Steel performance prediction system and method based on machine learning

Abstract

The present invention discloses a steel performance prediction system and method based on machine learning, which relates to the technical field of steel performance prediction, including obtaining basic information of steel, obtaining a steel production impact index based on the basic information of steel and based on the steel production environment assessment, classifying steel according to the steel production impact index, and obtaining first classification information of steel. The present invention evaluates the process nodes in the steel production process through the process node impact index, analyzes the degree of influence of different process nodes on steel performance, classifies steel under different production environments through the steel production impact index, improves the efficiency of steel performance prediction, further divides steel through the steel performance characteristic performance index, improves the accuracy of steel performance prediction, optimizes steel performance through the steel performance prediction model, effectively reduces the trial and error cost and production time, and improves production efficiency and cost-effectiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of steel performance prediction, and in particular to a steel performance prediction system and method based on machine learning. Background Art

[0002] With the continuous development of steel, steel is widely used in construction, machinery, automobiles, ships and other industrial fields. Among them, 70% of steel is used for construction, and the rest is used for various shafts, nuts and springs. Therefore, steel is of great significance to the development of the steel industry. Nowadays, my country's steel development has made great progress. Not only the production scale, but also the production process has made continuous breakthroughs. The production quality of my country's steel has been significantly improved compared with the previous quality. With the increasing demand for steel in my country's industrial development, the requirements for its performance have become more and more stringent.

[0003] At present, there are still some problems in predicting steel properties, such as being unable to accurately evaluate the impact of production process nodes on steel properties during the production of different types of steel, being unable to classify steels according to steel production process parameters, being unable to accurately analyze steel surface characteristics, being unable to accurately select steels with similar steel properties, and being unable to accurately predict steel properties. Summary of the invention

[0004] In order to solve the above technical problems, a steel performance prediction system and method based on machine learning are provided. The technical solution solves the problems raised in the above background technology that it is impossible to accurately evaluate the influence of production process nodes on steel performance in the production process of different types of steel, it is impossible to classify steels according to steel production process parameters, it is impossible to accurately analyze the surface characteristics of steels, it is impossible to accurately select steels with similar steel performance, and it is impossible to accurately predict steel performance.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] Steel performance prediction methods based on machine learning include:

[0007] Obtaining basic information about steel materials, wherein the basic information about steel materials includes information about steel material types and information about steel material production;

[0008] According to the basic information of steel, based on the assessment of steel production environment, obtain the steel production impact index;

[0009] Classify steel materials according to the steel production impact index and obtain the first classification information of steel materials;

[0010] Identify the surface characteristics of steel and obtain the surface characteristic data of steel;

[0011] According to the surface characteristic data of steel, the performance characteristic index of steel is obtained;

[0012] According to the first classification information of steel materials and the performance index of steel material performance characteristics, the second classification information of steel materials is obtained;

[0013] Acquire historical data of steel materials, wherein the historical data of steel materials include historical test data of steel material properties, data of steel material types and historical surface characteristic data of steel materials;

[0014] According to the historical data of steel, based on data processing, a training set of historical data of steel is obtained;

[0015] According to the historical data training set of steel, based on model training, a steel performance prediction model is obtained;

[0016] According to the second classification information of steel and the steel performance prediction model, the steel performance prediction data is obtained.

[0017] Preferably, the step of classifying steel materials according to the steel production impact index and obtaining first classification information of steel materials specifically includes:

[0018] According to the basic information of steel, steel production information is obtained, wherein the steel production information includes steel raw material information and steel production process information;

[0019] Acquire historical steel production data, wherein the historical steel production data includes historical steel performance data and historical steel production process node parameter information;

[0020] Based on the historical production data of steel, obtain the information on the impact of process nodes on steel performance;

[0021] According to the steel performance process node influence information, process node vector information is obtained, wherein the process node vector information includes a front-end control process node and a back-end influence process node of each process node;

[0022] Obtain the process node impact index based on process node vector information and historical steel performance data;

[0023] Obtain steel production impact index based on process node impact index and steel production information;

[0024] According to the steel production impact index, obtain the first classification information of steel;

[0025] Among them, the calculation formula of the process node impact index is:

[0026]

[0027] Where Q(x) is the process node impact index of the x-th process node, α iis the influence coefficient of the i-th front-end control process node on the process node, τ x is the steel performance influence coefficient of the xth process node, β j is the influence coefficient of the node on the jth back-end influencing process node, n is the total number of front-end controlled process nodes of the xth process node, and m is the total number of back-end influencing process nodes of the xth process node.

[0028] Preferably, obtaining the first classification information of steel materials according to the steel production impact index specifically includes:

[0029] Obtain steel type information based on basic steel information;

[0030] Based on the steel type information, steel is classified to obtain basic classification information of steel;

[0031] According to the steel production information, obtain the steel production process node information;

[0032] According to the steel type information, obtain the standard parameter information of the process node;

[0033] Obtain the steel production impact index according to the process node impact index, process node standard parameter information and steel production process node information;

[0034] Sort steel materials according to the steel production impact index and obtain steel material sorting information;

[0035] Based on the steel performance prediction requirements, the difference threshold of the steel production impact index is obtained;

[0036] Obtain the first classification information of steel materials according to the steel material production impact index, the steel material production impact index difference threshold, the steel material ranking information and the steel material basic classification information;

[0037] Among them, the calculation formula of steel production impact index is:

[0038]

[0039] Where E is the steel production impact index, Q(s) is the process node impact index of the s-th steel production process node, ω s is the sth steel production process node parameter, ω' s is the standard parameter of the sth steel production process node, and h is the total number of steel production process nodes.

