Gear precision correction method, device, equipment and storage medium
By building a gear database and optimizing the manufacturing process using machine learning algorithms, the nonlinear shrinkage problem in plastic gear manufacturing is solved, and high-precision and low-cost gear production is achieved, which is suitable for mass production.
Patent Information
- Application Number
- CN202411399402.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-09
- Publication Date
- 2025-01-03
- Estimated Expiration
- 2044-10-09
AI Technical Summary
In the prior art, nonlinear shrinkage caused by various factors in the manufacturing of plastic gears, it is difficult to form a unified shrinkage law and correction method, resulting in a long high-precision manufacturing cycle and high cost, and it is difficult to achieve mass production.
By constructing a database of gear deviation data, material characteristic data and manufacturing process data, using machine learning algorithms to establish deviation prediction models, optimize mold cavity and injection molding process parameters, realize gear deviation compensation, and reduce teeth by continuously updating models and strategies. The accuracy deviation of the distance.
Effectively reduce the nonlinear shrinkage of the gear, reduce the processing and forming time and manufacturing cost, and improve the accuracy and production efficiency of plastic gears.
Smart Images
Figure CN118917215B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of gear processing technology, and in particular to a gear precision correction method, device, equipment and storage medium. Background Art
[0002] At present, plastic gears are manufactured by injection molding, and there is nonlinear shrinkage in the manufacturing process. Factors that cause linear shrinkage include: the structure and parameters of plastic gears, mold design scheme, mold processing accuracy, mold assembly accuracy, injection molding process parameters and injection molding machine models and many other factors. In the prior art, although there are correction methods for single deviations such as pitch deviation, tooth profile deviation and helix deviation, there is a correlation between various deviations, which need to be considered comprehensively. At the same time, due to the large number of influencing factors, no statistically unified shrinkage law and correction method have been formed so far. This leads to a long manufacturing cycle and high cost for high-precision plastic gears, making it difficult to achieve mass production. Summary of the invention
[0003] The present application provides a gear precision correction method, device, equipment and storage medium for providing a production lead time to reduce the precision deviation of the gear, especially to reduce the precision deviation of the tooth pitch.
[0004] In a first aspect, an embodiment of the present application provides a gear precision correction method, the method comprising:
[0005] Gear deviation data, material property data and manufacturing process data are collected and sorted to establish a gear parameter database, wherein the gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth profile deviation parameters, and the manufacturing process data includes mold cavity parameters and injection molding process parameters;
[0006] Based on the gear parameter database, a machine learning algorithm is used to perform modeling and training to obtain a deviation prediction model;
[0007] According to the output result of the deviation prediction model, the mold cavity parameters and the injection molding process parameters are calculated and optimized to obtain a gear deviation compensation strategy;
[0008] Based on the gear deviation compensation strategy, the mold cavity parameters and the injection molding process parameters are adjusted and controlled to complete the gear deviation compensation;
[0009] Measure and analyze the gear samples after gear deviation compensation to obtain actual gear deviation data;
[0010] Based on the actual gear deviation data, the gear parameter database and the deviation prediction model are updated and optimized to update the gear deviation compensation strategy.
[0011] In a second aspect, an embodiment of the present application provides a gear precision correction device, the device comprising:
[0012] The first collection module is used to collect and organize gear deviation data, material property data and manufacturing process data to establish a gear parameter database, wherein the gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth profile deviation parameters, and the manufacturing process data includes mold cavity parameters and injection molding process parameters;
[0013] A model training module, used to perform modeling and training based on the gear parameter database using a machine learning algorithm to obtain a deviation prediction model;
[0014] A deviation compensation module, used to calculate and optimize the mold cavity parameters and the injection molding process parameters according to the output result of the deviation prediction model to obtain a gear deviation compensation strategy;
[0015] A parameter adjustment module, used to adjust and control the mold cavity parameters and the injection molding process parameters based on the gear deviation compensation strategy to complete the gear deviation compensation;
[0016] The second collection module is used to measure and analyze the gear sample after the gear deviation compensation is completed to obtain the actual gear deviation data;
[0017] An updating and optimization module is used to update and optimize the gear parameter database and the deviation prediction model based on the actual gear deviation data to update the gear deviation compensation strategy.
[0018] In a third aspect, an embodiment of the present application provides an electronic device, the electronic device comprising a memory and a processor;
[0019] The memory is used to store computer programs;
[0020] The processor is used to execute the computer program and implement the gear precision correction method as described in any one of the embodiments of the present application when executing the computer program.
[0021] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements a gear precision correction method as described in any one of the embodiments of the present application.
