the existing components in the product.

3.1. The Principle of Complex Product Function Category Classification

Compared with parameters and structures, functions have unique classification criteria. Using functional value as a criterion, only one category of function units obtained from function decomposition is the main function [21], which is the purpose of the product’s existence. The second category is the additional function [21]. The additional function is the function that supports the main function of the product. Problems with different categories have different impacts on the operation of the product. Problems with the main function will directly cause the product to stop running, while problems with additional functions will affect the efficiency of the product. For example, with the development of electronic devices, additional functions of the electric water heater such as “reducing energy consumption” and “timing shutdown” have emerged. However, for users, the main function of the electric water heater is always “improve water temperature”. When the “improve water temperature” function fails, the product will stop running directly. When the “timing shutdown” or “reduce energy consumption” function fails, the energy consumption of the electric water heater will increase, but the product will still run. Therefore, if the self-recovery concept is applied to the design process of the electric water heater, the first target is to ensure the realization of the function “improve water temperature”.

Self-recovery regulation for all functions in the product is the ideal result in the design stage but adding self-diagnosis and self-regulation functions to all functions in the product will significantly increase the design cost and complexity in the later manufacturing stage. Therefore, it is necessary to classify the functions in the product, prioritize the resources allocated for the main functions of the product, and arrange a reasonable self-recovery priority.

3.2. The Principle of Self-Diagnosis Function Design

The self-diagnosis function is the section that gives the product the ability to diagnose problems in the design stage. The self-diagnosis function has two main roles in the product design process under the self-recovery concept: first, to detect whether the function has problems; second, to initiate the self-regulation function after diagnosing the product problems. A reasonable self-diagnosis function is a necessary link to construct a functional period oriented to self-recovery and is the key to realizing self-recovery regulation.

Function expresses the relationship between the input parameters and output parameters of a technical system, which is generally expressed in the form of a verb and a noun [22]. The verb expresses the operation completed by the product, and the noun represents the operation object, in which the noun is measurable. When there are problems with a function, the output parameters of the function will be changed, and by detecting the change in the parameters of the function, it can be judged whether there are problems with the function. Some functions’ noun description is parameter information, where the noun can be directly confirmed as the diagnosis object. However, the noun description of some functions hides the parameter information, and thus the designer needs to identify the specific parameter information of the function according to the function’s working environment before diagnosing. For example, in the function “generate pressure”, the verb is “generate”, and the noun is “pressure”; the pressure itself is the parameter information, so it is feasible to diagnose problems with the function “generating pressure” by detecting the pressure. As for the function of “transport goods”, the noun is “goods”, the designer needs to identify the parameter of the noun “goods”, which may be “height” or “location”, or other parameters according to the environment of the function. After confirming the parameter hidden in the noun, the designer will find the corresponding detection parameter.

As there may be a coupling between functions, when the main function of the product has problems, the output parameters of its coupled functions will also be changed. When the complexity of the product is high and it is difficult to diagnose the main function directly, the coupling between functions can be used to diagnose the main function indirectly by diagnosing the functions coupled with the main function.

In order to visually express the parameter changes within the product from raw materials to goods and to accelerate the efficiency of self-diagnosis function design, the product function structure could be established by combining the function units obtained by the function decomposition in an orderly manner. The changes in the flow attributes of the function structure could be marked to express the functional output parameters.

3.3. The Principle of Self-Regulation Function Design Based on LT Dimension and Effect

The self-regulation function is the specific execution link of self-recovery regulation with the ultimate goal to maintain the operation of the main functions. When deciding the form of self-regulation function realization, the first step is to find the existing products that were referenced when designing new products. Through the existing product problems, potential problems in new products can be predicted. After clarifying the problem, work can continue on the self-regulation function design. The essence of the function can be expressed through the effect. When it is difficult to determine the form of self-regulation function realization directly, the conversion relationship between effect and function can be used to assist the realization form of self-regulation function determination.

The process of effect realization is the process of parameter transformation. An effect library is a summary of the scientific effects found at present. Through using effect library software, it will realize the retrieval of effects by parameter and function. Due to the non-uniformity of the product parameter dimension, it is difficult to directly determine the effect that realizes the self-regulation function, and it is also difficult to determine the effect that solves the problem. To reduce the difficulty of effect retrieval, the dimension will be transformed into the LT dimension, which uses the genetic idea [23] in the LT matrix for effect retrieval. The LT matrix is obtained by combining the LT dimensions, and part of the LT matrix is shown in

Table 1. Common LT matrix [24].

Time	Length
	L−2	L−1	L0	L1	L2	L3	L4	L5
T−5					L2T−5	Surface power	L4T−5	Power Temperature variety rate
T−4			L0T−4	Specific gravity Gradient of pressure	Pressure Strain Normal stress Elastic modulus	Surface tension Rigidity	Force Temperature gradient	Temperature Energy Force moment
T−3			L0T−3	Current Density	Electromagnetic field strength Ductility Dynamic viscosity	Current Loss mass	Momentum Impulse	Angular moment Angular impulse
T−2		L−1T−2	Mass density Angular acceleration	Acceleration Magnetic displacement Magnetic reluctance	Voltage	Mass Quantity of electricity	Magnetic flux	Moment of inertia
T−1	L−2T−1	Volume charge density	Frequency	Velocity	Velocity of change of the area	Loss volume	L4T−1	L5T−1
T0	L−2T0	Conductance	Dimensionless quantity	Length	Surface absorption	Volume of space	Moment of inertia of an area
T1	L−2T1	Conductivity	Period Duration	L1T1	L2T1	L3T1
T2	Magnetic permeability	L−1T2	L0T2	L1T2	L2T2

. In the table, the horizontal rows contain a series of integer subdivisions L, and the vertical rows contain a series of integer subdivisions T. The physical parameters in the table are expressed using the multiplication result of length L and time T, LⁿT^m. By logical multiplication of dimension x and dimension y with different technical characteristics in the LT matrix, the new LT dimension obtained will become the resource available for the retrieval effect [23].

