Product family quality improvement method and device based on artificial intelligence technology

Abstract

The invention provides a product family quality improvement method and device based on an artificial intelligence technology, and the method is characterized in that the method comprises the following steps: S1, collecting module quality characteristic data of M module instances; s2, constructing an initial quality characteristic correlation model, and constructing a quality characteristic data set according to historical product instances to train the initial quality characteristic correlation model to obtain a quality characteristic correlation model; s3, constructing a product family-oriented quality improvement model based on the quality characteristic correlation model; s4, solving the quality improvement model through a plurality of different intelligent optimization algorithms to obtain L module instance combination schemes; s5, comparing and analyzing the L module instance combination schemes to obtain an optimal module instance combination scheme as a combination scheme; and step S6, combining the M module instances into N product instances according to the combination scheme. In short, the method can improve the quality level of a final product.

Description

Technical Field

[0001] The present invention relates to the technical fields of artificial intelligence and quality improvement, and particularly relates to a method and device for product family quality improvement based on artificial intelligence technology. Background Art

[0002] The rapid development of economic globalization and information technology has made the competition among enterprises increasingly fierce, and customers' demands for products are becoming more and more personalized and diversified. In this context, enterprises need to not only meet customers' personalized needs but also control costs and improve product quality. The traditional large-scale production method cannot meet personalized needs, and the production mode of single-piece and small-batch production is too costly. It is an inevitable trend to rapidly develop and produce product systems of the same type but with different variants. Mass customization has become the mainstream production mode. The core of mass customization is the sharp increase in the variety and customization of products without increasing the corresponding costs. Product family development is one of the effective means to achieve mass customization. It improves the reusability of components and the efficiency of product development through modular technology and product platform development, facilitates the management of serialized products, and reduces the product development cycle and costs.

[0003] A product family includes multiple product variants and modules, and each product variant includes multiple modules. Different module combinations can obtain different types of products to meet the diverse needs of users. In addition to product diversity, quality is also one of the important indicators that customers pay attention to.

[0005] Existing technical solutions, such as option selection, have a deterministic relationship between the quality characteristics of the products and modules involved, which can be described by traditional mathematical formulas. When facing complex products, it is difficult to apply this method. There are also some solutions that use Bayesian networks to analyze quality characteristics. Other quality improvement methods are for single products, and those involving product family quality improvement are relatively rare. This method uses artificial intelligence technology to describe the relationship between quality characteristics, constructs a quality characteristic association model, constructs a quality improvement model for the product family based on this model, and uses intelligent optimization algorithms to solve the model to achieve the purpose of quality improvement. Summary of the Invention

[0006] The present invention is made to solve the above problems, and aims to provide a method and device for product family quality improvement based on artificial intelligence technology.