[0022] The embodiment of the present application provides a gear precision correction method, including: collecting and organizing gear deviation data, material property data and manufacturing process data, and establishing a gear parameter database, the gear deviation data including: pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters, and the manufacturing process data including: mold cavity parameters and injection molding process parameters; based on the gear parameter database, using a machine learning algorithm to perform modeling and training to obtain a deviation prediction model; according to the output result of the deviation prediction model, calculating and optimizing the mold cavity parameters and the injection molding process parameters to obtain a gear deviation compensation strategy; based on the gear deviation compensation strategy, adjusting and controlling the mold cavity parameters and the injection molding process parameters to complete gear deviation compensation; measuring and analyzing the gear samples after the gear deviation compensation to obtain actual gear deviation data; based on the actual gear deviation data, updating and optimizing the gear parameter database and the deviation prediction model to update the gear deviation compensation strategy. In the above method, a gear parameter database including gear deviation data, material property data and manufacturing process data is constructed. The gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters. A machine learning model is trained through the gear parameter database to obtain a gear deviation compensation strategy for the gear parameters. The data analysis capability of the large model is utilized to obtain an initial gear deviation compensation strategy to correct the manufacturing process parameters, and then the actual gear deviation data of the corrected gears are collected. The machine learning model is continuously updated and optimized to continuously update the gear deviation compensation strategy. The gear deviation compensation strategy gives a production lead time to reduce the nonlinear shrinkage of the gear, especially to reduce the accuracy deviation of the pitch, so as to reduce the processing time of the gear and reduce the manufacturing cost of the gear. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying any creative work.
[0024] Figure 1 A schematic flow chart of a gear precision correction method provided in an embodiment of the present application;
[0025] Figure 2 A schematic block diagram of a gear precision correction device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0027] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.
[0028] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0029] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0030] In the current deviation correction technology, there are different deviation correction methods for pitch deviation correction, tooth profile correction and helix correction. Because there is a certain correlation between different deviations, such as: pitch deviation has an impact on tooth profile deviation or helix deviation, but when formally correcting, the influence of pitch deviation (concentricity) is usually excluded first, the influence of pitch deviation is separated, and then the tooth profile deviation or helix deviation is corrected separately. Such non-comprehensive corrections are often difficult to reduce the accuracy deviation of gears with high precision and comprehensively.
[0031] There are many factors that affect the processing accuracy of plastic gears, and various factors will interact with each other, including:
[0032] Gear material: When the gear material is different, the shrinkage law will be completely different.
[0033] Production parameters: gear structure parameters and manufacturing process parameters.
[0034] The shrinkage law corresponding to each type of gear structural parameters is also different. For example, the gear structural parameters are: normal module 0.3~1.5mm, tooth width less than 15mm, tooth top circle diameter less than 70mm, and the maximum wall thickness of the gear is not more than 2.3mm.
[0035] The mold cavity parameters include: mold design parameters, mold processing accuracy and mold assembly accuracy. The injection molding process parameters include: injection molding speed parameters, injection molding temperature parameters and injection molding machine model.
[0036] Any change in the above factors may lead to different shrinkage results, and even if the correction method used is the same, plastic gears may produce different shrinkage laws under different manufacturing environments. Therefore, it is necessary to comprehensively consider the factors of the production process.
[0037] In order to more clearly introduce the technical solution of the present application, the technical solution of the present application will be introduced through specific embodiments below. It should be noted that the specific embodiments are used to expand the technical solution of the present application, but are not intended to limit the present application.
[0038] See also Figure 1 , Figure 1 The embodiment of the present application also provides a schematic flow chart of a gear precision correction method. Figure 1 As shown, an embodiment of the present application provides a method for controlling a pitch deviation of a gear, and the specific steps of the method include: S101-S106.
[0039] S101. Gear deviation data, material property data and manufacturing process data are collected and sorted to establish a gear parameter database. The gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters. The manufacturing process data includes mold cavity parameters and injection molding process parameters.
[0040] For example, in this application, since the contraction law will be completely different when the gear material is different, it is necessary to give a corresponding compensation result according to each material characteristic parameter in the material characteristic data. The error types are divided into three categories: tooth shape deviation, tooth surface deviation and pitch deviation.
[0041] Tooth profile deviation: For example, cylindrical gears. The overall shrinkage of cylindrical gears can be equivalent to a cylinder. According to the shrinkage principle of injection-molded plastic gears, errors such as head size, waist shrinkage, eccentricity, and non-verticality may occur based on the center hole (axis). Tooth profile deviation mainly includes tooth profile deviation and helix deviation. Tooth profile deviation includes: tooth profile inclination deviation, tooth profile shape deviation and tooth profile total deviation. Helix deviation includes: helix inclination deviation, helix shape deviation and helix total deviation.
