A heat treatment quality pre-judgment analysis system and a pre-judgment analysis method

Abstract

The present application relates to a kind of heat treatment quality pre-judgment analysis system and pre-judgment analysis method, system includes data processing module, heat treatment quality pre-judgment module, pre-judgment result evaluation module, kiss recommendation module and push optimization suggestion module;Method applies above-mentioned system, input is to be pre-judged workpiece information, data processing module is handled to form data set to be pre-judged workpiece information, heat treatment quality pre-judgment module exports quality pre-judgment result according to data set, evaluation module carries out evaluation to quality pre-judgment result, kiss recommendation module exports the product details of the product characteristic parameter in process sample library with the top five of the degree of agreement of to-be-judged workpiece product, push optimization suggestion module exports the best matching heat treatment process data model according to the selected workpiece emphasis requirement.The present application collects heat treatment process historical information data, carries out big model learning, carries out heat treatment result pre-judgment according to the provided workpiece information, solves the problem that heat treatment process is difficult to accurately predict product quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a heat treatment quality pre-judgment analysis system and a pre-judgment analysis method. BACKGROUND

[0002] With the transformation of manufacturing industry from traditional experience type to modern digitization, modeling and intelligentization, it is an inevitable trend to solve many quality problems in production at low cost and high quality by means of industrial database, data analysis and process simulation technology. As a key link affecting product quality, the quantitative relationship between heat treatment process data and product quality is complex, and it is difficult to determine efficiently and accurately by relying on traditional experience type means.

[0003] Therefore, based on big data modeling, the present application provides a heat treatment quality pre-judgment analysis system, which fully utilizes the technology of cognitive intelligence in mass data processing, efficiently solves the pain point of difficult accurate prediction of product quality in heat treatment process, and helps the control of heat treatment quality. SUMMARY

[0004] In view of the above problems, the present application provides a heat treatment quality pre-judgment analysis system and a pre-judgment analysis method, and the specific technical scheme is as follows:

[0005] A heat treatment quality pre-judgment analysis system, the construction steps are as follows:

[0006] S1, historical data acquisition and construction of heat treatment process data model based on product chemical composition-process parameters-mechanical property corresponding relationship;

[0007] S2, constructing a data processing module based on principal component analysis algorithm, determining the key influence factors of quality and extracting the process parameter bandwidth, processing the heat treatment process data model and establishing a process sample library;

[0008] S3, constructing a heat treatment quality pre-judgment module based on random forest algorithm, the specific construction steps are as follows,

[0009] S3.1, randomly selecting n samples from the process sample library as the training set of a single decision tree by using the resampling method;

[0010] S3.2, constructing a decision tree; starting from the root node, according to the selected best split feature and threshold, splitting the training data set into two sub-data sets; for each sub-data set, recursively performing the splitting process until the maximum depth of the tree is 5;

[0011] S3.3, repeating the process of training set selection and decision tree construction to obtain multiple decision trees, which together constitute a random forest;

[0012] S3.4, input new data set, take the average of multiple decision tree prediction values, output heat treatment quality pre-judgment result;

[0013] S3.5, repeat S2.1-S2.4 until the pre-judgment result meets the requirements and save the model weight;

[0014] S4, based on the comprehensive evaluation method, a quality pre-judgment result evaluation module is constructed, the performance parameters of the quality pre-judgment result are taken as evaluation indexes, the original quality pre-judgment result data set is normalized, different dimension index values are converted into dimensionless scores, and the comprehensive evaluation score of the heat treatment quality pre-judgment result is obtained;

[0015] S5, based on KNN near neighbor algorithm, an anastomosis recommendation module is constructed, and the top five product details in the process sample library that match the product feature parameters of the to-be-predicted workpiece are output;

[0016] S6, a push recommendation module is constructed, the workpiece performance is input, matching is performed in the process sample library, and the heat treatment process data model with the highest matching degree is output.

