Metal plate stamping forming performance prediction method based on big data

Abstract

The invention discloses a metal plate stamping forming performance prediction method based on big data, and the method comprises the steps: collecting test data and related data of a stamping forming metal plate, and carrying out the preprocessing of the test data and influence data; performing clustering analysis on the test data according to the condition similarity to obtain a sub-data set, and performing influence analysis on the related data according to the sub-data set to obtain a change coefficient; performing feature selection on the test data according to the change coefficient to obtain key features, and performing correlation analysis on the key features and the change coefficient to obtain a correlation coefficient; and constructing a metal plate stamping forming performance prediction model according to the correlation coefficient, inputting to-be-predicted data into the metal plate stamping forming performance prediction model, and outputting a prediction result. The method not only can improve the precision of metal sheet stamping forming performance prediction based on big data, but also has good interpretability, and can be directly applied to a big data performance prediction system.

Description

Technical Field

[0001] The present invention relates to the field of big data, and in particular to a method for predicting the stamping forming performance of metal sheets based on big data. Background Art

[0002] In the sheet metal processing industry, stamping is a widely used manufacturing process. However, the stamping performance of sheet metal is affected by many complex factors such as material properties, process parameters, environmental factors, etc., and it is difficult to accurately predict using traditional methods. Most existing prediction methods rely on empirical formulas or a limited number of experimental data, and have problems such as low prediction accuracy and narrow scope of application. With the rapid development of big data technology, it has become possible to use massive data to model and predict complex systems. Therefore, developing a sheet metal stamping performance prediction method based on big data is of great significance for improving production efficiency and reducing scrap rates. Summary of the invention

[0003] The purpose of the present invention is to provide a metal sheet stamping forming performance prediction method based on big data.

[0004] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0005] The present invention comprises the following steps:

[0006] Collecting test data and related data of stamped metal sheets, and preprocessing the test data and the influencing data;

[0007] Performing cluster analysis on the test data according to conditional similarity to obtain sub-data sets, and performing influence analysis on the relevant data according to the sub-data sets to obtain a coefficient of variation;

[0008] Performing feature selection on the test data according to the variation coefficient to obtain key features, and performing correlation analysis on the key features and the variation coefficient to obtain a correlation coefficient;

[0009] A metal sheet stamping forming performance prediction model is constructed according to the correlation coefficient, the metal sheet stamping forming performance prediction model is optimized, the data to be predicted is input into the metal sheet stamping forming performance prediction model, and the prediction result is output.

[0010] Furthermore, the sub-datasets are obtained by a mean clustering algorithm, wherein the expression of conditional similarity is:

[0011]

[0012] The jth test data is z j , the τth test data is z τ , test data zj and test data z τ The correlation degree is θ(z j ,z τ ), test data z j and test data z τ The similarity is ζ(z j ,z τ ), the adjustment constant is σ.

[0013] Furthermore, the method of performing an impact analysis on the relevant data according to the sub-data sets to obtain a coefficient of variation includes:

[0014] The performance parameters of metal sheets are obtained by dividing the data sets, taking the performance parameters as the result and the related data as the cause, and causal machine learning is used to perform impact analysis on the related data to obtain the coefficient of variation.

[0015] Furthermore, the method of performing feature selection on the test data according to the variation coefficient to obtain key features includes:

[0016] Extract features from the test data to obtain performance features, cluster the performance features to obtain classified data, and calculate the coefficient of variation and the mutual information value of the classified data:

[0017]

[0018] The growth rate is β, the error constant is X, the mutual information value is ψ, the current number of iterations is t, and the i-th classification data is k i , the bth coefficient of variation is The maximum value of mutual information is ψ max , classification data k i The marginal probability is p(k i ), coefficient of variation The marginal probability is p(k i ), classification data k i and coefficient of variation The joint probability of

[0019] Introduce the particle population and calculate the fitness of the particles:

[0020]

[0021] The true positive is TP, the true negative is TN, the false positive is FP, the false negative is FN, and the fitness of the wth particle is The number of particles is B, the number of search particles is B w ;