[0040] Preferably, the obtaining of the second classification information of steel materials according to the first classification information of steel materials and the performance index of steel material performance characteristics specifically includes:

[0041] According to the steel surface feature data, basic steel surface feature information and steel surface defect feature information are obtained, wherein the basic steel surface feature information includes steel surface roughness information and steel surface oxidation information;

[0042] Obtain the steel material performance characteristic index according to the steel material surface basic characteristic information and the steel material surface defect characteristic information;

[0043] According to the steel performance characteristic index and the first classification information of the steel, the second classification information of the steel is obtained;

[0044] Among them, the calculation formula of steel performance characteristic index is:

[0045]

[0046] In the formula, R is the performance index of steel performance characteristics, is the surface roughness of steel, σ is the coverage of the oxide layer on the steel surface, τ g is the influence coefficient of the g-th defect on the steel surface on the steel performance, μ g is the proportion of the gth defect on the steel surface, and z is the total number of defects on the steel surface.

[0047] Preferably, the step of obtaining a training set of historical steel data based on data processing according to historical steel data specifically includes:

[0048] According to the historical data of steel materials, obtain the historical production information of steel materials, the data of steel material types and the historical surface characteristics data of steel materials;

[0049] Classify historical steel data according to historical steel production information, steel type data and historical steel surface feature data, and obtain historical data classification information;

[0050] According to the historical data classification information, the steel is matched with the corresponding historical steel performance test data to obtain the historical matching data of the steel;

[0051] According to the historical matching data of steel materials, a historical characteristic performance index of steel material performance is obtained, wherein the historical characteristic performance index of steel material performance is calculated in the same manner as the steel material performance characteristic performance index, and represents the steel material performance characteristic performance index of the steel material in the historical data of steel materials;

[0052] According to the historical characteristic performance index of steel performance, the historical matching data of steel are sorted in ascending order of the historical characteristic performance index of steel performance, so as to obtain the historical sorting data of steel;

[0053] According to the historical sorting data of steel materials, a training set of historical data of steel materials is obtained.

[0054] Preferably, the obtaining of steel material performance prediction data according to the second steel material classification information and the steel material performance prediction model specifically includes:

[0055] Obtain historical sorting data of steel materials based on the historical data training set of steel materials;

[0056] According to the historical sorting data of steel materials, calibration historical data of steel materials is obtained, wherein the calibration historical data of steel materials are the first historical data of steel materials, the last historical data of steel materials, and the middle historical data of steel materials in the same category;

[0057] The median steel historical data includes the order median steel historical data and the index median steel historical data. The order median steel historical data is the median steel historical data of each category in the steel historical ranking data. The index median steel historical data is the steel historical data that is the same as or closest to the average value of the maximum value of the steel performance historical characteristic performance index and the minimum value of the steel performance historical characteristic performance index in each category.

[0058] According to the calibration steel historical data and the steel historical data training set, the existing prediction model is trained and tested to obtain the steel performance prediction model;

[0059] According to the second classification information of steel, obtain the calibration steel information;

[0060] According to the calibration steel information, based on the mechanical properties test, the calibration steel performance data is obtained;

[0061] Obtain steel performance prediction data based on calibrated steel performance data and a steel performance prediction model;

[0062] Among them, the steel performance prediction model is specifically as follows:

[0063]

[0064] Wherein, W(x, y) represents the performance prediction data of the yth steel of the xth category, R1(x) represents the steel performance characteristic performance index of the first steel in the xth category of the calibrated steel, R2(x) represents the steel performance characteristic performance index of the last steel in the xth category of the calibrated steel, R3(x) represents the steel performance characteristic performance index of the median steel in the xth category of the calibrated steel, R4(x) represents the steel performance characteristic performance index of the median steel in the xth category of the calibrated steel, θ xk is the steel performance data of the kth calibration steel of the xth category, E xy represents the steel production impact index of the yth steel product in the xth category, E k is the steel production impact index of the kth calibrated steel of the xth category, R xy Represents the steel performance characteristic index of the yth steel of the xth category, Rk It is the steel performance characteristic index of the kth calibrated steel of the xth category.