[0007] The present invention provides a product family quality improvement method based on artificial intelligence technology, which is used to obtain N product instances from M module instances and has the following characteristics, including the following steps: Step S1, collect the module quality characteristic data of M module instances; Step S2, construct an initial quality characteristic association model, and train the initial quality characteristic association model according to the quality characteristic data set constructed from historical product instances to obtain a quality characteristic association model; Step S3, construct a quality improvement model for the product family based on the quality characteristic association model; Step S4, solve the quality improvement model through multiple different intelligent optimization algorithms respectively to obtain L module instance combination schemes; Step S5, conduct a comparative analysis on the L module instance combination schemes to obtain the optimal module instance combination scheme as the combination scheme; Step S6, according to the combination scheme, combine M module instances into N product instances, where among the M module instances, there are multiple module instances with different module quality characteristic data but the same type and can be replaced with each other. The multiple intelligent algorithms in Step S4 are genetic algorithm and differential evolution algorithm. Each intelligent optimization algorithm in Step S4 includes the following steps: Step S4-1, generate multiple module instance combination schemes according to all module instances, and each module instance combination scheme includes N simulated product instances; Step S4-2, input the module quality characteristic data of all module instances corresponding to each simulated product instance into the quality improvement model to obtain the corresponding module instance combination scheme value; Step S4-3, optimize the module instance combination scheme according to all module instance combination scheme values; Step S4-4, repeat Step S4-2 to Step S4-3 until the corresponding termination condition is reached, then L optimized module instance combination schemes are obtained. In the product family quality improvement method based on artificial intelligence technology provided by the present invention, it may also have the following characteristics: Among them, in Step S4-2, the quality improvement model generates multiple product quality characteristic data corresponding to the simulated product instance according to the module quality characteristic data of all module instances corresponding to the simulated product instance, and generates the module instance combination scheme value according to the product quality characteristic data of all simulated product instances. The calculation expression of the module instance combination scheme value is: where MICPV is the module instance combination scheme value, and PQEV i is the product quality expected value PQEV of the i-th simulated product instance in the module instance combination scheme, m is the number of product quality characteristics, y j is the j-th product quality characteristic data, w j is the weight coefficient corresponding to y j , and L(y j ) is the calculation result of the standard quality loss function corresponding to y j . When the product quality characteristic data y j is a quality characteristic of the larger-the-better type, the calculation expression of L(y j ) is: where K is the mass loss coefficient and K ≤ 1, E(y j ) is the expectation of y j , y′ is the maximum achievable value of E(y j ), y * is the minimum acceptable value of E(y j ). When the product quality characteristic data y j is a smaller-the-better quality characteristic, the calculation expression of L(y j ) is: where y″ is the maximum achievable value of E(y j ), y ** is the minimum acceptable value of E(y j ). When the product quality characteristic data y j is a nominal-the-best quality characteristic, the calculation expression of L(y j ) is: where y + is the maximum achievable value of E(y j ), y - is the minimum acceptable value of E(y j ), y c is the target value of E(y j ), α is the mass loss coefficient when y c ≤ E(y j ) ≤ y + and α ≤ 1, β is the mass loss coefficient when y - ≤ E(y j ) ≤ y c and β ≤ 1.

[0008] In the product family quality improvement method based on artificial intelligence technology provided by the present invention, it may further have the following feature: wherein, in step S2, an initial quality characteristic association model is constructed through a neural network, and the quality characteristic association model includes an input layer, a hidden layer, and an output layer connected in sequence.

[0009] In the product family quality improvement method based on artificial intelligence technology provided by the present invention, it may also have the following characteristics: Among them, in step S4, obtaining the combined solution according to the genetic algorithm includes the following steps: Step T1, performing two-dimensional real number coding on the module quality characteristic data of M module instances to obtain coded data, which serves as an individual in the population; Step T2, initializing the population according to the coded data to obtain multiple module instance combination solutions; Step T3, combining the quality characteristic association model, evaluating the fitness of each module instance combination solution to obtain the corresponding module instance combination solution value; Step T4, performing selection operation, crossover operation, and mutation operation on the individuals in the population in sequence according to the module instance combination solution value to obtain multiple optimized module instance combination solutions; Step T5, repeating Step T3 to Step T4 until the termination condition is reached, then obtaining multiple optimized module instance combination solutions.

[0010] In the product family quality improvement method based on artificial intelligence technology provided by the present invention, it may also have the following characteristics: Among them, in step T4, the selection operation includes roulette wheel and elitist retention strategy, and the crossover parameter of the crossover operation and the mutation parameter of the mutation operation are optimized through design of experiments DOE.