[0022] The particle with the largest Fisher score in the classification data is used as the guide particle, and the speed and position of the particle are updated. The expression is:

[0023] V w (t+1)=α·V w (t)+λ1·r1[ps w (t)-D w (t)]+λ2·r2[gs(t)-D w (t)]

[0024] D w (t+1)=V w (t+1)+D w (t)

[0025] The velocity of the wth particle in the t+1th iteration is V w (t+1), the velocity of the wth particle in the tth iteration is V w (t), the position of the wth particle in the t+1th iteration is D w (t+1), the position of the wth particle in the tth iteration is D w (t), the inertia coefficient is α, the individual learning factor is λ1, the population learning factor is λ2, the random numbers from 0 to 1 are r1 and r2 respectively, and the local optimal position of the wth particle in the tth iteration is ps w (t), the global optimal position of the particle in the tth iteration is gs(t);

[0026] Calculate the diversity index of particles in the target space:

[0027]

[0028] The number of classified data in the cth category is N c , the value of the i-th classification data in the y-th category is k y,i , the overall mean of the i-th classification data is Classification data k i The average value in the cth category is Classification data k i The Fisher score is ζ(k i ), the local Fisher score is ζ ps (k i ), the global Fisher score is ζ gs (k i ), the number of classification data is N, and the diversity index is C q ;

[0029] Compute frequency thresholds for categorical data:

[0030]

[0031] The frequency threshold is fv, and the classification data is screened according to the frequency threshold to obtain high-frequency classification data and low-frequency classification data;

[0032] Calculate the reward and penalty factors:

[0033]

[0034] The average of the diversity index of the first three iterations in the current state is The control factor is μ, and the reward and punishment factor is φ;

[0035] A reward and punishment factor is used to impose a penalty on the particle speed of high-frequency classification data and reward the speed of low-frequency classification data. The iteration is continued until the maximum number of iterations is reached, otherwise the particle speed and position are updated;

[0036] The selection results of the particle pair performance characteristics are output as key features.

[0037] Furthermore, the method of performing correlation analysis on the key feature and the variation coefficient to obtain the correlation coefficient includes:

[0038] Calculate the correlation of key characteristics and coefficient of variation:

[0039]

[0040] The key feature vector of the sth test is u s , the coefficient of change vector for the sth test is h s , the key feature vector u s The weight vector is Variation coefficient vector h s The weight vector is Transposed to T, the key eigenvector u s and the coefficient of variation vector h s The covariance matrix of

[0041] Maximize the correlation and the expression is:

[0042]

[0043] The maximization function is According to the maximization function, the Lagrangian formula is constructed as follows:

[0044]

[0045] The key feature vector u s The characteristic value of Variation coefficient vector h sThe characteristic value of

[0046] Solve the Lagrangian formula to obtain the feature set with the maximum correlation, and calculate the correlation entropy of the key features in the feature set:

[0047]

[0048] The key feature vector u s and the coefficient of variation vector h s The correlation entropy is Key feature vector u s The parent node gene is x, and the coefficient vector h s The parent node gene is v, and the change coefficient vector is h s and the atomic sequence of the parent node gene v is h s →v, key feature vector u s and the atomic sequence of the parent node gene x is u s →x, atomic sequence h s →v and atomic sequence u s →x has a probability of appearing at the same time of p(u s →x,h s →v), atomic sequence h s →The probability of v appearing is p(u s →x);

[0049] Corrected probability, expression is:

[0050]

[0051] The atomic sequence u s →The counter of x is Atomic sequence h s →v and atomic sequence u s →x simultaneous occurrence counter is The number of atomic sequences is M;

[0052] Calculate the correlation coefficient by adjusting the correlation after the corrected probability:

[0053]

[0054] The correlation coefficient of the key feature u in the sth test is χ s (u), the number of key features in the sth test is n s , the variable control factor of the sth test is δ s , the correlation of the key feature u in the sth test after adjustment is The probability of the key feature u appearing is p(u), the norm is ‖·‖, and the correlation mean of the sth test is