[0065] Furthermore, a steel performance prediction system based on machine learning is proposed to implement the above-mentioned prediction method, including:

[0066] A main control module, the main control module is used to obtain a process node impact index based on process node vector information and historical steel performance data, obtain a steel production impact index based on the process node impact index and steel production information, obtain a steel performance characteristic performance index based on steel surface basic feature information and steel surface defect feature information, classify steels based on the steel production impact index, obtain first classification information of steels, perform surface feature recognition on steels, obtain steel surface feature data, and obtain second classification information of steels based on the first classification information of steels and the steel performance characteristic performance index;

[0067] An information acquisition module, wherein the information acquisition module is used to acquire basic information of steel, information on steel types, information on steel production, information on steel raw materials, information on steel production process, historical production data of steel, historical performance data of steel, parameter information on historical production process nodes of steel, historical data of steel, historical test data on steel performance, data on steel types, and historical surface feature data of steel, and transmit the data to the prediction module;

[0068] A prediction module, wherein the prediction module is used to obtain a training set of steel historical data based on data processing according to steel historical data, train and test an existing prediction model according to the calibrated steel historical data and the steel historical data training set, obtain a steel performance prediction model, obtain calibrated steel information according to the second classification information of steel, obtain calibrated steel performance data based on the calibrated steel information and mechanical property testing, and obtain steel performance prediction data according to the calibrated steel performance data and the steel performance prediction model;

[0069] A display module interacts with the main control module and is used to display the second classification information of steel, calibrated steel information, calibrated steel performance data and steel performance prediction data.

[0070] Optionally, the main control module specifically includes:

[0071] A control unit, the control unit is used to classify the steel according to the steel production impact index, obtain first classification information of the steel, identify the surface characteristics of the steel, obtain surface characteristic data of the steel, and obtain second classification information of the steel according to the first classification information of the steel and the performance characteristic index of the steel;

[0072] An information receiving unit, which interacts with the information acquisition module and the evaluation module to receive data and transmit it to the evaluation unit;

[0073] An evaluation unit is used to obtain a process node impact index based on process node vector information and historical steel performance data, obtain a steel production impact index based on the process node impact index and steel production information, and obtain a steel performance characteristic performance index based on steel surface basic characteristic information and steel surface defect characteristic information.

[0074] Optionally, the information acquisition module specifically includes:

[0075] A first acquisition unit, the first acquisition unit is used to acquire basic information of steel materials, information on steel types, information on steel production, information on steel raw materials, information on steel production processes, historical production data of steel materials, historical performance data of steel materials, and parameter information on historical production process nodes of steel materials;

[0076] The second acquisition unit is used to acquire historical data of steel materials, historical test data of steel material properties, data of steel material types and historical surface characteristic data of steel materials, and transmit the data to the prediction module.

[0077] Optionally, the prediction module specifically includes:

[0078] The model training unit is used to obtain a steel historical data training set based on data processing according to the steel historical data, train and test the existing prediction model according to the calibrated steel historical data and the steel historical data training set, and obtain a steel performance prediction model;

[0079] A steel material performance prediction unit is used to obtain calibrated steel material information according to the second classification information of steel material, obtain calibrated steel material performance data according to the calibrated steel material information and based on mechanical property testing, and obtain steel material performance prediction data according to the calibrated steel material performance data and a steel material performance prediction model.

[0080] Compared with the prior art, the present invention has the following beneficial effects:

[0081] The present invention proposes a steel performance prediction system and method based on machine learning, which evaluates the process nodes in the steel production process through the process node influence index, analyzes the influence of different process nodes on steel performance, and classifies steels under different production environments through the steel production influence index, thereby improving the efficiency of steel performance prediction. Steels are further divided through the steel performance characteristic performance index, thereby improving the accuracy of steel performance prediction. Steel performance is optimized through the steel performance prediction model, thereby effectively reducing trial and error costs and production time, and improving production efficiency and cost-effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1This is a flow chart of the steel performance prediction method based on machine learning proposed in the present invention;

[0083] Figure 2 This is a flow chart for obtaining the first classification information of steel in the present invention;

[0084] Figure 3 A flow chart for obtaining a historical data training set for steel in the present invention;

[0085] Figure 4 The flow chart for obtaining the steel performance prediction data in the present invention;

[0086] Figure 5 This is a structural block diagram of the steel performance prediction system based on machine learning proposed in the present invention. DETAILED DESCRIPTION

[0087] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0088] Reference Figure 1 - Figure 4 As shown, the steel performance prediction method based on machine learning in the embodiment of the present invention includes:

[0089] Obtaining basic information about steel materials, wherein the basic information about steel materials includes information about steel material types and information about steel material production;

[0090] According to the basic information of steel, based on the assessment of steel production environment, obtain the steel production impact index;

[0091] Classify steel materials according to the steel production impact index and obtain the first classification information of steel materials;

[0092] Specifically, steel is classified according to the steel production impact index to obtain the first classification information of steel, including:

[0093] According to the basic information of steel, steel production information is obtained, wherein the steel production information includes steel raw material information and steel production process information;

[0094] Acquire historical steel production data, wherein the historical steel production data includes historical steel performance data and historical steel production process node parameter information;

[0095] Based on the historical production data of steel, obtain the information on the impact of process nodes on steel performance;

[0096] According to the steel performance process node influence information, process node vector information is obtained, wherein the process node vector information includes a front-end control process node and a back-end influence process node of each process node;

[0097] Obtain the process node impact index based on process node vector information and historical steel performance data;

[0098] Obtain steel production impact index based on process node impact index and steel production information;

[0099] According to the steel production impact index, obtain the first classification information of steel;

[0100] Among them, the calculation formula of the process node impact index is:

[0101]

[0102] Where Q(x) is the process node impact index of the x-th process node, α i is the influence coefficient of the i-th front-end control process node on the process node, τ x is the steel performance influence coefficient of the xth process node, β j is the influence coefficient of the node on the jth back-end influencing process node, n is the total number of front-end controlled process nodes of the xth process node, and m is the total number of back-end influencing process nodes of the xth process node.