When applying the LT matrix with the effect library software to confirm the effect that realizes the self-regulation function, the first step is to describe the product problem, find the physical parameter corresponding to the problem, and convert the physical parameter into an LT parameter, recorded as L^a1T^b1. Then, a semi-effective solution can be found to assist in solving the product problem, which is the conceptual solution generated when analyzing the problem without considering the constraints. The physical parameters corresponding to the semi-effective solution are similarly transformed into LT parameters and recorded as L^a2T^b2. According to the genetic idea in the LT matrix [23], by multiplying the two LT parameters, as shown in Formula (1), the new parameters L^(a1+a2)T^(b1+b2) obtained will inherit the properties of the semi-effective solution with the possibility of solving the problem while inheriting the properties of the problem, making the new parameters more oriented to the problem itself.

$$\left({\mathrm{L}}^{\mathrm{a}1}{\mathrm{T}}^{\mathrm{b}1}\right)\cdot \left({\mathrm{L}}^{\mathrm{a}2}{\mathrm{T}}^{\mathrm{b}2}\right)={\mathrm{L}}^{(\mathrm{a}1+\mathrm{a}2)}{\mathrm{T}}^{(\mathrm{b}1+\mathrm{b}2)}$$

(1)

After obtaining the new LT parameters, the new LT parameters are reversed into physical parameters by the LT matrix, and then, the effect library is used to retrieve the effects to realize the self-regulation function. The design principle of the self-regulation function based on the LT dimension and effect is shown in Figure 1.

3.4. The Principle of the Functional Period Design Oriented to Self-Recovery

A functional period is a necessary characteristic of a stable product, specifically defined as a group of functional requirements that are repeated in the base cycle. A reasonable functional period will weaken the time-varying characteristics of the product and reduce the complexity of the product [25]. As the design goal under the self-recovery concept is to guarantee the safe and long-term running of the product, constructing a functional period in the product is the key to making the complexity change periodically, enhancing the intelligent level of the product, and realizing functional self-recovery.

In the process of constructing a functional period in the product, the product complexity needs to be extracted first. Additionally, the product complexity needs to be analyzed through one or more substance-field analyses, 39 standard engineering parameters, functional analysis, and laws of technological system evolution. After finishing the analysis, the corresponding function can be found using complexity features. If the product still has high complexity, the product complexity features can be converted into a standard problem, and the function of reduced complexity can be obtained using 76 standard solutions in TRIZ, the conflict resolution matrix, effects, and technology forecasting [26].

After obtaining the function, the functional period will be built according to the relationship of the function actions in the product. If the functional period is able to satisfy the functional requirements for the normal operation of the product, the function can be directly transformed into a functional period. If it is unable to satisfy the product operation requirements but the complexity has been reduced, the complexity existing in the product can be extracted again, the function can be found, and iteration can be performed until the product functional period satisfies the functional requirements for the normal operation of the product. After building the functional period for the normal operation of the product, based on the normal operation functional period, the functional requirements of the product in self-recovery regulation mode can be analyzed, self-diagnosis and self-regulation functions can be added, and iteration can be performed again until the functional period oriented to self-recovery is able to satisfy the functional requirements of self-recovery regulation. When the product is working properly, the initial functional period will be running, and when the product has problems, the functional period oriented to self-recovery will be running. The principle of self-recovery-oriented functional period construction is shown in Figure 2.

4. The Design Process Model of Complex Products under Functional Self-Recovery

Complex products under extreme operating conditions are difficult to repair by humans. To reduce human involvement at the later stage and improve the product’s intelligence level, a reasonable self-recovery process should be designed to maintain the safe and long-cycle operation of the product. Details of the application of the self-recovery concept to the complex product design process and the proposal of the complex product functional self-recovery design process model are described as follows.

Firstly, decompose the function of the new product, divide the product function units, find the principal solution to realize function units, establish the initial function model of the new product according to the new product function requirement and existing product function model, and confirm the correspondence between function units and components. Secondly, according to the correspondence between function units and components, calculate the value of different functions in the function model using a functional-level algorithm, divide the function categories, and identify the main functions of the product. Thirdly, establish the new product function structure, mark the output parameters of the main functions in the function structure, identify the functions coupled with the main functions in the product, and confirm the self-diagnosis function position and self-diagnosis form of the main functions. Next, using the LT matrix to predict the parameters to realize the self-regulation function, search the effects through the parameters in the effect library software to obtain the effects that realize the self-regulation function, combine the parameters in the function structure to select effects, and then derive the conceptual solution to realize the self-regulation function. After that, with reference to the conceptual solution of functional self-recovery, according to the functional value ranking, the components corresponding to the lower-value function units in the product will be used to execute the self-diagnosis function and self-regulation function for the main functions in the product. Finally, a functional period oriented to the self-recovery of the product will be established to reduce the complexity of the product. When the product has a problem or a trend of problem occurrence, it transitions from normal working mode to self-recovery regulation mode to maintain the safe and long-cycle operation of the main function in the product. The overall process is shown in Figure 3.