[0055] Furthermore, a method for constructing a metal sheet stamping forming performance prediction model according to the correlation coefficient includes:

[0056] The objective function is constructed according to the correlation coefficient, and the expression is:

[0057]

[0058] The objective function is No. The design parameters of stamping performance are The sth test The correlation coefficient of the uth key feature of the stamping forming performance is No. The predicted value of stamping performance is No. The actual value of stamping performance is

[0059] The sheet metal stamping forming performance prediction model includes anomaly screening algorithm, gradient boosting algorithm, and machine learning algorithm;

[0060] The anomaly screening algorithm screens the input data according to the degree of change to obtain variable data;

[0061] The gradient boosting algorithm constructs multiple weak prediction models through variable data, gradually iterates and optimizes the minimization loss function to obtain feature data, evaluates the importance of features based on the frequency of use of feature data in splitting nodes in the weak prediction model, and obtains feature data sets;

[0062] The machine learning algorithm is trained based on the historical data of the feature data set to obtain the changing rules of the stamping forming performance, and the stamping forming performance is predicted based on the changing rules.

[0063] Furthermore, the method for optimizing the metal sheet stamping forming performance prediction model comprises:

[0064] Introduce particle population to optimize parameters, calculate the fitness of particles, and take the position of the particle with the largest fitness as the search target;

[0065] Calculate the particle positions during the search phase:

[0066]

[0067] The initial position of the particle of dimension d at the tth iteration is The upper bound of the optimization is X max , the lower bound of the optimization is X min , the change control parameter is β, the random numbers from 0 to 1 are r1 and r2 respectively, and the position of the first particle arriving at the tth iteration is The position of the ath particle in the dth dimension at the t+1th iteration is

[0068] The particle position is updated using the expression:

[0069]

[0070] The search speed of the particle is ρ, and the position of the ath particle in the dth dimension at the tth iteration is The random numbers from 0 to 1 are r3 and r4 respectively, and the maximum number of iterations is t max , the updated position of the ath particle in the dth dimension at the t+1th iteration is

[0071] The convergence factor is used to update the particle position, and the expression is:

[0072]

[0073] The corrected position of the ath particle in the dth dimension at the t+1th iteration is The convergence factor is η, the Cauchy operator is ξ, and the random number between 0 and 1 is r5;

[0074] Update the convergence factor, the expression is:

[0075]

[0076] The initial value of the convergence factor is η bg , the control factor is ω, and the convergence factor is η;

[0077] Continue to iterate until the maximum number of iterations is reached, otherwise update the convergence factor.

[0078] In a second aspect, an embodiment of the present application further provides an electronic device, including:

[0079] A processor; and a memory arranged to store computer executable instructions, which when executed cause the processor to perform the method steps described in the first aspect.

[0080] In a third aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple applications, the electronic device executes the method steps described in the first aspect.

[0081] The beneficial effects of the present invention are:

[0082] The present invention is a metal sheet stamping forming performance prediction method based on big data. Compared with the prior art, the present invention has the following technical effects:

[0083] The present invention can improve the accuracy of metal sheet stamping forming performance prediction of big data through preprocessing, cluster analysis, influence analysis, feature selection, association analysis, model construction and model optimization steps, thereby improving the accuracy of metal sheet stamping forming performance prediction of big data, and optimizing the metal sheet stamping forming performance prediction of big data, which can greatly save resources and improve work efficiency, and can realize automatic prediction of metal sheet stamping forming performance, and perform data fusion and influence correction on the metal sheet stamping forming performance prediction of big data in real time, which is of great significance to the metal sheet stamping forming performance prediction of big data, can adapt to the metal sheet stamping forming performance prediction of big data with different standards and the metal sheet stamping forming performance prediction needs of different big data, and has a certain universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 A flowchart of the steps of a metal sheet stamping forming performance prediction method based on big data according to the present invention;