[0103] Specifically, according to the steel production impact index, the first classification information of steel is obtained, including:

[0104] Obtain steel type information based on basic steel information;

[0105] Based on the steel type information, steel is classified to obtain basic classification information of steel;

[0106] According to the steel production information, obtain the steel production process node information;

[0107] According to the steel type information, obtain the standard parameter information of the process node;

[0108] Obtain the steel production impact index according to the process node impact index, process node standard parameter information and steel production process node information;

[0109] Sort steel materials according to the steel production impact index and obtain steel material sorting information;

[0110] Based on the steel performance prediction requirements, the difference threshold of the steel production impact index is obtained;

[0111] Obtain the first classification information of steel materials according to the steel material production impact index, the steel material production impact index difference threshold, the steel material ranking information and the steel material basic classification information;

[0112] Among them, the calculation formula of steel production impact index is:

[0113]

[0114] Where E is the steel production impact index, Q(s) is the process node impact index of the s-th steel production process node, ω s is the sth steel production process node parameter, ω' s is the standard parameter of the sth steel production process node, and h is the total number of steel production process nodes.

[0115] In this scheme, the influence information of process nodes on steel performance is obtained through the historical production data of steel materials. The process node vector information is obtained based on the influence information of process nodes on steel performance. The process node influence index is obtained based on the process node vector information and the historical performance data of steel materials. The steel production influence index is obtained based on the process node influence index, the process node standard parameter information and the steel production process node information. The steel materials are sorted based on the steel production influence index to obtain the steel sorting information. The first classification information of steel materials is obtained based on the steel production influence index, the steel production influence index difference threshold, the steel sorting information and the basic classification information of steel materials.

[0116] It is understandable that in the steel production process, process node parameters are closely related to steel performance, and different process nodes have completely different degrees of influence on steel performance. For the same type of steel, the steel production process is basically the same, but there are certain differences in specific process node parameters. This is related to the specific production environment and production conditions and is unavoidable. Through the steel production impact index, steels with similar steel production conditions are classified, which improves the efficiency of steel performance prediction.

[0117] Identify the surface characteristics of steel and obtain the surface characteristic data of steel;

[0118] According to the surface characteristic data of steel, the performance characteristic index of steel is obtained;

[0119] According to the first classification information of steel materials and the performance index of steel material performance characteristics, the second classification information of steel materials is obtained;

[0120] Specifically, according to the first classification information of steel and the performance index of steel performance characteristics, the second classification information of steel is obtained, including:

[0121] According to the steel surface feature data, basic steel surface feature information and steel surface defect feature information are obtained, wherein the basic steel surface feature information includes steel surface roughness information and steel surface oxidation information;

[0122] Obtain the steel material performance characteristic index according to the steel material surface basic characteristic information and the steel material surface defect characteristic information;

[0123] According to the steel performance characteristic index and the first classification information of the steel, the second classification information of the steel is obtained;

[0124] Among them, the calculation formula of steel performance characteristic index is:

[0125]

[0126] In the formula, R is the performance index of steel performance characteristics, is the surface roughness of steel, σ is the coverage of the oxide layer on the steel surface, τ g is the influence coefficient of the g-th defect on the steel surface on the steel performance, μ g is the proportion of the gth defect on the steel surface, and z is the total number of defects on the steel surface.

[0127] In this scheme, the steel performance characteristic index is obtained through the basic characteristic information of the steel surface and the characteristic information of the steel surface defects. The second classification information of the steel is obtained according to the steel performance characteristic index and the first classification information of the steel. The steel is further divided according to the steel performance characteristic index to ensure that the steel performance of the same type of steel is similar, thereby improving the accuracy of steel performance prediction.

[0128] It should be noted that the specific steps to obtain the second classification information of steel materials are as follows:

[0129] Based on the steel performance prediction requirements, obtain the steel performance characteristic index difference threshold;

[0130] Sort each category of steel in the first classification information of steel according to the performance index of steel from small to large;

[0131] According to the steel performance characteristic performance index difference threshold, the adjacent steels in each category after sorting are evaluated and classified. If the difference in steel performance characteristic performance index of adjacent steels is less than the steel performance characteristic performance index difference threshold, the adjacent steels are of the same category. If the difference in steel performance characteristic performance index of adjacent steels is greater than the steel performance characteristic performance index difference threshold, the adjacent steels are of different categories, and the second classification information of steel is obtained.

[0132] It is understandable that the surface characteristics of steel have a certain impact on the performance of steel. For example, surface defects of steel such as pores, inclusions, and cracks may indicate quality problems in the manufacturing process or weaknesses that may occur during use. These defects may cause the local strength of the steel to weaken or be prone to cracking. At the same time, different steel surface roughness has different effects on steel surface defects. Even if the crack length and depth on the steel surface are the same, a higher surface roughness will lead to more severe stress concentration, resulting in faster crack propagation, thereby reducing the steel's resistance to crack propagation and fatigue life.