[0011] The present invention also provides a product family quality improvement device based on artificial intelligence technology, which is used to obtain N product instances according to M module instances, and has the following characteristics, including: a data acquisition unit, which is used to acquire the module quality characteristic data of M module instances; a storage unit, which is used to store a quality improvement model for the product family and multiple different intelligent optimization algorithms; a combination scheme generation unit, which is used to select multiple intelligent optimization algorithms from the storage unit to solve the quality improvement model respectively, and obtain L module instance combination schemes; an analysis unit, which is used to compare and analyze the L module instance combination schemes to obtain the optimal module instance combination scheme as the combination scheme; a combination control unit, which is used to control the combination device to combine M module instances into N product instances according to the combination scheme. Among them, among the M module instances, there are multiple module instances with different but same-type module quality characteristic data that can be replaced with each other. The quality improvement model is constructed based on a quality characteristic association model, and the quality characteristic association model is obtained by training an initial quality characteristic association model with a quality characteristic data set constructed according to historical product instances. The intelligent algorithms include a genetic algorithm and a differential evolution algorithm. Each intelligent optimization algorithm includes the following steps: Step S4-1, generating multiple module instance combination schemes according to all module instances, and each module instance combination scheme includes N simulated product instances; Step S4-2, inputting the module quality characteristic data of all module instances corresponding to each simulated product instance into the quality improvement model to obtain the corresponding module instance combination scheme value; Step S4-3, optimizing the module instance combination scheme according to all module instance combination scheme values; Step S4-4, repeating Step S4-2 to Step S4-3 until the corresponding termination condition is reached, then L optimized module instance combination schemes are obtained.

[0012] Functions and effects of the invention

[0013] According to the product family quality improvement method and device based on artificial intelligence technology involved in the present invention, because, on the one hand, a quality characteristic association model is constructed through the quality characteristic data of historical product instances, and then the module quality characteristic data of each module instance constituting the product instance is analyzed to predict the product quality characteristic of the product instance; on the other hand, multiple intelligent optimization algorithms are used to combine the module instances, and the module instance combination scheme value is analyzed and calculated according to the product quality characteristics corresponding to all product instances in the scheme, and then the combination scheme is evaluated, and the optimal combination scheme is obtained through the comparative analysis of the results of multiple algorithms to construct the optimal product instance. Therefore, the product family quality improvement method and device based on artificial intelligence technology of the present invention can improve the quality level of the final product. Brief description of the drawings

[0014] Figure 1 It is a schematic diagram of the quality characteristic association relationship between product instances and module instances in the embodiment of the present invention;

[0015] Figure 2 It is a schematic flowchart of the product family quality improvement method based on artificial intelligence technology in the embodiments of the present invention;

[0016] Figure 3 It is a schematic diagram of the quality characteristic data set in the embodiments of the present invention;

[0017] Figure 4 It is a schematic diagram of the working principle of the quality characteristic association model in the embodiments of the present invention;

[0018] Figure 5 It is a schematic diagram of the module instance combination scheme in the embodiments of the present invention;

[0019] Figure 6 It is a schematic flowchart of obtaining the combination scheme by the genetic algorithm in the embodiments of the present invention;

[0020] Figure 7 It is a block diagram of the product family quality improvement device based on artificial intelligence technology in the embodiments of the present invention. Detailed implementation manners

[0021] In order to make the technical means, creative features, achieved purposes and effects of the present invention easy to understand, the following embodiments will specifically describe the product family quality improvement method and device based on artificial intelligence technology of the present invention in conjunction with the accompanying drawings.

[0022] Figure 1 It is a schematic diagram of the quality characteristic association relationship between product instances and module instances in the embodiments of the present invention.

[0023] As Figure 1 shown, the product-module relationship is that a product is composed of different types of modules, and the quality characteristic relationship is that due to the differences in the quality characteristics of module instances corresponding to the same module, that is, module quality characteristics, different product instances of the same type of product obtained by combination have different quality characteristics, that is, product quality characteristics.

[0024] Therefore, by finding the optimal combination scheme, multiple module instances with different quality characteristics can be combined into multiple product instances with optimal comprehensive quality characteristics. Therefore, the product family quality improvement method based on artificial intelligence technology in this embodiment is used to obtain N product instances with the best quality according to M module instances that can form N product instances.

[0025] Among them, among the M module instances, there are multiple module instances with different module quality characteristic data but the same type and can be replaced with each other.

[0026] Figure 2It is a schematic flowchart of the product family quality improvement method based on artificial intelligence technology in the embodiments of the present invention.