[0085] Figure 2 It is a schematic diagram of the structure of an electronic device in an embodiment of this specification. DETAILED DESCRIPTION

[0086] The present invention is further described below by means of specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0087] The present invention provides a metal sheet stamping forming performance prediction method based on big data, comprising the following steps:

[0088] like Figure 1 As shown, in this embodiment, the following steps are included:

[0089] Collecting test data and related data of stamped metal sheets, and preprocessing the test data and the influencing data;

[0090] In the actual evaluation, taking high-strength aluminum alloy sheet as an example, the test data includes crack resistance, bulging performance, tensile performance, hole expansion performance, bending performance, composite forming performance, mold adhesion, and shape stability;

[0091] Relevant data include yield strength, tensile strength, elongation, hardening index, thickness direction coefficient, strain hardening index, temperature, humidity, material properties, stamping speed, die design, lubrication conditions, product shape, size, thickness, mechanical properties, chemical composition, microstructure;

[0092] Performing cluster analysis on the test data according to conditional similarity to obtain sub-data sets, and performing influence analysis on the relevant data according to the sub-data sets to obtain a coefficient of variation;

[0093] In the actual evaluation, the data sets are divided into: Category 1: rupture resistance, bulging performance, tensile performance, hole expansion performance, bending performance; Category 2: composite forming performance, mold adhesion, and shape stability; the coefficient of variation is 0.2752;

[0094] Performing feature selection on the test data according to the variation coefficient to obtain key features, and performing correlation analysis on the key features and the variation coefficient to obtain a correlation coefficient;

[0095] In the actual evaluation, the key features are maximum rupture load, cupping value, yield strength, tensile strength, elongation at break, limit hole expansion rate, minimum bending radius, composite forming index, die placement error, and springback rate; the total correlation coefficient of the test process is 0.2091;

[0096] A metal sheet stamping forming performance prediction model is constructed according to the correlation coefficient, the metal sheet stamping forming performance prediction model is optimized, the data to be predicted is input into the metal sheet stamping forming performance prediction model, and the prediction result is output.

[0097] In this embodiment, the sub-datasets are obtained by a mean clustering algorithm, wherein the expression of conditional similarity is:

[0098]

[0099] The jth test data is z j , the τth test data is z τ , test data z j and test data z τ The correlation degree is θ(z j ,z τ ), test data z j and test data z τ The similarity is ζ(z j ,z τ ), the adjustment constant is σ.

[0100] In this embodiment, the method of performing an impact analysis on the relevant data to obtain a coefficient of variation according to the sub-data sets includes:

[0101] The performance parameters of metal sheets are obtained by dividing the data sets, taking the performance parameters as the result and the related data as the cause, and causal machine learning is used to perform impact analysis on the related data to obtain the coefficient of variation.

[0102] In this embodiment, the method of performing feature selection on the test data according to the variation coefficient to obtain key features includes:

[0103] Extract features from the test data to obtain performance features, cluster the performance features to obtain classified data, and calculate the coefficient of variation and the mutual information value of the classified data:

[0104]

[0105] The growth rate is β, the error constant is X, the mutual information value is ψ, the current number of iterations is t, and the i-th classification data is k i , the bth coefficient of variation is The maximum value of mutual information is ψ max , classification data k i The marginal probability is p(k i ), coefficient of variation The marginal probability is p(k i ), classification data k i and coefficient of variation The joint probability of

[0106] Introduce the particle population and calculate the fitness of the particles:

[0107]

[0108] The true positive is TP, the true negative is TN, the false positive is FP, the false negative is FN, and the fitness of the wth particle is The number of particles is B, the number of search particles is B w ;

[0109] The particle with the largest Fisher score in the classification data is used as the guide particle, and the speed and position of the particle are updated. The expression is:

[0110] V w (t+1)=α·V w (t)+λ1·r1[ps w (t)-D w (t)]+λ2·r2[gs(t)-D w (t)]

[0111] D w (t+1)=V w (t+1)+D w (t)