4.1. The Initial Function Solving and Initial Function Model Construction for Complex Products

When confirming the corresponding product function–component relationship, the product function needs to be decomposed first, then the product’s function units need to be obtained. The initial principal solutions (component) corresponding to the function units need to be found. The new product development process is based on existing products, and most of the principal solutions corresponding to the functions of the new product also exist in the existing products. When it is difficult to find the principal solution of the function directly, by finding the same function in existing products, the principal solution of the existing product’s function can be introduced to assist in finding the principal solution of the new product function.

As introduced in Section 3.2, function units express the relationship of an action and a target through a verb and a noun, as expressed in Formula (2):

$$V+N\to A+T$$

(2)

where V is the verb, N is the noun, A is the action, and T is the action object, and the execution object of A can be obtained by finding the principal solution for the function unit. Therefore, after finding the principal solution, the function unit can be extended to Formula (3):

$$f=\{EO,A,T\}$$

(3)

where EO is the execution object, A is the action relationship, and T is the action object. This description of the function is the same as the method of expressing the action relationship between components in the function model. Therefore, the product function model is available to be established after finding the function unit’s principal solution, using the function model to calculate the value of different functions.

When building the function model of a new product, it is difficult to directly establish a function model of the new product by combining components, so function models of existing products that were referenced in finding the principal solution of function units can be introduced to assist in establishing the function model of the new product. Most of the new product development process is based on a single existing product, according to user requirements, when carrying out new product design. When analyzing this type of product, the initial function model of the new product can be established directly based on the principal solution results and the function model of the existing product. A small number of new products are developed based on multiple existing products, combining the functions of multiple existing products to design the new product. When analyzing this type of product, according to the principal solutions corresponding to the function units and the relationship between existing product components, the function models of several existing products can be combined through the trimming algorithm in TRIZ to establish an initial function model of the product [27]. After establishing the new product function model, the relationship of components in the function model can be expressed using Formula (3), and then the correspondence between components and functions can be confirmed with Formula (2).

4.2. Self-Recovery Prioritization of the Complex Product Function

Calculating the functional value of the product and confirming the main functions of the product are the key steps in identifying the self-recovery target and establishing the self-recovery priority. The functional level is a common criterion in functional value evaluation, and the value of the function is related to the connection relationship among components, products, and super-systems in the function model. After continuous development and modification by scholars [28,29], the rules of the functional level definitions are as follows:

(1)

If the component acts directly on the product, the functional level of its role is the main function, expressed by $A_0$.

(2)

If the component acts on the component that realizes the main function, the functional level of the component is a first-level additional function, expressed by $A_1$.

(3)

If the component acts on the component that realizes the $i-1$ level of additional function, the functional level of the component is an $i$-level additional function, expressed by $A_i$.

(4)

When the component acts with the super-system, the functional level of the component is expressed by $A_1$.

(5)

Set the function value of the lowest functional level in the product as 1.

(6)

RANK($A_{i-1}$) = RANK($A_i$) + 1.

(7)

RANK($A_0$) = RANK($A_1$) + 2.

(8)

For the function that acts on multiple functional components, the level is the sum of all the actions.

The rules of the functional level calculation are shown in Figure 4. After completing the functional level calculation of all components in the function model and sorting them by level, the function value in the product can be sorted according to the correspondence between the component and function. Then, the main functions of the product can be identified, and the priority to realize functional self-recovery can be established.

4.3. Self-Diagnosis Function Design Oriented to Self-Recovery

After confirming the function self-recovery priority of the product, the product function structure can be established according to the result of product function decomposition and the function structure of the existing product, expressing the transformation relationship of substance, energy, and signal in the product. Mark the main functions of the product in the function structure, analyze the functions coupled with the main functions, and summarize the output parameters of the main function and the function which is coupled with the main function. Find out the functions to diagnose different parameters and determine whether the existing components in the product can detect the output parameters of the main function and the function which is coupled with the main function; if there is a corresponding component, the self-diagnosis function will be realized by the existing components in the product. If the corresponding components do not exist in the product, add functions to diagnose different parameters and find the principal solutions of the functions. The principal solution with the lowest solution complexity will be selected first.

4.4. Self-Regulation Function Design Oriented to Self-Recovery

After completing the design of the self-diagnosis function, according to the product function requirements, carry out the design of the self-regulation function. First, determine the causes of the problems based on existing products, describe the problems, and find the semi-effective solution to solve the problem. After that, extract the physical parameters hidden in the description of the product problem and the semi-effective solution, convert the physical parameters to LT parameters, and then derive the new LT parameters through Formula (1). Reverse this LT parameter into a physical parameter, select the parameter-effect retrieval mode in the effect library to retrieve this parameter, and record the effect set as $A$.