[0042] Tooth surface deviation: Plastic gears are soft in texture, and tooth surface damage often occurs during transportation, usually in local locations. After analysis, it was found that because plastic gears are not smoothly ejected from the mold cavity, there is sometimes drag damage in the middle of the tooth profile surface in the direction of the tooth width, and generally all tooth profile surfaces and helix surfaces will experience this. When this occurs, the tooth profile surface deviation or helix surface deviation is large, and the probability of tooth profile error identification is the highest. In addition, when analyzing tooth surface deviation, roughness also needs to be considered. Roughness has a great influence on gear transmission efficiency and noise. By measuring the tooth profile and helix curve data, the roughness of the tooth surface can be statistically analyzed using the roughness statistical method.
[0043] Pitch deviation: single pitch deviation and cumulative pitch deviation.
[0044] According to the above-listed deviation types and manufacturing process parameters, data collection is performed to obtain production parameters, pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters.
[0045] S102. Based on the gear parameter database, a machine learning algorithm is used to perform modeling and training to obtain a deviation prediction model.
[0046] Exemplarily, a deviation prediction model is developed based on the gear parameter database established in step S101. The model uses a machine learning algorithm, such as a support vector machine (SVM) or a deep neural network (DNN), to process multi-dimensional input parameters. The input of the deviation prediction model includes: gear deviation data, material property data, manufacturing process data, etc., and the output is the predicted manufacturing process compensation parameters. During the model training process, supervised learning is performed using historical data in the gear parameter database, and the model parameters are continuously optimized through the back propagation algorithm. At the same time, a cross-validation mechanism is introduced to ensure the generalization ability of the model. In addition, the deviation prediction model should also include an adaptive learning module that can dynamically adjust the model parameters according to the newly added actual gear deviation data to improve the prediction accuracy.
[0047] It should be noted that a visual interface needs to be established to facilitate operators to input parameters and obtain prediction results.
[0048] In some embodiments, based on the gear parameter database, a machine learning algorithm is used for modeling and training to obtain a deviation prediction model, and the specific steps include: S1020-S1029.
[0049] S1020, standardize the data in the gear parameter database to obtain a standardized data set.
[0050] S1021. Use principal component analysis to perform dimensionality reduction processing on the standardized data set to obtain dimensionality reduction feature data.
[0051] In some embodiments, principal component analysis is used to perform dimensionality reduction processing on a standardized data set to obtain reduced-dimensionality feature data, including: constructing a multidimensional feature matrix based on the standardized data set; constructing a covariance matrix based on the multidimensional feature matrix, and performing eigenvalue decomposition on the covariance matrix to obtain a first feature sequence, the first feature sequence including: contribution eigenvalues and contribution eigenvectors corresponding to the contribution eigenvalues; sorting the first feature sequence from large to small according to the contribution eigenvalues to obtain a second feature sequence; accumulating the contribution eigenvalues of the second feature sequence in sequence, and stopping the accumulation when the accumulation reaches a preset accumulation threshold, and setting the accumulated contribution eigenvalues of the second feature sequence as target eigenvalues; generating a reduced-dimensionality projection matrix based on the contribution eigenvectors corresponding to the target eigenvalues; and obtaining the reduced-dimensionality feature data by multiplying the multidimensional feature matrix and the reduced-dimensionality projection matrix.
[0052] The above scheme uses principal component analysis (PCA) to reduce the dimensionality of the data, and obtains the contribution eigenvectors sorted by contribution eigenvalues by performing eigenvalue decomposition on the covariance matrix. The contribution eigenvectors whose cumulative contribution values reach the preset cumulative threshold are selected to form a dimensionality reduction projection matrix, which is multiplied with the original eigenmatrix to achieve data dimensionality reduction. This method can remove data redundancy and reduce computational complexity, while retaining the main feature information of the data and improving the efficiency and accuracy of subsequent machine learning models.
[0053] S1022. Randomly divide the dimension-reduced feature data to obtain a training set, a validation set, and a test set.
[0054] S1023. Based on the training set, a multi-layer perceptual neural network structure is constructed, and the number of network layers is set to 5, with the number of neurons in each layer being 100, 80, 60, 40, and 20 respectively.
[0055] S1024, initializing the multi-layer perception neural network, assigning network weights using the Xavier initialization method, and obtaining a first neural network model.
[0056] Exemplarily, when the Xavier initialization method is used, for each layer's weight W, a random sample is taken from a normal distribution with a mean of 0 and a variance of 1 / n, where n is the number of input neurons in the layer. This method can make the output variance of each layer approximately equal, which helps the gradient remain stable during back propagation. For the bias term, it is usually initialized to 0 or a small positive number. Through Xavier initialization, the convergence speed of the network can be effectively accelerated, and the risk of gradient disappearance or explosion can be reduced, thereby obtaining a good initial neural network model, laying the foundation for the subsequent training process.
[0057] S1025. Based on the first neural network model, the back propagation algorithm is used for model training, and the learning rate is set to 0.001, the batch size is 64, and the number of iterations is 1000.
[0058] S1026. After every 100 iterations, the model is evaluated using the validation set, the root mean square error is calculated, and the learning rate is dynamically adjusted according to the evaluation results to obtain the second neural network model.