[0112] The velocity of the wth particle in the t+1th iteration is V w (t+1), the velocity of the wth particle in the tth iteration is V w(t), the position of the wth particle in the t+1th iteration is D w (t+1), the position of the wth particle in the tth iteration is D w (t), the inertia coefficient is α, the individual learning factor is λ1, the population learning factor is λ2, the random numbers from 0 to 1 are r1 and r2 respectively, and the local optimal position of the wth particle in the tth iteration is ps w (t), the global optimal position of the particle in the tth iteration is gs(t);

[0113] Calculate the diversity index of particles in the target space:

[0114]

[0115] The number of classified data in the cth category is N c , the value of the i-th classification data in the y-th category is k y,i , the overall mean of the i-th classification data is Classification data k i The average value in the cth category is Classification data k i The Fisher score is ζ(k i ), the local Fisher score is ζ ps (k i ), the global Fisher score is ζ gs (k i ), the number of classification data is N, and the diversity index is C q ;

[0116] Compute frequency thresholds for categorical data:

[0117]

[0118] The frequency threshold is fv, and the classification data is screened according to the frequency threshold to obtain high-frequency classification data and low-frequency classification data;

[0119] Calculate the reward and penalty factors:

[0120]

[0121] The average of the diversity index of the first three iterations in the current state is The control factor is μ, and the reward and punishment factor is φ;

[0122] A reward and punishment factor is used to impose a penalty on the particle speed of high-frequency classification data and reward the speed of low-frequency classification data. The iteration is continued until the maximum number of iterations is reached, otherwise the particle speed and position are updated;

[0123] The selection results of the particle pair performance characteristics are output as key features.

[0124] In this embodiment, the method of performing correlation analysis on the key feature and the variation coefficient to obtain the correlation coefficient includes:

[0125] Calculate the correlation of key characteristics and coefficient of variation:

[0126]

[0127] The key feature vector of the sth test is u s , the coefficient of change vector for the sth test is h s , the key feature vector u s The weight vector is Variation coefficient vector h s The weight vector is Transposed to T, the key eigenvector u s and the coefficient of variation vector h s The covariance matrix of

[0128] Maximize the correlation and the expression is:

[0129]

[0130]

[0131] The maximization function is According to the maximization function, the Lagrangian formula is constructed as follows:

[0132]

[0133] The key feature vector u s The characteristic value of Variation coefficient vector h s The characteristic value of

[0134] Solve the Lagrangian formula to obtain the feature set with the maximum correlation, and calculate the correlation entropy of the key features in the feature set:

[0135]

[0136] The key feature vector u s and the coefficient of variation vector h s The correlation entropy is Key feature vector u s The parent node gene is x, and the coefficient vector h s The parent node gene is v, and the change coefficient vector is h s and the atomic sequence of the parent node gene v is h s→v, key feature vector u s and the atomic sequence of the parent node gene x is u s →x, atomic sequence h s →v and atomic sequence u s →x has a probability of appearing at the same time of p(u s →x,h s →v), atomic sequence h s →The probability of v appearing is p(u s →x);

[0137] Corrected probability, expression is:

[0138]

[0139] The atomic sequence u s →The counter of x is Atomic sequence h s →v and atomic sequence u s →x simultaneous occurrence counter is The number of atomic sequences is M;

[0140] Calculate the correlation coefficient by adjusting the correlation after the corrected probability:

[0141]

[0142] The correlation coefficient of the key feature u in the sth test is χ s (u), the number of key features in the sth test is n s , the variable control factor of the sth test is δ s , the correlation of the key feature u in the sth test after adjustment is The probability of the key feature u appearing is p(u), the norm is ‖·‖, and the correlation mean of the sth test is

[0143] In this embodiment, the method for constructing a metal sheet stamping forming performance prediction model according to the correlation coefficient includes:

[0144] The objective function is constructed according to the correlation coefficient, and the expression is:

[0145]