To reduce the results of the parameter-effect mapping caused by physical parameters under the same LT scale and limit the range of effect selections, combine the effect library function-effect retrieval mode to confirm the effect that realizes the self-regulation function. The function-effect search is carried out in the following two ways: first, with the help of the standard function set [30] to retrieve an additional function, by retrieving the effect to realize the additional function to eliminate the cause of the problem, the result of which will be recorded as $B$. Second, using the standard function set to search the main function itself to find the effect of realizing the main function of the product, the result of which will be recorded as $C$. The sets obtained by the two retrieval methods are taken as a union, which is recorded as the function-effect retrieval set, $B\cup C$. After obtaining the parameter-effect retrieval set A and function-effect retrieval set $B\cup C$, the two are recorded as $D$ by taking the intersection set, $D=A\cap \left(B\cup C\right)$. Set $D$ is the effect that can be used to realize the self-regulation function, as shown in Figure 5.

After obtaining the effect set $D$ for the realization of the self-regulation function, the input parameters required for the realization of the effect are recorded as set $E$. Record the parameter inside the product which is marked in the function structure as set $F$. Take the intersection of set $E$ and set $F$ to obtain set $M$, $M=E\cap F$. When set $M$ is not empty, the effect in set $D$ corresponding to the parameters in set $M$ is the effect that can be used to realize the self-regulation function, which uses the parameters in the product, recorded as set $N$, as shown in Figure 6.

If set $N$ is empty, starting from the output of the effect in set $D$, reverse the deduction to compare set $E$ with set $F$, and select the effect with the highest degree of overlap (if a high degree of overlap is found in the later analysis, but the feasibility is not strong, select the one with the next highest degree of overlap). The degree of overlap calculation method is shown in Formula (4):

$$W=\frac{f}{n}\times 100\%$$

(4)

where $W$ is the degree of overlap, $n$ is the input parameter required for effect realization, and $f$ is the parameter existing in the product that overlaps with the input parameter required for effect realization.

After selecting the effect with the highest degree of overlap, analyze the missing input parameters for the effect realization. Take the input parameters as the output parameters to retrieve the effect in the effect library; if the product contains the required input parameters for the new retrieved effect realization, combine effects as an effect chain. Take the effect chain as the conceptual solution for the self-regulation function realization. If the required input parameters for the new retrieved effect realization are missing in the product, take the missing input parameters as the output parameter again for the effect retrieval until all the input parameters required for the effect realization in the product and the effect chain can output the target parameters, as shown in Figure 7. The effect chain formed in this way will be the conceptual design for the realization of the self-regulation function. In the process of retrieving the effect to realize the self-regulation function, if the overlap of all effects is 0 in one link, the function decomposition needs to be carried out again to establish a new function structure and add the parameters required for effect realization in the new function structure.

4.5. Complex Product Function Resolution Oriented to Self-Recovery

The results of the principal solution corresponding to the function are usually not the only results, and different principal solutions will be selected from different perspectives. To improve the design efficiency, the principal solution in the existing product is introduced as a reference in the initial function solving. After combining the self-recovery concept, it is necessary to resolve some functions and reselect the principal solution, making the principal solution realize the self-diagnosis and self-regulation functions based on realizing the original function.

Before resolving the function, analyze whether the self-diagnosis function and self-regulation function of the product can be realized through the existing principal solution in the product according to the realization form of the self-diagnosis function determined in Section 3.3 and the realization effect of the self-regulation function determined in Section 3.4. If the two functions can be realized, there is no need to resolve the two functions. If they cannot be realized, the principal solution of some functions in the product will be selected again that combine with the function output parameters marked in the function structure. The selected principal solution executes the original function when the main function is in normal operation and executes the self-regulation function when the self-diagnosis function detects problems in the main function, forming a complex product design scheme under the self-recovery concept.

4.6. Functional Period Construction Oriented to Self-Recovery

After the product function solving under the self-recovery concept is completed, it is necessary to establish the product functional period, introduce the function for the position with high complexity in the product, and reduce the product complexity. After introducing the function for the high-complexity position and combining the function solving results oriented to self-recovery, rebuild the product function model, recalculate the functional value in the product, and judge whether the high-value functions required by the product operation have changed. If not, continue to establish the complex product functional period oriented to self-recovery. If there is any change, enter the functional self-recovery design process model again, iterate repeatedly until the high-value functions required by the product operation do not change, and establish a complex product functional period oriented to the self-recovery according to the iteration results.

Constructing the functional period in the product will reduce the complexity of the product, on the one hand, and express the operating state of the product under different working conditions, on the other hand. First, according to the functional relationship between the principal solution and the function execution sequence information contained in the function structure, construct the initial functional period of the product. Then, according to the function solving results oriented to self-recovery, change the execution mode and execution time of the principal solution to construct the functional period oriented to the self-recovery of the product. When the main functions of the product do not have problems, the initial functional period of the product will be selected. When the main functions of the product have problems, the functional period oriented to self-recovery will be selected to maintain the operation of the main functions of the product, as shown in Figure 8.