[0059] S1027, perform pruning operation on the second neural network model, set the threshold to 0.01, delete the connections whose absolute weight values are less than the threshold, and obtain the third neural network model.
[0060] S1028. Use the test set to perform performance test on the third neural network model and calculate the prediction accuracy and F1 score.
[0061] S1029. When the prediction accuracy is greater than a preset accuracy threshold and the F1 score is greater than a preset score value, the third neural network is saved as a deviation prediction model.
[0062] The above model construction process systematically improves the accuracy and robustness of the deviation prediction model through multiple links such as data preprocessing, network structure design, and training strategy optimization, providing reliable data support for subsequent gear deviation compensation.
[0063] S103. According to the output result of the deviation prediction model, the mold cavity parameters and the injection molding process parameters are calculated and optimized to obtain the gear deviation compensation strategy.
[0064] Exemplarily, according to the output results of the deviation prediction model wheel developed in step S102, a corresponding gear deviation compensation strategy is formulated. First, the allowable range of gear deviation is set, and the upper and lower limits are determined according to the requirements of different accuracy levels. Then, based on the output results of the deviation prediction model, the specific value that needs to be compensated is calculated. The compensation strategy includes two aspects: one is mold design compensation, which pre-compensates for the expected shrinkage by adjusting the mold cavity size; the other is process parameter compensation, which controls the actual shrinkage by adjusting parameters such as injection pressure, holding time, and cooling time. At the same time, a compensation amount database is established to record the compensation effects under different parameter combinations to provide a basis for subsequent optimization.
[0065] S104. Based on the gear deviation compensation strategy, the mold cavity parameters and the injection molding process parameters are adjusted and controlled to complete the gear deviation compensation.
[0066] Exemplarily, gear deviation compensation is implemented according to the compensation strategy formulated in step S103. First, according to the mold design compensation plan, high-precision processing equipment is used to accurately modify the mold cavity to ensure that the dimensional error after modification is controlled within ±0.001mm. Secondly, according to the process parameter compensation plan, the relevant parameters of the injection molding machine are adjusted, including but not limited to: adjusting the melt temperature, injection pressure, holding time, cooling time, etc. In the process of implementing the compensation, a real-time monitoring system is used to continuously record the key parameters of the injection molding process, such as cavity pressure, temperature changes, etc. At the same time, a closed-loop control mechanism is introduced to dynamically adjust the injection molding parameters through real-time feedback from sensors to cope with possible fluctuations in the production process.
[0067] S105, measuring and analyzing the gear sample after the gear deviation compensation is completed to obtain actual gear deviation data.
[0068] Exemplarily, the gear samples produced after the compensation in step S104 are precisely measured. A high-precision gear measuring instrument is used to measure in accordance with the GB / T 10095-2008 standard, and the measurement accuracy is not less than 0.001 mm. The measurement items include single pitch deviation, adjacent pitch deviation, cumulative pitch deviation, etc. At the same time, a three-coordinate measuring machine is used to scan the overall shape of the gear to obtain the three-dimensional point cloud data of the gear. The measured data is compared with the theoretical design value to calculate the actual deviation value. Then, the measured data is processed using statistical analysis software to calculate the mean, standard deviation, distribution characteristics and other statistical quantities of the deviation to obtain the actual gear deviation data. In addition, a correlation analysis is performed to study the relationship between the pitch deviation and each process parameter, and identify the key factors affecting the deviation.
[0069] S106. Based on the actual gear deviation data, the gear parameter database and the deviation prediction model are updated and optimized to update the gear deviation compensation strategy.
[0070] Exemplarily, based on the measurement and analysis results in step S105, closed-loop optimization and continuous improvement are implemented. First, the actual gear deviation data is fed back to the parameter database in step S101 to update the database and improve the timeliness and representativeness of the data. Secondly, the prediction model in step S102 is retrained using the newly added data to improve the accuracy and adaptability of the model. Then, based on the analysis results, the compensation strategy in step S103 is optimized to adjust the compensation amount or compensation method. If it is found that certain process parameters or deviation types have a very high correlation with gear deviation, these parameters are optimized in a focused manner.
[0071] For example, through statistical analysis of deviation types, it can be found that pitch deviation occurs most frequently in deviation cases. Therefore, it can be seen that pitch deviation has the greatest impact on the deviation correction plan and needs to be analyzed in detail.
[0072] At the same time, we optimized the key control points in the production process, such as improving the mold structure, upgrading the injection molding equipment, etc. We established a continuous improvement mechanism, held technical review meetings regularly, discussed the optimization plan and formulated an improvement plan. Finally, we formulated new production standards and quality control specifications, solidified the optimization results into standard operating procedures, and ensured the continued effectiveness of the pitch deviation control method.