[0146] The objective function is No. The design parameters of stamping performance are The sth test The correlation coefficient of the uth key feature of the stamping forming performance is No. The predicted value of stamping performance is No. The actual value of stamping performance is

[0147] The sheet metal stamping forming performance prediction model includes anomaly screening algorithm, gradient boosting algorithm, and machine learning algorithm;

[0148] The anomaly screening algorithm screens the input data according to the degree of change to obtain variable data;

[0149] The gradient boosting algorithm constructs multiple weak prediction models through variable data, gradually iterates and optimizes the minimization loss function to obtain feature data, evaluates the importance of features based on the frequency of use of feature data in splitting nodes in the weak prediction model, and obtains feature data sets;

[0150] The machine learning algorithm is trained based on the historical data of the feature data set to obtain the changing rules of the stamping forming performance, and the stamping forming performance is predicted based on the changing rules.

[0151] In this embodiment, the method for optimizing the metal sheet stamping forming performance prediction model includes:

[0152] Introduce particle population to optimize parameters, calculate the fitness of particles, and take the position of the particle with the largest fitness as the search target;

[0153] Calculate the particle positions during the search phase:

[0154]

[0155] The initial position of the particle of dimension d at the tth iteration is The upper bound of the optimization is X max , the optimal lower bound is X min , the change control parameter is β, the random numbers from 0 to 1 are r1 and r2 respectively, and the position of the first particle arriving at the tth iteration is The position of the ath particle in the dth dimension at the t+1th iteration is

[0156] The particle position is updated using the expression:

[0157]

[0158] The search speed of the particle is ρ, and the position of the ath particle in the dth dimension at the tth iteration is The random numbers from 0 to 1 are r3 and r4 respectively, and the maximum number of iterations is t max , the updated position of the ath particle in the dth dimension at the t+1th iteration is

[0159] The convergence factor is used to update the particle position, and the expression is:

[0160]

[0161] The corrected position of the ath particle in the dth dimension at the t+1th iteration is The convergence factor is η, the Cauchy operator is ξ, and the random number between 0 and 1 is r5;

[0162] Update the convergence factor, the expression is:

[0163]

[0164] The initial value of the convergence factor is η bg , the control factor is ω, and the convergence factor is η;

[0165] Continue to iterate until the maximum number of iterations is reached, otherwise update the convergence factor.

[0166] Figure 2 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 2 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.

[0167] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 2 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0168] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0169] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a metal sheet stamping forming performance prediction device based on big data at the logical level. The processor executes the program stored in the memory and is specifically used to execute any of the above-mentioned metal sheet stamping forming performance prediction methods based on big data.

[0170] The above application Figure 1 The embodiment shown discloses a metal sheet stamping forming performance prediction method based on big data, which can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0171] The electronic device may also perform Figure 1 A metal sheet stamping forming performance prediction method based on big data is proposed and implemented Figure 1 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0172] An embodiment of the present application also proposes a computer-readable storage medium, which stores one or more programs, and the one or more programs include instructions. When the instructions are executed by an electronic device including multiple applications, any one of the aforementioned metal sheet stamping forming performance prediction methods based on big data is executed.

[0173] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0174] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0175] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0176] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0177] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0178] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0179] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0180] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0181] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0182] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A metal sheet stamping forming performance prediction method based on big data, characterized in that: The following steps are involved: Collecting test data and related data of stamped metal sheets, and preprocessing the test data and the influencing data; Performing cluster analysis on the test data according to conditional similarity to obtain sub-data sets, and performing influence analysis on the relevant data according to the sub-data sets to obtain a coefficient of variation; Performing feature selection on the test data according to the variation coefficient to obtain key features, and performing correlation analysis on the key features and the variation coefficient to obtain a correlation coefficient; A metal sheet stamping forming performance prediction model is constructed according to the correlation coefficient, the metal sheet stamping forming performance prediction model is optimized, the data to be predicted is input into the metal sheet stamping forming performance prediction model, and the prediction result is output.