5. Engineering Examples

With the global warming trend becoming more and more obvious, the use of clean energy has become the focus of national research. Solar power generation technology, also known as photovoltaic power generation technology, is one of the important sources of clean energy. By the end of June 2022, the Chinese cumulative installed capacity of photovoltaic power generation reached 340 million kilowatts. At the same time, European and American countries are also increasing the target of the new PV installed capacity. It is estimated that from 2022 to 2025, the annual average new PV installed capacity will reach 232 million kilowatts to 286 million kilowatts [31]. Most photovoltaic power generation systems are located in the mountains and Gobi Desert with abundant solar energy resources. The environment is extreme, and human activities are rare. When the product has problems, it is difficult to repair manually. Therefore, it is necessary to improve the intelligent level of the current photovoltaic power generation systems, integrate the self-recovery concept into the design process of photovoltaic power generation systems, develop an intelligent photovoltaic power generation system as a product, extend the safe operation cycle of the system, and reduce human participation in the later stage.

5.1. The Initial Function Solving and Initial Function Model Construction for Intelligent Photovoltaic Power Generation System

The function of the intelligent photovoltaic power generation system needs to be decomposed. The total function of the system is to generate electric energy. The function decomposition result is shown in Figure 9.

After the function decomposition, find the component of function units using Formulas (2) and (3). The action A and the action object T of each function unit are analyzed to find the executive object EO and confirm the corresponding relationship between the functions and the components. To improve the design efficiency, search the existing product of the intelligent photovoltaic power generation system. After comparison, it is found that the intelligent photovoltaic power generation system can be improved by the solar tracking system. Therefore, the principal solution of the intelligent photovoltaic power generation system can be confirmed by referring to the components in the solar tracking system. The corresponding relationships between functions and components after the function solving are shown in

Table 2. Function–component correspondence.

Function Units	${\mathit{V}}_{\mathit{i}}$	${\mathit{N}}_{\mathit{i}}$	$\mathit{E}{\mathit{O}}_{\mathit{i}} \mathbf{Source}$	$\mathit{E}{\mathit{O}}_{\mathit{i}}$	${\mathit{A}}_{\mathit{i}}$	${\mathit{T}}_{\mathit{i}}$
$f_1$	detect	wind speed	direct function solving	wind speed sensor	detect	wind speed
$f_2$	detect	wind direction	direct function solving	wind direction sensor	detect	wind direction
$f_3$	detect	light intensity	solar tracking system	light sensor	detect	light intensity
$f_4$	change	angle	solar tracking system	motor 1	turn	solar panel
$f_5$	change	orientation	solar tracking system	motor 2	turn	solar panel
$f_6$	detect	angle	solar tracking system	attitude sensor 1	detect	panel angle
$f_7$	detect	orientation	solar tracking system	attitude sensor 2	detect	panel orientation
$f_8$	transform	solar energy	solar tracking system	solar panel	transform	solar energy
$f_9$	interactive	data	solar tracking system	control system	interactive	signal and data
$f_{10}$	detect	voltage	solar tracking system	voltage sensor	detect	battery
$f_{11}$	storage	electricity	solar tracking system	battery	storage	electricity

.

After the corresponding relationships between functions and components are confirmed, the function model of the intelligent photovoltaic power generation system can be established to express the components’ relationships between each other, to obtain the self-recovery priority later. Since the solar tracking system is the existing product of the intelligent photovoltaic system, the function model of the intelligent photovoltaic power generation system can be established by referring to the function model of the solar tracking system. Based on the components in Table 2, add the super-system “wind”, “dust”, “solar energy”, and the product “electricity”. The initial function model of the intelligent photovoltaic power generation system is shown in Figure 10.

5.2. Self-Recovery Prioritization of Intelligent Photovoltaic Power Generation System

Refer to the functional level rules in Section 3.2 to calculate the corresponding function value of components in the intelligent photovoltaic power generation system. Take the solar panel as an example. The solar panel “tracks” and “converts” the super-system resource “solar energy”, and the functional level is marked as $A_1+A_1$; the solar panel “generates” the product “electricity”, and the functional level is marked as $A_0$. Finally, the total functional level is $A_0+A_1+A_1$. According to the functional level rules, the lowest functional level in the system is $A_3$, so $A_3$ is assigned a value of 1. $A_0$, $A_1$, and $A_2$ are assigned values of 5, 3, and 2, respectively. The final solar panel functional level calculation result is 11. At the same time, since the solar panel is directly connected to the system, the corresponding functions of the solar panel are divided into main functions. The calculation process of the functional level corresponding to other components in the system is the same.

After the functional level calculation of components in the system is completed, combine the corresponding relationships between components and functions. The self-recovery priority is sorted according to the function category and functional level, as shown in

Table 3. Self-recovery priority of the intelligent photovoltaic power generation system.

Element	Corresponding Function	Value of Functional Level	Calculation Rules
solar panel	transform solar energy	$A_0+A_1+A_1$ = 11	$A_3$ = 1 $A_2$ = 2 $A_1$ = 3 $A_0$ = 5
battery	stored electricity	$A_0$ = 5
control system	interactive data (drive motor 1)	$A_2+A_2$ = 4
control system	interactive data (drive motor 2)	$A_2+A_2$ = 4
light sensor	detect light intensity	$A_1+A_3$ = 4
posture sensor 1/2	detect angle/orientation	$A_1+A_3$ = 4
wind direction sensor	detect wind direction	$A_1+A_3$ = 4
wind speed sensor	detect wind speed	$A_1+A_3$ = 4
motor 1, 2	change angle/orientation	$A_1$ = 3
voltage sensor	detect voltage	$A_1$ = 3

.