[0073] In some embodiments, based on actual gear deviation data, the gear parameter database and the deviation prediction model are updated and optimized to update the gear deviation compensation strategy. The specific steps include: S1061-S1067.
[0074] S1061. Statistically analyze the actual gear deviation data, calculate the mean, standard deviation and skewness of each deviation parameter, and obtain deviation statistical characteristics;
[0075] S1062. Based on the statistical characteristics of the deviation, a kernel density estimation method is used to fit the deviation distribution to obtain a deviation probability density function;
[0076] Exemplarily, a kernel density estimation method is used to fit the deviation distribution, wherein the kernel density estimation method is a Gaussian kernel function, and its specific formula is:
[0077] ;
[0078] Where h is the bandwidth parameter, and the optimal value can be determined by cross-validation method.
[0079] S1063, performing weighted update on the deviation data in the gear parameter database according to the deviation probability density function, setting the update weight to 0.7, and obtaining an updated gear parameter database;
[0080] Exemplarily, the specific formula of the deviation probability density function is:
[0081] .
[0082] Through statistical analysis and probability density estimation of actual gear deviation data, the deviation distribution characteristics can be described more accurately, providing a reliable data basis for subsequent model updates and strategy optimization.
[0083] S1064. Based on the updated gear parameter database, the deviation prediction model is updated online using an incremental learning algorithm, the learning rate is set to 0.01, the number of iterations is set to 500, and an updated deviation prediction model is obtained;
[0084] Exemplarily, an incremental learning algorithm is used to update the model parameters of the deviation prediction model online: using the stochastic gradient descent method, the specific formula for updating the model parameters is: ,in, for, is the learning rate, is the loss function.
[0085] Since there is already a deviation prediction model trained based on the original data, when new data arrives, the model is updated through the new actual gear deviation data, and these new data are used to update the existing model instead of retraining from scratch. During the update process, the parameters of the model are slightly adjusted according to the new data to adapt to the new information. Specifically, the learning rate is used to control the model's acceptance of new information. A smaller learning rate means that the model is more stable, but it adapts to new data more slowly; a larger learning rate is the opposite. Iterative update: Through multiple iterations (500 times), the model is gradually adjusted to better fit the overall distribution containing new and old data. Through the above method, there is no need to reprocess all historical data, and the model can be continuously optimized to adapt to minor changes in the production process, retaining historical data while adapting to new data trends.
[0086] By continuously updating the database, models and strategies, a closed-loop self-optimization system is formed, which can continuously improve the accuracy and stability of gear deviation control.
[0087] S1065, performing sensitivity analysis on the updated deviation prediction model, calculating the influence of each input parameter on the prediction result, and obtaining the parameter sensitivity ranking;
[0088] S1066, sorting the compensation parameters in the gear deviation compensation strategy according to the parameter sensitivity, setting the priority threshold to 0.1, and screening out high priority compensation parameters;
[0089] Through sensitivity analysis and priority sorting, the key parameters that have the greatest impact on deviations can be identified, thereby improving the pertinence and efficiency of compensation strategies.
[0090] S1067. Based on the high priority compensation parameters, the genetic algorithm is used to optimize the gear deviation compensation strategy. The population size is set to 100, the crossover probability is set to 0.8, the mutation probability is set to 0.1, and the number of iterations is set to 200 to obtain the updated gear deviation compensation strategy.
[0091] Global optimization: Using genetic algorithms to optimize gear deviation compensation strategies can find the global optimal solution in complex parameter space and avoid falling into local optimality.
[0092] The embodiment of the present application provides a gear precision correction method, including: collecting and organizing gear deviation data, material property data and manufacturing process data, and establishing a gear parameter database, the gear deviation data including: pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters, and the manufacturing process data including: mold cavity parameters and injection molding process parameters; based on the gear parameter database, using a machine learning algorithm to perform modeling and training to obtain a deviation prediction model; according to the output result of the deviation prediction model, calculating and optimizing the mold cavity parameters and the injection molding process parameters to obtain a gear deviation compensation strategy; based on the gear deviation compensation strategy, adjusting and controlling the mold cavity parameters and the injection molding process parameters to complete gear deviation compensation; measuring and analyzing the gear samples after the gear deviation compensation to obtain actual gear deviation data; based on the actual gear deviation data, updating and optimizing the gear parameter database and the deviation prediction model to update the gear deviation compensation strategy. In the above method, a gear parameter database including gear deviation data, material property data and manufacturing process data is constructed. The gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters. A machine learning model is trained through the gear parameter database to obtain a gear deviation compensation strategy for the gear parameters. The data analysis capability of the large model is utilized to obtain an initial gear deviation compensation strategy to correct the manufacturing process parameters, and then the actual gear deviation data of the corrected gears are collected. The machine learning model is continuously updated and optimized to continuously update the gear deviation compensation strategy. The gear deviation compensation strategy gives a production lead time to reduce the nonlinear shrinkage of the gear, especially to reduce the accuracy deviation of the pitch, so as to reduce the processing time of the gear and reduce the manufacturing cost of the gear.