2. According to the big data-based metal sheet stamping performance prediction method of claim 1, it is characterized in that: The sub-datasets are obtained by the mean clustering algorithm, where the expression of conditional similarity is: The jth test data is z j , the τth test data is z τ , test data z j and test data z τ The correlation degree is θ(z j ,z τ ), test data z j and test data z τ The similarity is ζ(z j ,z τ ), the adjustment constant is σ.

3. According to the big data-based metal sheet stamping performance prediction method of claim 1, it is characterized in that: The method of performing an impact analysis on the relevant data according to the sub-data sets to obtain a coefficient of variation includes: The performance parameters of metal sheets are obtained by dividing the data sets, taking the performance parameters as the result and the related data as the cause, and causal machine learning is used to perform impact analysis on the related data to obtain the coefficient of variation.

4. According to the big data-based metal sheet stamping forming performance prediction method of claim 1, it is characterized in that: The method of performing feature selection on the test data according to the variation coefficient to obtain key features includes: Extract features from the test data to obtain performance features, cluster the performance features to obtain classified data, and calculate the coefficient of variation and the mutual information value of the classified data: The growth rate is β, the error constant is X, the mutual information value is ψ, the current number of iterations is t, and the i-th classification data is k i , the bth coefficient of variation is The maximum value of mutual information is ψ max , classification data k i The marginal probability is p(k i ), coefficient of variation The marginal probability is p(k i ), classification data k i and the coefficient of variation θ b The joint probability of Introduce the particle population and calculate the fitness of the particles: The true positive is TP, the true negative is TN, the false positive is FP, the false negative is FN, and the fitness of the wth particle is The number of particles is B, the number of search particles is B w ; The particle with the largest Fisher score in the classification data is used as the guide particle, and the speed and position of the particle are updated. The expression is: V w (t+1)=α·V w (t)+λ1·r1[ps w (t)-D w (t)]+λ2·r2[gs(t)-D w (t)] D w (t+1)=V w (t+1)+D w (t) The velocity of the wth particle in the t+1th iteration is V w (t+1), the velocity of the wth particle in the tth iteration is V w (t), the position of the wth particle in the t+1th iteration is D w (t+1), the position of the wth particle in the tth iteration is D w (t), the inertia coefficient is α, the individual learning factor is λ1, the population learning factor is λ2, the random numbers from 0 to 1 are r1 and r2 respectively, and the local optimal position of the wth particle in the tth iteration is ps w (t), the global optimal position of the particle in the tth iteration is gs(t); Calculate the diversity index of particles in the target space: The number of classified data in the cth category is N c , the value of the i-th classification data in the y-th category is k y,i , the overall mean of the i-th classification data is Classification data k i The average value in the cth category is Classification data k i The Fisher score is ζ(k i ), the local Fisher score is ζ ps (k i ), the global Fisher score is ζ gs (k i ), the number of classification data is N, and the diversity index is C q ; Compute frequency thresholds for categorical data: The frequency threshold is fv, and the classification data is screened according to the frequency threshold to obtain high-frequency classification data and low-frequency classification data; Calculate the reward and penalty factors: The average of the diversity index of the first three iterations in the current state is The control factor is μ, and the reward and punishment factor is φ; A reward and punishment factor is used to impose a penalty on the particle speed of high-frequency classification data and reward the speed of low-frequency classification data. The iteration is continued until the maximum number of iterations is reached, otherwise the particle speed and position are updated; The selection results of the particle pair performance characteristics are output as key features.