5.3. Self-Diagnosis Function Design of Intelligent Photovoltaic Power Generation System

When solving the function of the intelligent photovoltaic power generation system, the solar tracking system is taken as an existing product. The function structure of the intelligent photovoltaic power generation system can be established by referring to the function structure of the solar tracking system and combining the function decomposition results, as shown in Figure 11. In the self-recovery prioritization, “transform solar energy” and “store electricity” in the system are divided into the main functions. The self-recovery priority is the highest, so the self-diagnosis function is supposed to be added to the “transform solar energy” and “store electricity” functions.

In the function structure, the output parameter of the “transform solar energy” function is voltage, and the output parameter of the “store electricity” function is electric energy. It can be seen from the function structure that the “store electricity” function is coupled with the “detect voltage” function. When the output parameter of the “store electricity” function changes, the output parameter of the “detect voltage” function also changes. It can be judged whether the “store electricity” function operates normally through the change of the output parameter of the “detect voltage” function.

The parameters directly diagnosed by the self-diagnosis function are voltage (transform solar energy), electric energy (store electricity), indirect diagnosis voltage (transform solar energy), and voltage (detect voltage). The detection of voltage changes can satisfy the diagnosis requirements, so the “detect voltage” function is used as the self-diagnosis function for the “store electricity” and “transform solar energy” functions. At the same time, there is a “detect voltage” function in the system, which is designed to detect the voltage of the battery. After the self-diagnosis design, there is no need to add new functions. Based on the detection of the battery voltage, add the output voltage of the solar panel as the detection object.

5.4. Self-Regulation Function Design of Intelligent Photovoltaic Power Generation System

The existing product of the intelligent photovoltaic power generation system is the solar tracking system. Summarize the problems that occurred in the solar tracking system and predict the potential problems of the intelligent photovoltaic power generation system. After analysis, the main functions of the intelligent photovoltaic power generation system are “store electricity” and “transform solar energy”. The “store electricity” function is stable, but “transform solar energy” is unstable, which is affected by the floating dust in the environment. To maintain the normal operation of the intelligent photovoltaic power generation system, add the self-regulation function to the “solar energy conversion” function.

First, predict the parameters corresponding to the self-regulation function of the system using the LT matrix. The problems that occur in “transform solar energy” can be described as follows: The environment contains a large amount of dust. When the dust contacts the solar panel, the solar panel will be deformed at the micro level. The dust will gather on the surface of the solar panel, causing the photosensitive area to decrease. The parameter of the problem is extracted as the area (L²T⁰). The semi-effective solution to this problem can be described as follows: The solar panel is an ideal rigid body with an infinite elastic modulus. When the dust contacts the battery panel, the panel will not produce any deformation, so the dust cannot gather and adhere and will naturally fall under the action of gravity. The extracted semi-effective solution parameter is the elastic modulus (L²T⁻⁴). According to Formula (1), the LT parameter corresponding to the self-regulation function is L⁴T⁻⁴. In the LT matrix, the physical parameters corresponding to the LT parameters are “force” and “temperature gradient”. Since the working environment of the intelligent photovoltaic power generation system is outdoors, it is not practical to use the temperature gradient for cleaning, so the LT parameter is converted into the physical parameter “force”. A total of 82 effects are obtained by searching the parameter “force” in the effect library and recorded as set $A$.

After the parameter-effect retrieval is completed, the function-effect retrieval is supposed to be carried out by using the effect library, and the intersection of the two will be taken to narrow the effect selection range. Select “converse solar energy to electric energy” in the effect library software. A total of four effects are obtained and recorded as set $C$. The cause for the problem of “transform solar energy” is dust adhesion. Therefore, select “remove divided solid” in the effect library software, and a total of 90 effects are obtained, which are recorded as set $B$. Unionize the function-effect retrieval result, $B\cup C$, and a total of 94 effects are obtained. Set $D$ is obtained by taking the intersection of the parameter-effect set and the function-effect set $B\cup C$, $D=A\cap \left(B\cup C\right)$, and a total of 22 effects are obtained. Among the effects in set $D$, the feasibility of changing materials is low, and the system does not contain magnetic fields. After removing the effect that requires changing materials and magnetic fields, there remain nine effects in set $N$, as shown in

Table 4. Elements included in set N.

Effect	Required Parameters	Remarks
Jet effect	$Q=\sqrt{10P}$	$P$: pressure
Archimedes’ law	$F=G=\rho \mathrm{g}V$	$\rho$: density, $V$: volume
Bernoulli’s law	$p+\rho gh+\frac{1}{2}\rho v^2=K$	$P$: pressure, $\rho$: density, $h$: height, $V$: volume $K$: constant
Centrifugal force	$F=\frac{m{v}^2}{r}$	$m$: mass, $v$: linear velocity, $r$: radius
Newton’s second law	$F=\mathrm{ma}$	$m$: mass, $a$: acceleration
Gravitation	$F=\frac{GMm}{R^2}$	$G$: scale factor, $M$: mass of object 1, $m$: mass of object 2, $R$: distance between two objects
Inertia	$M$	All objects with weight have inertia
Wind	none	Wind resources exist in the environment
Fluid drag law	$F=Kv$	$K$: scale factor, $v$: velocity

.