[0027] As Figure 2 shown, the product family quality improvement method based on artificial intelligence technology in this embodiment includes the following steps:

[0028] Step S1, collect the module quality characteristic data of M module instances.

[0029] Step S2, construct an initial quality characteristic association model, and train the initial quality characteristic association model according to the quality characteristic data set constructed from historical product instances to obtain a quality characteristic association model, wherein the quality characteristic association model is constructed by a neural network.

[0030] Figure 3 It is a schematic diagram of the quality characteristic data set in the embodiments of the present invention.

[0031] As Figure 3 shown, the quality characteristic data set includes the module quality characteristic data and product quality characteristic data of l historical product instances collected for a single product. Each product has x 1 ~x m module quality characteristics, is the specific value of the module quality characteristic x j corresponding to the i-th historical product instance, and has y 1 ~y n product quality characteristics, is the specific value of the product quality characteristic y j corresponding to the i-th historical product instance.

[0032] Figure 4 It is a schematic diagram of the working principle of the quality characteristic association model in the embodiments of the present invention.

[0033] As Figure 4 shown, the quality characteristic association model 2 includes an input layer 21, a hidden layer 22, and an output layer 23 connected in sequence. Input the module quality characteristic data of all module instances corresponding to a single product instance into the quality characteristic association model 2. After being processed by the input layer 21, the hidden layer 22, and the output layer 23 in sequence, multiple product quality characteristic data corresponding to this product instance are obtained.

[0034] Step S3, construct a quality improvement model for the product family based on the quality characteristic association model.

[0035] Figure 5 It is a schematic diagram of the module instance combination scheme in the embodiments of the present invention.

[0036] As Figure 5As shown, different module instances belonging to the same module can be replaced with each other, so as to combine multiple different product instances. Each module instance combination scheme has its corresponding comprehensive quality characteristic, that is, the module instance combination scheme value. When there are many products and modules, there are a large number of module instance combination schemes, and it is difficult for humans to obtain the optimal solution, that is, the optimal combination scheme. Therefore, the product family quality improvement method based on artificial intelligence technology in this embodiment performs scheme combination through an intelligent optimization algorithm, that is, solves the quality improvement model and mines the optimal combination scheme. The specific process is as shown in steps S4 to S5:

[0037] Step S4, solve the quality improvement model through multiple different intelligent optimization algorithms respectively, and obtain L module instance combination schemes. Among them, the multiple intelligent algorithms are genetic algorithm and differential evolution algorithm.

[0038] Among them, each intelligent optimization algorithm includes the following steps:

[0039] Step S4-1, generate multiple module instance combination schemes according to all module instances. Each module instance combination scheme includes N simulated product instances.

[0040] Step S4-2, input the module quality characteristic data of all module instances corresponding to each simulated product instance into the quality improvement model, and obtain the corresponding module instance combination scheme value.

[0041] Among them, in step S4-2, the quality improvement model generates multiple product quality characteristic data corresponding to the simulated product instance according to the module quality characteristic data of all module instances corresponding to the simulated product instance, and generates the module instance combination scheme value according to the product quality characteristic data of all simulated product instances.

[0042] The calculation expression of the module instance combination scheme value is:

[0043]

[0044]

[0045] In the formula, MICPV is the module instance combination scheme value, and PQEV i is the product quality expectation value of the i-th initial product instance in the module instance combination scheme, m is the total number of product quality characteristics of the initial product instance, y j is the j-th product quality characteristic corresponding to the initial product instance, w j is the weight coefficient corresponding to y j and L(y j ) is the calculation result of the standard quality loss function corresponding to y j .

[0046] Where MICPV is the value of the module instance combination scheme, and PQEV i is the expected product quality PQEV of the i-th simulated product instance in the module instance combination scheme, m is the number of product quality characteristics, y j is the data of the j-th product quality characteristic, w j is the weight coefficient corresponding to y j , and L(y j ) is the calculation result of the standard quality loss function corresponding to y j .