[0133] Acquire historical data of steel materials, wherein the historical data of steel materials include historical test data of steel material properties, data of steel material types and historical surface characteristic data of steel materials;

[0134] According to the historical data of steel, based on data processing, a training set of historical data of steel is obtained;

[0135] Specifically, based on the historical data of steel, a training set of historical data of steel is obtained based on data processing, which specifically includes:

[0136] According to the historical data of steel materials, obtain the historical production information of steel materials, the data of steel material types and the historical surface characteristics data of steel materials;

[0137] Classify historical steel data according to historical steel production information, steel type data and historical steel surface feature data, and obtain historical data classification information;

[0138] According to the historical data classification information, the steel is matched with the corresponding historical steel performance test data to obtain the historical matching data of the steel;

[0139] According to the historical matching data of steel materials, a historical characteristic performance index of steel material performance is obtained, wherein the historical characteristic performance index of steel material performance is calculated in the same manner as the steel material performance characteristic performance index, and represents the steel material performance characteristic performance index of the steel material in the historical data of steel materials;

[0140] According to the historical characteristic performance index of steel performance, the historical matching data of steel are sorted in ascending order of the historical characteristic performance index of steel performance, so as to obtain the historical sorting data of steel;

[0141] According to the historical sorting data of steel materials, a training set of historical data of steel materials is obtained.

[0142] In this scheme, the historical data of steel are classified through the historical production information of steel, the type data of steel and the historical surface characteristic data of steel to obtain the classification information of historical data, wherein the method for obtaining the classification information of historical data is the same as the method for obtaining the second classification information of steel, the steel production influence index of the historical data of steel is obtained through the process node influence index in the historical data of steel, the historical steel is classified through the steel production influence index of the historical steel, and the first classification information of the historical steel data is obtained, the steel performance characteristic performance index of the steel in the historical data of steel is obtained through the historical surface characteristic data of steel, that is, the historical characteristic performance index of steel performance, and the historical data classification information is obtained through the historical characteristic performance index of steel performance and the first classification information of historical steel data.

[0143] According to the historical data training set of steel, based on model training, a steel performance prediction model is obtained;

[0144] According to the second classification information of steel and the steel performance prediction model, the steel performance prediction data is obtained.

[0145] Specifically, according to the second classification information of steel and the steel performance prediction model, the steel performance prediction data is obtained, including:

[0146] Obtain historical sorting data of steel materials based on the historical data training set of steel materials;

[0147] According to the historical sorting data of steel materials, calibration historical data of steel materials is obtained, wherein the calibration historical data of steel materials are the first historical data of steel materials, the last historical data of steel materials, and the middle historical data of steel materials in the same category;

[0148] The median steel historical data includes the median steel historical data in the order and the median steel historical data in the index. The median steel historical data in the order is the median steel historical data in the order of each category in the steel historical ranking data. The median steel historical data in the index is the steel historical data in each category that is the same as or closest to the average value of the maximum value of the steel performance historical characteristic performance index and the minimum value of the steel performance historical characteristic performance index. The first steel historical data is the steel historical data with the minimum value of the steel performance historical characteristic performance index, and the last steel historical data is the steel historical data with the maximum value of the steel performance historical characteristic performance index.

[0149] According to the calibration steel historical data and the steel historical data training set, the existing prediction model is trained and tested to obtain the steel performance prediction model;

[0150] According to the second classification information of steel, obtain the calibration steel information;

[0151] According to the calibration steel information, based on the mechanical properties test, the calibration steel performance data is obtained;

[0152] Obtain steel performance prediction data based on calibrated steel performance data and a steel performance prediction model;

[0153] Among them, the steel performance prediction model is specifically as follows:

[0154]

[0155] Wherein, W(x, y) represents the performance prediction data of the yth steel of the xth category, R1(x) represents the steel performance characteristic performance index of the first steel in the xth category of the calibrated steel, R2(x) represents the steel performance characteristic performance index of the last steel in the xth category of the calibrated steel, R3(x) represents the steel performance characteristic performance index of the median steel in the xth category of the calibrated steel, R4(x) represents the steel performance characteristic performance index of the median steel in the xth category of the calibrated steel, θ xk is the steel performance data of the kth calibration steel of the xth category, E xy represents the steel production impact index of the yth steel product in the xth category, E k is the steel production impact index of the kth calibrated steel of the xth category, R xy Represents the steel performance characteristic index of the yth steel of the xth category, R k It is the steel performance characteristic index of the kth calibrated steel of the xth category.

[0156] In this scheme, through the steel historical data training set, the steel historical sorting data is obtained, and the calibrated steel historical data is obtained according to the steel historical sorting data. According to the calibrated steel historical data and the steel historical data training set, the existing prediction model is trained and tested to obtain the steel performance prediction model, and according to the second classification information of the steel, the calibrated steel information is obtained. According to the calibrated steel information, based on the mechanical property test, the calibrated steel performance data is obtained, and according to the calibrated steel performance data and the steel performance prediction model, the steel performance prediction data is obtained. The steel performance is optimized through the steel performance prediction model, which effectively reduces the trial and error cost and production time, and improves the production efficiency and cost-effectiveness.