[0017] Further, S2 includes the following steps:

[0018] S2.1, standardizing the heat treatment process parameter features, converting the data into a distribution with a mean of 0 and a standard deviation of 1;

[0019] S2.2, for n heat treatment process parameter features, an n x n covariance matrix is constructed;

[0020] S2.3, performing eigenvalue decomposition on the covariance matrix C to obtain eigenvectors and eigenvalues;

[0021] S2.4, sorting the eigenvalues according to their size, and selecting the eigenvectors corresponding to the top 10 largest eigenvalues as principal components;

[0022] S2.5, using the selected 10 principal components as new coordinate axes, projecting the original heat treatment process parameter data onto these coordinate axes to obtain a new data set and key influencing factors;

[0023] S2.6, performing maximum, minimum, mean, and standard deviation statistical analysis on the reduced data to determine the data range bandwidth of the key process parameters.

[0024] Further, in S3.2, when each node of the decision tree is split, 10 features are randomly selected from all features, and one best split feature is selected from the 10 features for splitting.

[0025] Further, in S4, for the physical workpieces that have started production, the deviation of process data from the preset process data is executed as a deduction item; for the physical workpieces that have not started production and the virtual workpieces, the matching degree of the feature information with the data in the process data model is also taken as a scoring item.

[0026] Further, in S5, the Euclidean distance between the process parameters of the input workpiece and each sample in the process sample library is calculated.

[0027] Further, in S6, the user selectively inputs the workpiece information and performance requirements, the system matches the input conditions with the information in the process sample library, and outputs the best-matched hot processing process data model according to the weight value and the matching degree.

[0028] A hot treatment quality pre-judgment analysis method, the application of the above-mentioned hot treatment quality pre-judgment analysis system, the information of the workpiece to be pre-judged is input, the data processing module processes the information of the workpiece to be pre-judged to form a data set, the hot treatment quality pre-judgment module outputs the quality pre-judgment result according to the data set, the evaluation module evaluates the quality pre-judgment result, the fitting recommendation module outputs the product details of the top five product feature parameters in the process sample library that are consistent with the workpiece product to be pre-judged, and the optimization suggestion module outputs the best-matched hot treatment process data model according to the selected workpiece requirement.

[0029] The beneficial effects of the present application are as follows:

[0030] The present application collects hot treatment process historical information data, performs large model learning, and pre-judges the hot treatment result according to the provided workpiece information, thereby solving the problem that the product quality is difficult to accurately pre-judge in the hot treatment process, and helping to control the hot treatment quality. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0032] Figure 1 is a schematic diagram of the hot treatment quality pre-judgment analysis method of the present application;

[0033] Figure 2 is a schematic diagram of the actual workpiece processing of the present application;

[0034] Figure 3 is a schematic diagram of the virtual workpiece processing of the present application. DETAILED DESCRIPTION

[0035] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" and the like are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined with "first", "second" and the like can be explicitly or implicitly included one or more. In the description of the present application, unless otherwise stated, the meaning of "multiple" is two or more.

[0036] The present application provides the following specific embodiments:

[0037] In the first aspect, the present application provides a heat treatment quality prediction analysis system, and the construction steps are as follows:

[0038] S1, data acquisition: a large amount of multi-dimensional data is collected from actual production, including production process data, process data, quality data, equipment data and the like, which comprehensively reflects various factors of the heat treatment process.

[0039] The collected multi-dimensional historical data is deposited in the data center, which follows the data management concept and requirements of DAMA and DCMM, and is guided by data planning and management consulting methods to carry out heat processing process parameter data architecture design and data model design, realize the unified collection and management of multi-source heterogeneous data, complete the construction of heat processing process parameter data warehouse. According to the management activities of data standard, data quality and data security, metadata management, etc., deposit heat processing process parameter data assets, support data subscription and service publishing based on asset directory, realize process monitoring of the system, and have diversified analysis ability of business, data and technology, can build enterprise-level service ability, based on platform data, further build a large database.

[0040] Based on the data center and the corresponding relationship between the historical product chemical composition-process parameters-mechanical properties, a heat processing process data model is established, and a related type process sample library is established, which contains heat treatment process parameters and corresponding mechanical property data of different types of workpieces, providing basis for subsequent data analysis and quality prediction.