[0047] When the product quality characteristic data y j is a quality characteristic of the larger-the-better type, the calculation expression of L(y j ) is:

[0048]

[0049] Where K is the quality loss coefficient and K ≤ 1, E(y j ) is the expectation of y j , y' is the maximum achievable value of E(y j ), and y * is the minimum acceptable value of E(y j ).

[0050] When the product quality characteristic data y j is a quality characteristic of the smaller-the-better type, the calculation expression of L(y j ) is:

[0051]

[0052] Where y″ is the maximum achievable value of E(y j ), and y ** is the minimum acceptable value of E(y j ).

[0053] When the product quality characteristic data y j is a quality characteristic of the nominal-the-best type, the calculation expression of L(y j ) is:

[0054]

[0055] Where y + is the maximum achievable value of E(y j ), y - is the minimum acceptable value of E(y j ), y c is the target value of E(y j ), α is the quality loss coefficient when y c ≤ E(y j ) ≤ y + and α ≤ 1, and β is the quality loss coefficient when y- ≤ E(y j ) ≤ y c when the mass loss coefficient and β ≤ 1.

[0056] Step S4-3, optimize the module instance combination scheme according to all module instance combination scheme values;

[0057] Step S4-4, repeat Step S4-2 to Step S4-3 until the corresponding termination condition is met, then L optimized module instance combination schemes are obtained.

[0058] Figure 6 It is a schematic flowchart of obtaining the combination scheme by the genetic algorithm in the embodiments of the present invention.

[0059] As Figure 6 shown, the steps of obtaining the combination scheme according to the genetic algorithm include the following:

[0060] Step T1, perform two-dimensional real number encoding on the module quality characteristic data of M module instances to obtain encoded data, which serves as an individual in the population.

[0061] Step T2, initialize the population according to the encoded data to obtain multiple module instance combination schemes.

[0062] Step T3, combine with the quality improvement model to evaluate the fitness of each module instance combination scheme to obtain the corresponding module instance combination scheme value.

[0063] Step T4, perform selection operation, crossover operation and mutation operation on the individuals in the population in sequence according to the module instance combination scheme value to obtain multiple optimized module instance combination schemes.

[0064] Among them, the selection operation includes roulette wheel and elite retention strategy, and the crossover parameter of the crossover operation and the mutation parameter of the mutation operation are optimized through design of experiments DOE.

[0065] Step T5, repeat Step T3 to Step T4 until the termination condition is reached, then multiple optimized module instance combination schemes are obtained.

[0066] Step S5, conduct a comparative analysis on the L module instance combination schemes to obtain the optimal module instance combination scheme as the combination scheme.

[0067] Step S6, combine M module instances into N product instances according to the combination scheme.

[0068] In this embodiment, a product family quality improvement device based on artificial intelligence technology is also provided, which is used to obtain N optimal product instances according to M module instances that can form N product instances. Among the M module instances, there are multiple module instances with different module quality characteristic data but the same type and can be replaced with each other.

[0069] Figure 7 It is a block diagram of the product family quality improvement device based on artificial intelligence technology in the embodiment of the present invention.

[0070] As Figure 7 shown, the product family quality improvement device 1 includes a data acquisition unit 10, a storage unit 20, a combination scheme generation unit 30, an analysis unit 40, and a combination control unit 50.

[0071] The data acquisition unit 10 is used to acquire the module quality characteristic data of M module instances.

[0072] The storage unit 20 is used to store the quality improvement model for the product family and multiple different intelligent optimization algorithms.

[0073] Among them, the intelligent algorithms include genetic algorithms and differential evolution algorithms, etc.

[0074] Among them, the quality improvement model is constructed based on the quality characteristic association model, and the quality characteristic association model is obtained by training the initial quality characteristic association model with the quality characteristic data set constructed according to historical product instances.

[0075] The combination scheme generation unit 30 is used to select multiple intelligent optimization algorithms from the storage unit to solve the quality improvement model respectively, and obtain L module instance combination schemes.