[0157] It should be noted that, in the present embodiment, the first steel historical data, the last steel historical data and the median steel historical data in the same category of steel historical data are used as the benchmark data of the steel performance prediction model, that is, the steel historical data with the largest steel performance historical characteristic performance index, the steel historical data with the smallest steel performance historical characteristic performance index, the steel historical data in the middle of the ranking of the steel performance historical characteristic performance index and the steel historical data with the same or closest average value of the maximum value of the steel performance historical characteristic performance index and the minimum value of the steel performance historical characteristic performance index in the category are selected from the same category of steel historical data. These four types of data are used as the benchmark data, and the calibrated steel performance data are obtained by performing steel performance tests on the calibrated steels of each category in the second classification information of steels, and then the performance of the steels of this category is predicted by the calibrated steel performance data, thereby improving the accuracy of the steel performance prediction, while avoiding the steel performance test on each steel and reducing the detection cost.

[0158] Reference Figure 5 As shown, further, combined with the above-mentioned steel performance prediction method based on machine learning, a steel performance prediction system based on machine learning is proposed, including:

[0159] A main control module, the main control module is used to obtain a process node impact index based on process node vector information and historical steel performance data, obtain a steel production impact index based on the process node impact index and steel production information, obtain a steel performance characteristic performance index based on steel surface basic feature information and steel surface defect feature information, classify steels based on the steel production impact index, obtain first classification information of steels, perform surface feature recognition on steels, obtain steel surface feature data, and obtain second classification information of steels based on the first classification information of steels and the steel performance characteristic performance index;

[0160] An information acquisition module, wherein the information acquisition module is used to acquire basic information of steel, information on steel types, information on steel production, information on steel raw materials, information on steel production process, historical production data of steel, historical performance data of steel, parameter information on historical production process nodes of steel, historical data of steel, historical test data on steel performance, data on steel types, and historical surface feature data of steel, and transmit the data to the prediction module;

[0161] A prediction module, wherein the prediction module is used to obtain a training set of steel historical data based on data processing according to steel historical data, train and test an existing prediction model according to the calibrated steel historical data and the steel historical data training set, obtain a steel performance prediction model, obtain calibrated steel information according to the second classification information of steel, obtain calibrated steel performance data based on the calibrated steel information and mechanical property testing, and obtain steel performance prediction data according to the calibrated steel performance data and the steel performance prediction model;

[0162] A display module interacts with the main control module and is used to display the second classification information of steel, calibrated steel information, calibrated steel performance data and steel performance prediction data.

[0163] Main control module, including:

[0164] A control unit, the control unit is used to classify the steel according to the steel production impact index, obtain first classification information of the steel, identify the surface characteristics of the steel, obtain surface characteristic data of the steel, and obtain second classification information of the steel according to the first classification information of the steel and the performance characteristic index of the steel;

[0165] An information receiving unit, which interacts with the information acquisition module and the evaluation module to receive data and transmit it to the evaluation unit;

[0166] An evaluation unit is used to obtain a process node impact index based on process node vector information and historical steel performance data, obtain a steel production impact index based on the process node impact index and steel production information, and obtain a steel performance characteristic performance index based on steel surface basic characteristic information and steel surface defect characteristic information.

[0167] Information acquisition module, specifically including:

[0168] A first acquisition unit, the first acquisition unit is used to acquire basic information of steel materials, information on steel types, information on steel production, information on steel raw materials, information on steel production processes, historical production data of steel materials, historical performance data of steel materials, and parameter information on historical production process nodes of steel materials;

[0169] The second acquisition unit is used to acquire historical data of steel materials, historical test data of steel material properties, data of steel material types and historical surface characteristic data of steel materials, and transmit the data to the prediction module.

[0170] Prediction module, including:

[0171] The model training unit is used to obtain a steel historical data training set based on data processing according to the steel historical data, train and test the existing prediction model according to the calibrated steel historical data and the steel historical data training set, and obtain a steel performance prediction model;

[0172] A steel material performance prediction unit is used to obtain calibrated steel material information according to the second classification information of steel material, obtain calibrated steel material performance data according to the calibrated steel material information and based on mechanical property testing, and obtain steel material performance prediction data according to the calibrated steel material performance data and a steel material performance prediction model.

[0173] In summary, the advantages of the present invention are: obtaining a process node influence index through process node vector information and steel historical performance data, evaluating the process nodes in the steel production process through the process node influence index, analyzing the degree of influence of different process nodes on steel performance, obtaining a steel production influence index through the process node influence index and steel production information, classifying steels under different production environments through the steel production influence index, thereby improving the efficiency of steel performance prediction, obtaining a steel performance characteristic performance index through steel surface basic feature information and steel surface defect feature information, further dividing steels through the steel performance characteristic performance index, ensuring that the steel performance of the same type of steel is similar, and improving the accuracy of steel performance prediction, training and testing the existing prediction model through calibrating steel historical data and steel historical data training sets, obtaining a steel performance prediction model, optimizing steel performance through the steel performance prediction model, effectively reducing trial and error costs and production time, and improving production efficiency and cost-effectiveness.