[0041] S2, based on the principal component analysis algorithm, a data processing module is constructed, the key influence factors of quality are determined, and the process parameter bandwidth is extracted, the heat processing process data model is processed, and a process sample library is established.

[0042] The specific steps are as follows, S2.1, standardizing the heat treatment process parameter characteristics, that is, converting the data into a distribution with a mean of 0 and a standard deviation of 1, and the calculation formula is as follows:

[0043]

[0044] Where μ is the mean of all sample data, and σ is the standard deviation of all sample data.

[0045] S2.2, for n heat treatment process parameter characteristics, an n*n covariance matrix C is constructed, where each element C ij represents the covariance between features i and j, where the heat treatment process parameter set is assumed to be a three-dimensional data set of x, y, and z, and the covariance matrix is a 3*3 matrix as shown in the following figure:

[0046]

[0047] If the covariance is positive, the two parameter variables of the heat treatment process are positively correlated, and if the covariance is negative, the two parameters of the heat treatment process are negatively correlated.

[0048] S2.3, perform eigenvalue decomposition on the covariance matrix C to obtain eigenvectors and eigenvalues.

[0049] S2.4, sort the eigenvalues according to their size, and select the eigenvectors corresponding to the top 10 largest eigenvalues as the principal components.

[0050] S2.5, use the selected 10 principal components (i.e., eigenvectors) as new coordinate axes to project the original heat treatment process parameter data onto these coordinate axes to obtain a new data set and key influencing factors.

[0051] S2.6, perform maximum, minimum, mean, and standard deviation statistical analysis on the reduced dimension data to determine the data range bandwidth of the key process parameters.

[0052] S3, build a heat treatment quality prediction module based on the random forest algorithm, and the specific construction steps are as follows:

[0053] S3.1, randomly select n samples from the process sample library as the training set of a single decision tree using the sampling method with replacement.

[0054] S3.2, construct a decision tree; starting from the root node, split the training data set into two sub-data sets according to the selected best split feature and threshold; for each sub-data set, recursively perform the split process until the maximum depth of the tree is 5. When each node of the decision tree is split, 10 features are randomly selected from all features, and one of the 10 features is selected as the best split feature for splitting.

[0055] S3.3, repeat the process of training set selection and decision tree construction to obtain multiple decision trees which together constitute a random forest.

[0056] S3.4, input a new data set, take the average of the prediction values of multiple decision trees, and output the heat treatment quality pre-judgment result.

[0057] S3.5, repeat S2.1-S2.4 until the pre-judgment result meets the requirements and the model weight is saved.

[0058] S4, construct a quality pre-judgment result evaluation module based on the comprehensive evaluation method, take the hardness, impact toughness, fatigue strength, metallographic structure, decarburization depth, deformation, cracking, overheating and overburning degree and other performance parameters of the quality pre-judgment result as evaluation indexes, normalize the original quality pre-judgment result data set, convert the index values of different dimensions into dimensionless scores, and obtain the comprehensive evaluation score of the heat treatment quality pre-judgment result. For physical workpieces that have started production, deviations in process data from pre-set process data are used as deduction items; for physical workpieces and virtual workpieces that have not started production, the matching degree of feature information and process data model data is also used as a scoring item, and the scores of all indexes are added to obtain the comprehensive evaluation score of the heat treatment quality pre-judgment result.

[0059] Finally, output the quality pre-judgment result, including performance prediction result, quality pre-judgment actual score and effect evaluation, and the item with the most points lost.

[0060] S5, construct a matching recommendation module based on KNN nearest neighbor algorithm, output the top five product details in the process sample library that match the product feature parameters of the workpiece to be pre-judged, as follows: for process parameter input samples, including material grade, chemical composition, process information (such as quenching / normalizing temperature, holding time, cooling method, sample type), calculate the Euclidean distance between the input process parameters and each sample in the historical heat process parameter data set. The formula is as follows:

[0061]

[0062] Let K = 5, that is, output the top 5 product details in terms of product feature parameter matching degree, the system supports direct viewing of the main feature parameters, and clicking can view the actual process and quality information of the whole process.