[0076] Among them, each intelligent optimization algorithm includes the following steps:

[0077] Step S4-1, generate multiple module instance combination schemes according to all module instances, and each module instance combination scheme includes N simulated product instances.

[0078] Step S4-2, input the module quality characteristic data of all module instances corresponding to each simulated product instance into the quality improvement model to obtain the corresponding module instance combination scheme value.

[0079] Step S4-3, optimize the module instance combination scheme according to all module instance combination scheme values.

[0080] Step S4-4, repeat steps S4-2 to S4-3 until the corresponding termination condition is reached, then L optimized module instance combination schemes are obtained.

[0081] The analysis unit 40 is used to compare and analyze L module instance combination schemes, and obtain the optimal module instance combination scheme as the combination scheme.

[0082] The combination control unit 50 is used to control the combination device to combine M module instances into N optimal product instances according to the combination scheme.

[0083] Functions and effects of the embodiment

[0084] According to the product family quality improvement method and device based on artificial intelligence technology involved in this embodiment, on the one hand, a quality characteristic association model is constructed through the quality characteristic data of historical product instances, and then the module quality characteristic data of each module instance constituting the product instance is analyzed to predict the product quality characteristics of the product instance; on the other hand, multiple intelligent optimization algorithms are used to combine the module instance schemes, and the module instance combination scheme value is calculated according to the product quality characteristics corresponding to all product instances in the scheme, and then the combination scheme is evaluated, and the optimal combination scheme is obtained through the comparative analysis of the results of multiple algorithms to construct the optimal product instance. In short, this method can improve the quality level of the final product.

[0085] The above embodiments are preferred cases of the present invention and are not used to limit the protection scope of the present invention.

Claims

1. A product family quality improvement method based on artificial intelligence technology, which is used to obtain N product instances according to M module instances, characterized in that, it includes the following steps: Step S1, collect the module quality characteristic data of the M module instances; Step S2, construct an initial quality characteristic association model, and train the initial quality characteristic association model according to the quality characteristic data set constructed by historical product instances to obtain a quality characteristic association model; Step S3, construct a quality improvement model for the product family based on the quality characteristic association model; Step S4, solve the quality improvement model respectively through multiple different intelligent optimization algorithms to obtain L module instance combination schemes; Step S5, conduct a comparative analysis on the L module instance combination schemes to obtain the optimal module instance combination scheme as the combination scheme; Step S6, according to the combination scheme, combine the M module instances into the N product instances, wherein, among the M module instances, there are multiple module instances with different module quality characteristic data but the same type and can be replaced with each other, the multiple intelligent algorithms in the step S4 are genetic algorithm and differential evolution algorithm, each of the intelligent optimization algorithms in the step S4 includes the following steps: Step S4-1, generate multiple module instance combination schemes according to all the module instances, and each module instance combination scheme includes N simulated product instances; Step S4-2, input the module quality characteristic data of all the module instances corresponding to each simulated product instance into the quality improvement model to obtain the corresponding module instance combination scheme value; Step S4-3, optimize the module instance combination scheme according to all the module instance combination scheme values; Step S4-4, repeat the step S4-2 to the step S4-3 until the corresponding termination condition is reached, then L optimized module instance combination schemes are obtained.