Comparing the parameters marked in the function structure, it can be seen that there is no parameter pressure in the system and the system volume does not change. Archimedes’ law and Bernoulli’s law are removed from set $N$. In Table 4, the law of universal gravitation, Newton’s second law, and fluid resistance are the objective expressions of the laws of mechanics, which are less enlightening for the realization form of the self-regulation function. It is difficult to generate high-frequency and high-speed conditions in the intelligent photovoltaic power generation system, and the feasibility of using inertia to remove dust is low. The above effects or laws are removed in set $N$. Finally, set $N$ contains centrifugal force and wind.

Set $N$ is not an empty set. The function structure contains all input parameters needed by centrifugal force and wind force. The overlap degree $W$ calculated using Formula (4) is 100%. Firstly, analyze the feasibility of using centrifugal force to remove dust: dust will be adsorbed on the solar panel by friction force. It is required that the centrifugal force be greater than the friction force. When the radius is constant, greater centrifugal force requires a higher linear velocity and greater angular velocity. However, a large angular velocity in the intelligent photovoltaic power generation system will damage it, so the feasibility is low. Analyze the feasibility of using wind power to remove dust. The wind power resources are also abundant in the mountains and Gobi [32] Desert. In addition, there is no need to introduce new parameters to remove the floating dust on the surface of the panel through wind power, and there is no need to resolve the function. Finally, wind power is used to remove the dust as the self-regulation function corresponding to “transform solar energy”.

5.5. The Function Resolution of Intelligent Photovoltaic Power Generation System Oriented to Self-Recovery

Confirm that the self-diagnosis function corresponding to the “transform solar energy” function is “detect voltage”. The function “detect voltage” already exists in the system. The voltage sensor is one of the components in the initial solution of the system so that the voltage sensor can satisfy the functional requirements.

It is confirmed that the self-regulation function corresponding to the “transform solar energy” function is “remove dust by wind power”, but the function does not exist in the function decomposition result of the system. Therefore, analyze whether the component obtained from the initial function solving of the system can additionally realize the function “remove dust by wind power” based on satisfying the original functional requirements. No internal parameters are required to realize the “remove dust by wind power” function. However, since wind resources exist in the environment and the power and direction are difficult to control, the intelligent photovoltaic power generation system needs to change the angle (ω) to keep the orientation of the battery panel consistent with the wind direction. In the function structure of the intelligent photovoltaic power generation system, the output parameter of the “drive motor 2” function is horizontal angle information ω. The original function of motor 2 is to “change direction”. By controlling the solar panel to aim at the maximum light intensity point, improve the efficiency of the “transform solar energy” function to generate the product “electricity”. It can be seen from the analysis that “remove dust by wind power” can be realized by detecting the direction of wind power and changing the direction of the system. There are already existing components for detecting wind power and wind direction in the system: wind speed and wind direction sensors. Motor 2 can realize the self-regulation functions, “change direction” and “remove dust by wind power”. In the intelligent photovoltaic power generation system, the self-diagnosis function and self-regulation function can be realized by the components in the initial principal solution of the system. There is no need to select the principal solution again.

5.6. Functional Period Construction of Intelligent Photovoltaic Power Generation System Oriented to Self-Recovery

According to the self-recovery concept, the intelligent photovoltaic power generation system should ensure the realization of the main function, “transform solar energy”. The self-diagnosis function of the system is “detect voltage”, and the self-regulation function is “remove dust by wind power”. In the process of constructing the functional period, the complexity of the self-regulation function “remove dust by wind power” is high, and the natural wind does not fully satisfy the functional requirements of the system. At this time, substance-field analysis is used as the analysis tool, and 76 standard solutions are selected as the problem solution tool to obtain the function that reduces the complexity. The solution process is shown in Figure 12.

There is no need to transform the natural wind and the solar panel. The action F between the natural wind and the solar panel will be transformed. The standard solution 2.1.1 series connection substance-field model is adopted, and the “amplify wind” function, $f_{13}$, is added to construct a new functional period. When seeking the principal solution of the wind amplification device, considering that the tapered nozzle structure in a hair drier can realize the wind amplification function, the tapered nozzle structure is selected to be the wind amplification device, as shown in Figure 13.

The realization of the self-diagnosis function and self-regulation function of the system does not introduce the new structure. The construction of the functional period requires the structure “wind amplification device” as a component to realize the function “amplify wind”. According to the function requirements, the voltage sensor is connected to the solar panel in the function model, and the new component “wind amplification device” is introduced. The function model of the intelligent photovoltaic power generation system changes, as shown in Figure 14.

In the self-recovery prioritization, the function “detect voltage” changes, the functional level changes to 6, the wind amplification functional level is 3, and the function categories are all additional functions. The main functions of the system are not changed, and no iteration is required. The final function model is shown in Figure 15.