[0174] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A steel performance prediction method based on machine learning, characterized in that: include: Obtaining basic information about steel materials, wherein the basic information about steel materials includes information about steel material types and information about steel material production; According to the basic information of steel, based on the assessment of steel production environment, obtain the steel production impact index; Classify steel materials according to the steel production impact index and obtain the first classification information of steel materials; Identify the surface characteristics of steel and obtain the surface characteristic data of steel; According to the surface characteristic data of steel, the performance characteristic index of steel is obtained; According to the first classification information of steel materials and the performance index of steel material performance characteristics, the second classification information of steel materials is obtained; Acquire historical data of steel materials, wherein the historical data of steel materials include historical test data of steel material properties, data of steel material types and historical surface characteristic data of steel materials; According to the historical data of steel, based on data processing, a training set of historical data of steel is obtained; According to the historical data training set of steel, based on model training, a steel performance prediction model is obtained; Obtain steel performance prediction data according to the second classification information of steel and the steel performance prediction model; The step of classifying steel materials according to the steel production impact index and obtaining first classification information of steel materials specifically includes: According to the basic information of steel, steel production information is obtained, wherein the steel production information includes steel raw material information and steel production process information; Acquire historical steel production data, wherein the historical steel production data includes historical steel performance data and historical steel production process node parameter information; Based on the historical production data of steel, obtain the information on the impact of process nodes on steel performance; According to the steel performance process node influence information, process node vector information is obtained, wherein the process node vector information includes a front-end control process node and a back-end influence process node of each process node; Obtain the process node impact index based on process node vector information and historical steel performance data; Obtain steel production impact index based on process node impact index and steel production information; According to the steel production impact index, obtain the first classification information of steel; Among them, the calculation formula of the process node impact index is: In the formula, is the process node impact index of the x-th process node, is the influence coefficient of the i-th front-end control process node on the process node, is the steel performance influence coefficient of the xth process node, is the influence coefficient of the node on the jth back-end influencing process node, n is the total number of front-end controlled process nodes of the xth process node, and m is the total number of back-end influencing process nodes of the xth process node.

2. The steel performance prediction method based on machine learning according to claim 1, characterized in that: The obtaining of the first classification information of steel materials according to the steel production impact index specifically includes: Obtain steel type information based on basic steel information; Based on the steel type information, steel is classified to obtain basic classification information of steel; According to the steel production information, obtain the steel production process node information; According to the steel type information, obtain the standard parameter information of the process node; Obtain the steel production impact index according to the process node impact index, process node standard parameter information and steel production process node information; Sort steel materials according to the steel production impact index and obtain steel material sorting information; Based on the steel performance prediction requirements, the difference threshold of the steel production impact index is obtained; Obtain the first classification information of steel materials according to the steel material production impact index, the steel material production impact index difference threshold, the steel material ranking information and the steel material basic classification information; Among them, the calculation formula of steel production impact index is: Where E is the steel production impact index, is the process node impact index of the s-th steel production process node, is the parameter of the s-th steel production process node, is the standard parameter of the sth steel production process node, and h is the total number of steel production process nodes.

3. The steel material properties prediction method based on machine learning according to claim 1, characterized in that: The obtaining of the second classification information of the steel material according to the first classification information of the steel material and the performance characteristic index of the steel material specifically includes: According to the steel surface feature data, basic steel surface feature information and steel surface defect feature information are obtained, wherein the basic steel surface feature information includes steel surface roughness information and steel surface oxidation information; Obtain the steel material performance characteristic index according to the steel material surface basic characteristic information and the steel material surface defect characteristic information; According to the steel performance characteristic index and the first classification information of the steel, the second classification information of the steel is obtained; Among them, the calculation formula of steel performance characteristic index is: In the formula, R is the performance index of steel performance characteristics, is the surface roughness of steel, is the coverage of the oxide layer on the steel surface, is the influence coefficient of the g-th defect on the steel surface on the steel performance, is the proportion of the g-th defect on the steel surface, is the total number of surface defects on steel.

4. The steel material properties prediction method based on machine learning according to claim 1, characterized in that: The obtaining of a training set of historical steel data based on data processing according to the historical steel data specifically includes: According to the historical data of steel materials, obtain the historical production information of steel materials, the data of steel material types and the historical surface characteristics data of steel materials; Classify historical steel data according to historical steel production information, steel type data and historical steel surface feature data, and obtain historical data classification information; According to the historical data classification information, the steel is matched with the corresponding historical steel performance test data to obtain the historical matching data of the steel; According to the historical matching data of steel materials, a historical characteristic performance index of steel materials is obtained, wherein the historical characteristic performance index of steel materials is calculated in the same manner as the steel material performance index of steel materials, and represents the steel material performance index of steel materials in the historical data of steel materials; According to the historical characteristic performance index of steel performance, the historical matching data of steel are sorted in ascending order of the historical characteristic performance index of steel performance, so as to obtain the historical sorting data of steel; According to the historical sorting data of steel materials, a training set of historical data of steel materials is obtained.