[0093] See also Figure 2 , Figure 2 The embodiment of the present application further provides a schematic block diagram of a gear precision correction device, wherein the gear precision correction device 200 is used to execute the aforementioned gear precision correction method. The gear precision correction device 200 can be configured in a server.
[0094] Among them, the server can be an independent server or a server cluster, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN) and big data and artificial intelligence platforms.
[0095] like Figure 2 As shown, the gear precision correction device 200 includes: a first collection module 201, a model training module 202, a deviation compensation module 203, a parameter adjustment module 204, a second collection module 205 and an update optimization module 206.
[0096] The first collection module 201 is used to collect and organize gear deviation data, material property data and manufacturing process data to establish a gear parameter database. The gear deviation data includes: pitch deviation parameters, tooth surface deviation parameters and tooth shape deviation parameters. The manufacturing process data includes: mold cavity parameters and injection molding process parameters.
[0097] The model training module 202 is used to perform modeling and training based on the gear parameter database using a machine learning algorithm to obtain a deviation prediction model.
[0098] In some embodiments, when the model training module 202 is used to implement modeling and training based on the gear parameter database using a machine learning algorithm to obtain a deviation prediction model, it is specifically used to implement: standardizing the data in the gear parameter database to obtain a standardized data set; using the principal component analysis method to reduce the dimension of the standardized data set to obtain reduced-dimensional feature data; randomly dividing the reduced-dimensional feature data to obtain a training set, a validation set, and a test set; based on the training set, constructing a multi-layer perceptual neural network structure, setting the number of network layers to 5, and the number of neurons in each layer to 100, 80, 60, 40, and 20 respectively; initializing the multi-layer perceptual neural network, using the Xavier initialization method to assign network weights, and obtaining a first neural network. network model; based on the first neural network model, the back propagation algorithm is used for model training, the learning rate is set to 0.001, the batch size is 64, and the number of iterations is 1000; after every 100 iterations, the model is evaluated using the validation set, the root mean square error is calculated, and the learning rate is dynamically adjusted according to the evaluation results to obtain the second neural network model; the second neural network model is pruned, the threshold is set to 0.01, and the connections whose absolute weights are less than the threshold are deleted to obtain the third neural network model; the third neural network model is performance tested using the test set, and the prediction accuracy and F1 score are calculated; when the prediction accuracy is greater than the preset accuracy threshold, and the F1 score is greater than the preset score value, the third neural network is saved as a deviation prediction model.
[0099] In some embodiments, when the model training module 202 is used to implement dimensionality reduction processing on a standardized data set using principal component analysis to obtain reduced-dimensionality feature data, it is specifically used to implement: constructing a multidimensional feature matrix based on the standardized data set; constructing a covariance matrix based on the multidimensional feature matrix, and performing eigenvalue decomposition on the covariance matrix to obtain a first feature sequence, the first feature sequence including: contribution eigenvalues and contribution eigenvectors corresponding to the contribution eigenvalues; sorting the first feature sequence from large to small according to the contribution eigenvalues to obtain a second feature sequence; accumulating the contribution eigenvalues of the second feature sequence in sequence, and stopping the accumulation when the accumulation reaches a preset accumulation threshold, and setting the accumulated contribution eigenvalues of the second feature sequence as the target eigenvalues; generating a reduced-dimensionality projection matrix based on the contribution eigenvectors corresponding to the target eigenvalues; and obtaining reduced-dimensionality feature data by multiplying the multidimensional feature matrix and the reduced-dimensionality projection matrix.
[0100] The deviation compensation module 203 is used to calculate and optimize the mold cavity parameters and injection molding process parameters according to the output results of the deviation prediction model to obtain the gear deviation compensation strategy.
[0101] The parameter adjustment module 204 is used to adjust and control the mold cavity parameters and the injection molding process parameters based on the gear deviation compensation strategy to complete the gear deviation compensation.
[0102] The second collection module 205 is used to measure and analyze the gear sample after the gear deviation compensation is completed to obtain actual gear deviation data.
[0103] The update optimization module 206 is used to update and optimize the gear parameter database and the deviation prediction model based on the actual gear deviation data to update the gear deviation compensation strategy.