[0063] S6, a push optimization suggestion module is constructed, the workpiece focuses on performance is input, matching is carried out in the process sample library, and the highest matching degree of the hot working process data model is output. Specifically, the user can select one of the input material brand or chemical composition, and one of the hardness requirement or performance requirement (including sample type), or all condition items can be input. The system matches the input condition with the information in the process sample library, and according to the weight value and the matching degree, the best matching hot working process data model is output.

[0064] In a second aspect, as shown in Figure 1 The present application provides a heat treatment quality prediction analysis method, which applies the heat treatment quality prediction analysis system, inputs the workpiece information to be predicted, the workpiece information includes material brand, chemical composition, process information (quenching / normalizing temperature, holding time, cooling method, sample type), part number (only for physical workpiece), hardness and physical and chemical performance standard, a data processing module processes the workpiece information to be predicted to form a data set, a heat treatment quality prediction module outputs a quality prediction result according to the data set, an evaluation module evaluates the quality prediction result, a fitting recommendation module outputs the top five product details in the process sample library that are most consistent with the product characteristic parameters of the workpiece to be predicted, and a push optimization suggestion module outputs the best matching heat treatment process data model according to the selected workpiece focus requirement.

[0065] As shown in Figure 2 The input workpiece is an actual workpiece (a workpiece that has started production and implementation and a workpiece that has not started production and implementation), process production information of the physical workpiece is obtained; the executed nodes are displayed when the prediction result is output; the input information of the physical workpiece, the actual execution information and the actual typical product data bandwidth information in the process data model are matched; the quality prediction result information such as hardness result and mechanical property result is output; the prediction result is comprehensively scored; the top five product details with the highest product characteristic parameter consistency are output; the actual process and quality information of the whole process is viewed by clicking; whether the predicted hardness and physical and chemical performance meet the requirements is obtained by comparing the predicted hardness and physical and chemical performance with the standard requirements; the deviation of the execution process data from the preset process data is taken as a deduction item; the final quality prediction result is output; the item with the most deductions is output; the push optimization suggestion is output: the material brand, chemical composition, hardness requirement and performance requirement of the physical workpiece are input; the input condition is matched with the typical category material / hardness / performance data bandwidth information in the process optimization model; the best matching typical category furnace matching process and process execution information can be used as the process push optimization basic data; the push optimization process parameters such as quenching / normalizing temperature range, holding time, cooling method, tempering temperature range, holding time, etc. are output.

[0066] As shown in Figure 3As shown, the input workpiece is a virtual workpiece (a different type of workpiece that is not manufactured into a physical entity, but has a quality prediction requirement), the material grade, chemical composition, virtual process information (quenching / normalizing temperature, holding time, cooling method, sample type), hardness, and physicochemical performance standard of the virtual workpiece are input; the input information is systematically matched with the actual typical product data bandwidth information or the calculated data bandwidth information in the process data model; the hardness result, mechanical property result, and other quality prediction result information are output; the prediction results are comprehensively scored; the product characteristics parameter of the 5 products with the highest matching degree are output; after clicking, the actual process and quality information of the whole process can be viewed; the predicted hardness and physicochemical performance result are compared with the standard requirement to determine whether the requirement is met; the matching degree of the characteristic information with the data in the process data model is also used as a scoring item; the final quality prediction result is output; the item with the most lost points is output; the optimization suggestion is output: the material grade, chemical composition, hardness requirement, and performance requirement of the virtual workpiece are input; the input conditions are matched with the typical category material / hardness / performance data bandwidth information in the process optimization model; the best matching typical category furnace matching process and process execution information can be used as the process optimization basic data; the optimization process parameters, such as quenching / normalizing temperature range, holding time, cooling method, tempering temperature range, holding time, etc., are output.

[0067] By collecting heat treatment process historical information data, large model learning is performed, and heat treatment result prediction is performed according to the provided workpiece information, so that the problem of difficult accurate prediction of product quality in the heat treatment process is solved, and the heat treatment quality control is facilitated.

[0068] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application.