2. The product family quality improvement method based on artificial intelligence technology according to claim 1, characterized in that: wherein, in the step S4-2, the quality improvement model generates multiple product quality characteristic data corresponding to the simulated product instance according to the module quality characteristic data of all the module instances corresponding to the simulated product instance, and generates the module instance combination scheme value according to the product quality characteristic data of all the simulated product instances, the calculation expression of the module instance combination scheme value is: Where MICPV is the value of the module instance combination scheme, and PQEV i is the expected product quality of the i-th simulated product instance in the module instance combination scheme, m is the number of product quality characteristics, and y j is the data of the j-th product quality characteristic, and w j is the weight coefficient corresponding to y j , and L(y j ) is the calculation result of the standard quality loss function corresponding to y j . Product quality characteristic data y j When it is a larger-the-better quality characteristic, L(y j ) is calculated by the following expression: where K is the mass loss coefficient and K ≤ 1, E(y j ) is the expectation of y j , y' is the maximum achievable value of E(y j ), and y * is the minimum acceptable value of E(y j ). Product quality characteristic data y j When it is a smaller-the-better quality characteristic, the calculation expression of L(y j ) is as follows: where y″ is the maximum achievable value of E(y j ), y ** is the minimum acceptable value of E(y j ), Product quality characteristic data y j When it is a target-seeking quality characteristic, the calculation expression of L(y j ) is as follows: where y + is the maximum achievable value of E(y j ), y - is the minimum acceptable value of E(y j ), y c is the target value of E(y j ), α is the quality loss coefficient when y c ≤ E(y j ) ≤ y + and α ≤ 1, β is the quality loss coefficient when y - ≤ E(y j ) ≤ y c and β ≤ 1.

3. The product family quality improvement method based on artificial intelligence technology according to claim 1, characterized in that: wherein, in the step S2, the initial quality characteristic association model is constructed through a neural network, the quality characteristic association model includes an input layer, a hidden layer and an output layer connected in sequence.

4. The product family quality improvement method based on artificial intelligence technology according to claim 1 ， characterized in that: wherein, in the step S4, obtaining the module instance combination scheme according to the genetic algorithm includes the following steps: Step T1: Perform two-dimensional real number encoding on the module quality characteristic data of the M module instances to obtain encoded data, which serves as individuals in the population. Step T2: Initialize the population based on the encoded data to obtain multiple module instance combination schemes. Step T3: Combine with the quality improvement model to evaluate the fitness of each module instance combination scheme, and obtain the corresponding module instance combination scheme value. Step T4: Perform selection operation, crossover operation, and mutation operation on the individuals in the population in sequence according to the module instance combination scheme value to obtain multiple optimized module instance combination schemes. Step T5: Repeat Step T3 to Step T4 until the termination condition is reached, then obtain multiple optimized module instance combination schemes.

5. The product family quality improvement method based on artificial intelligence technology according to claim 1, characterized in that: Wherein, In Step T4, the selection operation includes roulette wheel and elitist retention strategy. The crossover parameter of the crossover operation and the mutation parameter of the mutation operation are optimized through design of experiments (DOE).

6. A product family quality improvement device based on artificial intelligence technology, used to obtain N product instances from M module instances. It is characterized in that It includes: A data acquisition unit, used to acquire the module quality characteristic data of the M module instances. A storage unit, used to store the quality improvement model for the product family and multiple different intelligent optimization algorithms. A combination scheme generation unit, used to select multiple of the intelligent optimization algorithms from the storage unit to solve the quality improvement model respectively, and obtain L module instance combination schemes. An analysis unit, used to perform comparative analysis on the L module instance combination schemes to obtain the optimal module instance combination scheme as the combination scheme. A combination control unit, used to control the combination device to combine the M module instances into the N product instances according to the combination scheme. Wherein, among the M module instances, there are multiple module instances with different but same-type module quality characteristic data that can be replaced with each other. The quality improvement model is constructed based on the quality characteristic association model. The quality characteristic association model is obtained by training the initial quality characteristic association model with the quality characteristic data set constructed according to historical product instances. The intelligent algorithms include genetic algorithm and differential evolution algorithm. Each intelligent optimization algorithm includes the following steps: Step S4-1: Generate multiple module instance combination schemes according to all the module instances, and each module instance combination scheme includes N simulated product instances. Step S4-2: Input the module quality characteristic data of all the module instances corresponding to each simulated product instance into the quality improvement model to obtain the corresponding module instance combination scheme value. Step S4-3: Optimize the module instance combination scheme according to all the module instance combination scheme values. Step S4-4: Repeat Step S4-2 to Step S4-3 until the corresponding termination condition is reached, then obtain L optimized module instance combination schemes.

Images