[0104] In some embodiments, when the update optimization module 206 is used to update and optimize the gear parameter database and the deviation prediction model based on the actual gear deviation data to update the gear deviation compensation strategy, it is specifically used to implement: statistical analysis of the actual gear deviation data, calculating the mean, standard deviation and skewness of each deviation parameter, and obtaining the deviation statistical characteristics; based on the deviation statistical characteristics, using the kernel density estimation method to fit the deviation distribution to obtain the deviation probability density function; according to the deviation probability density function, weighted update of the deviation data in the gear parameter database, setting the update weight to 0.7, and obtaining the updated gear parameter database; based on the updated gear parameter database, using the incremental learning algorithm The deviation prediction model is updated online by the method, the learning rate is set to 0.01, the number of iterations is set to 500, and the updated deviation prediction model is obtained; a sensitivity analysis is performed on the updated deviation prediction model, the influence of each input parameter on the prediction result is calculated, and the parameter sensitivity ranking is obtained; according to the parameter sensitivity ranking, the compensation parameters in the gear deviation compensation strategy are prioritized, the priority threshold is set to 0.1, and high priority compensation parameters are screened out; based on the high priority compensation parameters, the gear deviation compensation strategy is optimized by the genetic algorithm, the population size is set to 100, the crossover probability is set to 0.8, the mutation probability is set to 0.1, and the number of iterations is set to 200, and the updated gear deviation compensation strategy is obtained.
[0105] An embodiment of the present application provides an electronic device, which includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement a gear precision correction method such as any one of the embodiments of the present application when executing the computer program.
[0106] An embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements any gear precision correction method as described in the embodiments of the present application.
[0107] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.
Claims
1. A gear precision correction method, characterized in that: include: Gear deviation data, material property data and manufacturing process data are collected and sorted to establish a gear parameter database, wherein the gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth profile deviation parameters, and the manufacturing process data includes mold cavity parameters and injection molding process parameters; The data in the gear parameter database are standardized to obtain a standardized data set; a multidimensional feature matrix is constructed according to the standardized data set; a covariance matrix is constructed according to the multidimensional feature matrix, and the covariance matrix is eigenvalue decomposed to obtain a first feature sequence, wherein the first feature sequence includes: a contribution eigenvalue and a contribution eigenvector corresponding to the contribution eigenvalue; the first feature sequence is sorted from large to small according to the contribution eigenvalue to obtain a second feature sequence; the contribution eigenvalues of the second feature sequence are accumulated in sequence, and when the accumulation reaches a preset accumulation threshold, the accumulation is stopped, and the accumulated contribution eigenvalues of the second feature sequence are set as target eigenvalues; a dimensionality reduction projection matrix is generated according to the contribution eigenvector corresponding to the target eigenvalue; reduced dimensionality feature data is obtained by multiplying the multidimensional feature matrix and the reduced dimensionality projection matrix; the reduced dimensionality feature data is randomly divided to obtain a training set, a validation set and a test set; a multi-layer perceptron neural network structure is constructed based on the training set, and a network is set The number of layers is 5, and the number of neurons in each layer is 100, 80, 60, 40, and 20 respectively; the multilayer perceptual neural network is initialized, and the network weights are assigned using the Xavier initialization method to obtain a first neural network model; based on the first neural network model, the model is trained using the back propagation algorithm, and the learning rate is set to 0.001, the batch size is 64, and the number of iterations is 1000 times; after every 100 iterations, the model is evaluated using the validation set, the root mean square error is calculated, and the learning rate is dynamically adjusted according to the evaluation result to obtain a second neural network model; the second neural network model is pruned, the threshold is set to 0.01, and the connections with absolute weight values less than the threshold are deleted to obtain a third neural network model; the third neural network model is performance tested using the test set to calculate the prediction accuracy and F1 score; when the prediction accuracy is greater than the preset accuracy threshold, and the F1 score is greater than the preset score value, the third neural network model is saved as a deviation prediction model; According to the output result of the deviation prediction model, the mold cavity parameters and the injection molding process parameters are calculated and optimized to obtain a gear deviation compensation strategy; Based on the gear deviation compensation strategy, the mold cavity parameters and the injection molding process parameters are adjusted and controlled to complete the gear deviation compensation; Measure and analyze the gear samples after gear deviation compensation to obtain actual gear deviation data; The actual gear deviation data is statistically analyzed, and the mean, standard deviation and skewness of each deviation parameter are calculated to obtain the deviation statistical characteristics; based on the deviation statistical characteristics, the kernel density estimation method is used to fit the deviation distribution to obtain the deviation probability density function; according to the deviation probability density function, the deviation data in the gear parameter database is weighted updated, and the update weight is set to 0.7 to obtain the updated gear parameter database; based on the updated gear parameter database, the incremental learning algorithm is used to update the deviation prediction model online, and the learning rate is set to 0.01 and the number of iterations is 500. An updated deviation prediction model is obtained; a sensitivity analysis is performed on the updated deviation prediction model, the influence of each input parameter on the prediction result is calculated, and a parameter sensitivity ranking is obtained; according to the parameter sensitivity ranking, the compensation parameters in the gear deviation compensation strategy are prioritized, the priority threshold is set to 0.1, and high-priority compensation parameters are screened out; based on the high-priority compensation parameters, a genetic algorithm is used to optimize the gear deviation compensation strategy, the population size is set to 100, the crossover probability is set to 0.8, the mutation probability is set to 0.1, and the number of iterations is set to 200 times, and an updated gear deviation compensation strategy is obtained.