5. The metal sheet stamping forming performance prediction method based on big data according to claim 1 is characterized in that: The method of performing correlation analysis on the key feature and the variation coefficient to obtain the correlation coefficient comprises: Calculate the correlation of key characteristics and coefficient of variation: The key feature vector of the sth test is u s , the coefficient of change vector for the sth test is h s , the key feature vector u s The weight vector is Variation coefficient vector h s The weight vector is Transposed to T, the key eigenvector u s and the coefficient of variation vector h s The covariance matrix of Maximize the correlation and the expression is: The maximization function is According to the maximization function, the Lagrangian formula is constructed as follows: The key feature vector u s The characteristic value of Variation coefficient vector h s The characteristic value of Solve the Lagrangian formula to obtain the feature set with the maximum correlation, and calculate the correlation entropy of the key features in the feature set: The key feature vector u s and the coefficient of variation vector h s The correlation entropy is Key feature vector u s The parent node gene is x, and the coefficient vector h s The parent node gene is v, and the change coefficient vector is h s and the atomic sequence of the parent node gene v is h s →v, key feature vector u s and the atomic sequence of the parent node gene x is u s →x, atomic sequence h s →v and atomic sequence u s →x has a probability of appearing at the same time of p(u s →x,h s →v), atomic sequence h s →The probability of v appearing is p(u s →x); Corrected probability, expression is: The atomic sequence u s →The counter of x is Atomic sequence h s →v and atomic sequence u s →x simultaneous occurrence counter is The number of atomic sequences is M; Calculate the correlation coefficient by adjusting the probability correction: The correlation coefficient of the key feature u in the sth test is χ s (u), the number of key features in the sth test is n s , the variable control factor of the sth test is δ s , the correlation of the key feature u in the sth test after adjustment is The probability of the key feature u appearing is p(u), the norm is ‖·‖, and the correlation mean of the sth test is 6. The metal sheet stamping forming performance prediction method based on big data according to claim 1 is characterized in that: The method for constructing a metal sheet stamping forming performance prediction model according to the correlation coefficient comprises: The objective function is constructed according to the correlation coefficient, and the expression is: The objective function is No. The design parameters of stamping performance are The sth test The correlation coefficient of the uth key feature of the stamping forming performance is No. The predicted value of stamping performance is No. The actual value of stamping performance is The sheet metal stamping forming performance prediction model includes anomaly screening algorithm, gradient boosting algorithm, and machine learning algorithm; The anomaly screening algorithm screens the input data according to the degree of change to obtain variable data; The gradient boosting algorithm constructs multiple weak prediction models through variable data, gradually iterates and optimizes the minimization loss function to obtain feature data, evaluates the importance of features based on the frequency of use of feature data in splitting nodes in the weak prediction model, and obtains feature data sets; The machine learning algorithm is trained based on the historical data of the feature data set to obtain the changing rules of the stamping forming performance, and the stamping forming performance is predicted based on the changing rules.

7. The metal sheet stamping forming performance prediction method based on big data according to claim 1 is characterized in that: The method for optimizing the metal sheet stamping forming performance prediction model comprises: Introduce particle population to optimize parameters, calculate the fitness of particles, and take the position of the particle with the largest fitness as the search target; Calculate the particle positions during the search phase: The initial position of the particle of dimension d at the tth iteration is The upper bound of the optimization is X max , the optimal lower bound is X min , the change control parameter is β, the random numbers from 0 to 1 are r1 and r2 respectively, and the position of the first particle arriving at the tth iteration is The position of the ath particle in the dth dimension at the t+1th iteration is The particle position is updated using the expression: The search speed of the particle is ρ, and the position of the ath particle in the dth dimension at the tth iteration is The random numbers from 0 to 1 are r3 and r4 respectively, and the maximum number of iterations is t max , the updated position of the ath particle in the dth dimension at the t+1th iteration is The convergence factor is used to update the particle position, and the expression is: The corrected position of the ath particle in the dth dimension at the t+1th iteration is The convergence factor is η, the Cauchy operator is ξ, and the random number between 0 and 1 is r5; Update the convergence factor, the expression is: The initial value of the convergence factor is η bg , the control factor is ω, and the convergence factor is η; Continue to iterate until the maximum number of iterations is reached, otherwise update the convergence factor.

8. An electronic device comprising: processor; as well as A memory arranged to store computer executable instructions, wherein when the executable instructions are executed, the processor executes the method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, enables the electronic device to execute the method according to any one of claims 1 to 6.

Images