Claims

1. A heat treatment quality prediction and analysis system, characterized in that: The construction steps are as follows: S1. Collect historical data and construct a thermal processing data model based on the correspondence between product chemical composition, process parameters and mechanical properties; S2. Construct a data processing module based on principal component analysis algorithm, determine key quality influencing factors and extract process parameter bandwidth, process the hot processing process data model and establish a process sample library; S3. Construct a heat treatment quality prediction module based on the random forest algorithm. The specific construction steps are as follows. S3.

1. Using sampling with replacement, n samples are randomly selected from the process sample library as the training set for a single decision tree. S3.2 Construct a decision tree; Starting from the root node, split the training dataset into two subsets based on the selected best splitting feature and threshold; For each subset, recursively perform the splitting process until the maximum depth of the tree is 5; S3.3, Repeat the process of selecting the training set and constructing the decision tree to obtain multiple decision trees, which together form a random forest; S3.4 Input a new dataset, take the average of the predictions from multiple decision trees, and output the heat treatment quality prediction result; S3.5 Repeat S2.1-S2.4 until the prediction result meets the requirements and save the model weights; S4. Construct a quality prediction result evaluation module based on the comprehensive evaluation method. Use the performance parameters of the quality prediction result as the evaluation index, normalize the original quality prediction result dataset, and convert the index values ​​of different dimensions into dimensionless scores to obtain the comprehensive evaluation score of the heat treatment quality prediction result. S5. Construct a matching recommendation module based on the KNN nearest neighbor algorithm, and output the details of the top five products in the process sample library that have the highest matching degree with the feature parameters of the workpiece to be predicted. S6. Construct an optimization suggestion module, input the workpiece's key performance characteristics, match them in the process sample library, and output the hot working process data model with the highest matching degree.

2. The heat treatment quality prediction and analysis system according to claim 1, characterized in that: S2 includes the following steps: S2.1 Standardize the characteristics of heat treatment process parameters and convert the data into a distribution with a mean of 0 and a standard deviation of 1. S2.2 For n heat treatment process parameter characteristics, construct an n×n covariance matrix; S2.3 Perform eigenvalue decomposition on the covariance matrix C to obtain eigenvectors and eigenvalues; S2.4 Sort the eigenvalues ​​according to their size and select the eigenvectors corresponding to the 10 largest eigenvalues ​​as principal components; S2.5 Using the selected 10 principal components as new coordinate axes, project the original heat treatment process parameter data onto these coordinate axes to obtain a new dataset and key influencing factors; S2.6 Perform statistical analysis on the maximum, minimum, mean, and standard deviation of the dimensionality-reduced data to determine the data range bandwidth of key process parameters.

3. The heat treatment quality prediction and analysis system according to claim 1, characterized in that: In S3.2, when each node of the decision tree splits, 10 features are randomly selected from all features, and then the best splitting feature is selected from these 10 features for splitting.

4. The heat treatment quality prediction and analysis system according to claim 1, characterized in that: In S4, for physical workpieces that have already started production, deviations from the preset process data are used as a deduction item; for physical workpieces and virtual workpieces that have not yet started production, the degree of matching between the feature information and the data in the process data model is also used as a scoring item.

5. The heat treatment quality prediction and analysis system according to claim 1, characterized in that: In S5, the Euclidean distance between the process parameters of the input workpiece and each sample in the process sample library is calculated.

6. The heat treatment quality prediction and analysis system according to claim 1, characterized in that: In S6, users selectively input workpiece information and performance requirements. The system matches the input conditions with information in the process sample library and outputs the best-matched hot working process data model based on the weight value and matching degree.

7. A method for predicting and analyzing the quality of heat treatment, using the heat treatment quality prediction and analysis system described in any one of claims 1-6, characterized in that: The workpiece information to be predicted is input, the data processing module processes the workpiece information to form a dataset, the heat treatment quality prediction module outputs the quality prediction result based on the dataset, the evaluation module evaluates the quality prediction result, the matching recommendation module outputs the details of the top five products in the process sample library that match the product feature parameters of the workpiece to be predicted, and the optimization suggestion module outputs the best matching heat treatment process data model based on the selected workpiece's emphasis requirements.

Images