2. The gear accuracy correction method according to claim 1, characterized in that: The mold cavity parameters include: mold design parameters, mold processing accuracy and mold assembly accuracy; the injection molding process parameters include: injection molding speed parameters, injection molding temperature parameters and injection molding machine model.
3. The gear accuracy correction method according to claim 1, characterized in that: The pitch deviation parameters include: single pitch deviation parameters and cumulative pitch deviation parameters.
4. A gear precision correction device, characterized in that: The gear precision correction device comprises: The first collection module is used to collect and organize gear deviation data, material property data and manufacturing process data to establish a gear parameter database, wherein the gear deviation data includes pitch deviation parameters, tooth surface deviation parameters and tooth profile deviation parameters, and the manufacturing process data includes mold cavity parameters and injection molding process parameters; The model training module is used to perform standardization processing on the data in the gear parameter database to obtain a standardized data set; construct a multidimensional feature matrix according to the standardized data set; construct a covariance matrix according to the multidimensional feature matrix, perform eigenvalue decomposition on the covariance matrix to obtain a first feature sequence, wherein the first feature sequence includes: a contribution eigenvalue and a contribution eigenvector corresponding to the contribution eigenvalue; sort the first feature sequence from large to small according to the contribution eigenvalue to obtain a second feature sequence; accumulate the contribution eigenvalues of the second feature sequence in sequence, and stop accumulating when the accumulation reaches a preset accumulation threshold, and set the accumulated contribution eigenvalues of the second feature sequence as the target eigenvalue; generate a dimensionality reduction projection matrix according to the contribution eigenvector corresponding to the target eigenvalue; obtain dimensionality reduction feature data by multiplying the multidimensional feature matrix and the dimensionality reduction projection matrix; randomly divide the dimensionality reduction feature data to obtain a training set, a validation set and a test set; and construct a multi-layer perceptual neural network structure based on the training set. The number of network layers is set to 5, and the number of neurons in each layer is 100, 80, 60, 40, and 20 respectively; the multilayer perceptual neural network is initialized, and the network weights are assigned using the Xavier initialization method to obtain a first neural network model; based on the first neural network model, the model is trained using the back propagation algorithm, and the learning rate is set to 0.001, the batch size is 64, and the number of iterations is 1000 times; after every 100 iterations, the model is evaluated using the validation set, the root mean square error is calculated, and the learning rate is dynamically adjusted according to the evaluation results to obtain a second neural network model; the second neural network model is pruned, the threshold is set to 0.01, and the connections whose absolute weight value is less than the threshold are deleted to obtain a third neural network model; the third neural network model is performance tested using the test set to calculate the prediction accuracy and F1 score; when the prediction accuracy is greater than the preset accuracy threshold, and the F1 score is greater than the preset score value, the third neural network model is saved as a deviation prediction model; A deviation compensation module, used to calculate and optimize the mold cavity parameters and the injection molding process parameters according to the output results of the deviation prediction model to obtain a gear deviation compensation strategy; A parameter adjustment module, used to adjust and control the mold cavity parameters and the injection molding process parameters based on the gear deviation compensation strategy to complete the gear deviation compensation; The second collection module is used to measure and analyze the gear sample after the gear deviation compensation is completed to obtain the actual gear deviation data; The update optimization module is used to perform statistical analysis on the actual gear deviation data, calculate the mean, standard deviation and skewness of each deviation parameter, and obtain the deviation statistical characteristics; based on the deviation statistical characteristics, the kernel density estimation method is used to fit the deviation distribution to obtain the deviation probability density function; according to the deviation probability density function, the deviation data in the gear parameter database is weighted updated, and the update weight is set to 0.7 to obtain the updated gear parameter database; based on the updated gear parameter database, the incremental learning algorithm is used to update the deviation prediction model online, and the learning rate is set to 0.01 and the number of iterations is set to 5 00 times to obtain an updated deviation prediction model; perform sensitivity analysis on the updated deviation prediction model, calculate the influence of each input parameter on the prediction result, and obtain the parameter sensitivity ranking; according to the parameter sensitivity ranking, prioritize the compensation parameters in the gear deviation compensation strategy, set the priority threshold to 0.1, and screen out high-priority compensation parameters; based on the high-priority compensation parameters, use a genetic algorithm to optimize the gear deviation compensation strategy, set the population size to 100, the crossover probability to 0.8, the mutation probability to 0.1, and the number of iterations to 200 times to obtain an updated gear deviation compensation strategy.
5. An electronic device, characterized in that: The electronic device comprises a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the gear precision correction method as described in any one of claims 1 to 3 when executing the computer program.
6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the gear precision correction method according to any one of claims 1 to 3.