The working mode of the system is as follows: The intelligent photovoltaic power generation system diagnoses the system operation by detecting the output voltage of the solar panel. First, perform the initial functional period: when the environment information detected by the wind power and wind direction sensor does not harm the system, start to detect the maximum light intensity point, then drive the motor to rotate the solar panel, change the system orientation, and then convert the solar energy when the solar panel is detected to rotate to the maximum light intensity point. The self-diagnosis function is realized by detecting the voltage of the solar panel. When the detected voltage of the solar panel is in the predicted voltage range, it is considered that the system’s “transform solar energy” function is running properly, and the initial functional period will be executed. When the detected voltage is lower than the predicted voltage range, it is considered that the dust accumulation causes the problem in the system’s “transform solar energy” function. The functional period oriented to self-recovery will be performed: based on the initial functional period, the system’s directional device changes from light tracing to the self-regulation function “remove dust by wind power”. In addition, when there is no light at night and the system needs to remove dust, the self-regulation function will be performed to prevent dust from adhering to the surface of the panel. When the detected output voltage of the solar panel returns to the normal level or the output voltage does not change significantly in a certain time, the initial functional period will be executed again. The initial functional period of the system and the functional period oriented to self-recovery are shown in Figure 16.

6. Experimental Results

We verified the wind amplification effect of the wind amplification device through simulation. The fluid simulation of the product was carried out using SolidWorks2016 software. The wall surface condition was set as the insulation wall surface, and the roughness was 60 μm. Under the condition of standard atmospheric pressure, an ambient temperature of 293.20 k, and a fluid type of air, a wind force of 3 M/S was input. The simulation results show that the wind amplification device has a significant wind amplification effect, as shown in Figure 17. After confirming that the wind amplification device has a significant wind amplification effect, the final prototype is shown in Figure 18.

It is difficult to directly measure the degree of dust adhesion, so the degree of dust adhesion can be expressed indirectly by the output voltage of the battery panel. Let the dust fall naturally above the system to make it adhere to the surface of the battery panel. The lower the output voltage of the battery panel, the higher the degree of dust adhesion. After different degrees of dust adhered to the surface of the panel, we recorded the output voltage of the no-dust intelligent photovoltaic power generation system, the dust-adhered intelligent photovoltaic power generation system, and the intelligent photovoltaic power generation system after performing the functional period oriented to self-recovery after 10 min under the same light intensity. The data of the experiment are shown in

Table 5. Experimental data.

Group Number	${\mathit{U}}_{\mathbf{1}}$	${\mathit{U}}_{\mathbf{2}}$	${\mathit{U}}_{\mathbf{3}}$	$\mathbf{\Delta }\mathit{\eta }$
1	6.26 V	5.12 V	5.54 V	6.71%
2	6.06 V	5.38 V	5.79 V	6.77%
3	5.83 V	4.85 V	5.25 V	6.86%
4	5.44 V	4.63 V	5.01 V	6.99%
5	7.11 V	5.61 V	6.12 V	7.17%
6	7.67 V	6.37 V	6.96 V	7.69%
7	7.43 V	6.45 V	6.81 V	4.84%
8	7.35 V	6.28 V	6.75 V	6.39%
9	7.20 V	6.16 V	6.62 V	6.39%
10	7.04 V	6.03 V	6.49 V	6.53%
11	6.33 V	5.12 V	5.60 V	7.58%
12	6.12 V	5.38 V	5.86 V	7.84%
13	5.88 V	4.85 V	5.32 V	7.99%
14	5.88 V	4.85 V	5.12 V	4.59%
15	5.88 V	4.85 V	5.43 V	9.86%

.

To visually express the improvement in the self-recovery ability of the system, taking the improvement in the power generation efficiency as an indicator, the output voltage of the no-dust intelligent photovoltaic power generation system was recorded as $U_1$, and the output voltage of the dust-adhered solar panel was recorded as $U_2$. After performing the functional period oriented to self-recovery, the output voltage of the intelligent photovoltaic power generation system was recorded as $U_3$, and the formula of the power generation efficiency $\mathsf{\Delta }\eta$ improvement is as follows:

$$\mathsf{\Delta }\eta =\frac{U_3-U_2}{U_1}$$

(5)

The result of the experiment shows that the functional period oriented to self-recovery has significant cleaning ability. After performing the functional period oriented to self-recovery, the power generation efficiency of the system was significantly improved.

7. Conclusions and Discussion

This paper applied the functional-level product research method to the product design process under the self-recovery concept in order to reduce the problem that the current application of the self-recovery concept is mostly focused on the structural and parametric levels, which may lead to the waste of some resources in the product. By extending the application of the self-recovery concept to the functional level, a functional self-recovery design process model of the product was proposed. Using tools such as function models, function structures, functional-level algorithms, the LT matrix, and an effect library, the main functions of the product were identified, and the implementation scheme of the self-diagnosis and self-regulation functions corresponding to the main functions were determined. Finally, this paper demonstrated the effectiveness of the product functional self-recovery design process model through an intelligent photovoltaic power generation system. Through the design process model, the efficiency of product resource utilization can be improved, and the self-recovery concept can be applied to the product design process from more perspectives to reduce the possibility of problems.

Although this paper studied the feasibility and rationality of applying the self-recovery concept to the design process of complex products from the functional perspective, there are still some shortcomings worthy of further study. For example, the conversion between physical parameters and LT parameters is not complete, it is difficult to determine all the input parameters of effect realization, and the product integration problem may occur in the process of establishing a new product function model by using the existing product function model, resulting in a large deviation between the initial function model and the final function model. Our team will continuously improve the model and conduct in-depth research on the current shortcomings.