5. The steel performance prediction method based on machine learning according to claim 1, characterized in that: The obtaining of steel material performance prediction data according to the second steel material classification information and the steel material performance prediction model specifically includes: Obtain historical sorting data of steel materials based on the historical data training set of steel materials; According to the historical sorting data of steel materials, calibration historical data of steel materials is obtained, wherein the calibration historical data of steel materials are the first historical data of steel materials, the last historical data of steel materials, and the middle historical data of steel materials in the same category; The median steel historical data includes the order median steel historical data and the index median steel historical data. The order median steel historical data is the median steel historical data of each category in the steel historical ranking data. The index median steel historical data is the steel historical data that is the same as or closest to the average value of the maximum value of the steel performance historical characteristic performance index and the minimum value of the steel performance historical characteristic performance index in each category. According to the calibration steel historical data and the steel historical data training set, the existing prediction model is trained and tested to obtain the steel performance prediction model; According to the second classification information of steel, obtain the calibration steel information; According to the calibration steel information, based on the mechanical properties test, the calibration steel performance data is obtained; Obtain steel performance prediction data based on calibrated steel performance data and a steel performance prediction model; Among them, the steel performance prediction model is specifically as follows: In the formula, represents the performance prediction data of the yth steel of the xth category, It represents the steel performance characteristic index of the first steel in the x-th type of calibrated steel. It represents the steel performance characteristic index of the last steel in the x-th category of calibrated steel. It represents the steel performance characteristic index of the median steel in the x-th category of the calibrated steel. It represents the steel performance characteristic index of the median steel in the x-type calibrated steel. is the steel performance data of the kth calibrated steel of the xth category, represents the steel production impact index of the yth steel product in the xth category, is the steel production impact index of the kth calibrated steel of the xth category, It represents the steel performance characteristic index of the yth steel of the xth category, It is the steel performance characteristic index of the kth calibrated steel of the xth category.

6. A steel performance prediction system based on machine learning, used to implement the prediction method according to any one of claims 1 to 5, characterized in that: include: A main control module, the main control module is used to obtain a process node impact index based on process node vector information and historical steel performance data, obtain a steel production impact index based on the process node impact index and steel production information, obtain a steel performance characteristic performance index based on steel surface basic feature information and steel surface defect feature information, classify steels based on the steel production impact index, obtain first classification information of steels, perform surface feature recognition on steels, obtain steel surface feature data, and obtain second classification information of steels based on the first classification information of steels and the steel performance characteristic performance index; An information acquisition module, wherein the information acquisition module is used to acquire basic information of steel, information on steel types, information on steel production, information on steel raw materials, information on steel production process, historical production data of steel, historical performance data of steel, parameter information on historical production process nodes of steel, historical data of steel, historical test data on steel performance, data on steel types, and historical surface feature data of steel, and transmit the data to the prediction module; A prediction module, wherein the prediction module is used to obtain a training set of steel historical data based on data processing according to steel historical data, train and test an existing prediction model according to the calibrated steel historical data and the steel historical data training set, obtain a steel performance prediction model, obtain calibrated steel information according to the second classification information of steel, obtain calibrated steel performance data based on the calibrated steel information and mechanical property testing, and obtain steel performance prediction data according to the calibrated steel performance data and the steel performance prediction model; A display module interacts with the main control module and is used to display the second classification information of steel, calibrated steel information, calibrated steel performance data and steel performance prediction data.

7. The steel material properties prediction system based on machine learning according to claim 6, characterized in that: The main control module specifically includes: A control unit, the control unit is used to classify the steel according to the steel production impact index, obtain first classification information of the steel, identify the surface characteristics of the steel, obtain surface characteristic data of the steel, and obtain second classification information of the steel according to the first classification information of the steel and the performance characteristic index of the steel; An information receiving unit, which interacts with the information acquisition module and the evaluation module to receive data and transmit it to the evaluation unit; An evaluation unit is used to obtain a process node impact index based on process node vector information and historical steel performance data, obtain a steel production impact index based on the process node impact index and steel production information, and obtain a steel performance characteristic performance index based on steel surface basic characteristic information and steel surface defect characteristic information.

8. The steel material properties prediction system based on machine learning according to claim 6, characterized in that: The information acquisition module specifically includes: A first acquisition unit, the first acquisition unit is used to acquire basic information of steel materials, information on steel types, information on steel production, information on steel raw materials, information on steel production processes, historical production data of steel materials, historical performance data of steel materials, and parameter information on historical production process nodes of steel materials; The second acquisition unit is used to acquire historical data of steel materials, historical test data of steel material properties, data of steel material types and historical surface characteristic data of steel materials, and transmit the data to the prediction module.

9. The steel material properties prediction system based on machine learning according to claim 6, characterized in that: The prediction module specifically includes: The model training unit is used to obtain a steel historical data training set based on data processing according to the steel historical data, train and test the existing prediction model according to the calibrated steel historical data and the steel historical data training set, and obtain a steel performance prediction model; A steel material performance prediction unit is used to obtain calibrated steel material information according to the second classification information of steel material, obtain calibrated steel material performance data according to the calibrated steel material information and based on mechanical property testing, and obtain steel material performance prediction data according to the calibrated steel material performance data and a steel material performance prediction model.

Images