CN114386325B - Strip steel mechanical property forecasting method based on rule optimizing - Google Patents

Abstract

The invention discloses a band steel mechanical property forecasting method based on rule optimization, which comprises the following steps: collecting characteristic variables and mechanical properties of the strip steel, forming a sample library, and calculating the contribution degree of each characteristic variable to the mechanical properties; selecting a preset number of feature variables with total contribution degree of a preset value from high to low as rule composition features, and calculating standard deviation of the rule composition features; combining each rule composition characteristic with a standard deviation threshold value of each rule composition characteristic to obtain a plurality of different rules; calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording; and predicting the mechanical properties of the strip steel by utilizing an optimal rule to obtain a prediction result. The method provided by the invention has very strong generalization capability, can accurately forecast the mechanical properties of the strip steel, can also guide the on-site production, and reflects the abnormal problems of an on-site data acquisition system.

Description

Strip steel mechanical property forecasting method based on rule optimizing

Technical Field

The invention relates to the technical field of strip steel mechanical property prediction, in particular to a strip steel mechanical property prediction method based on rule optimizing.

Background

The detection of the mechanical properties of the strip steel is an important link of the steel production flow, and the currently adopted detection of the mechanical properties of the strip steel is mainly detected by manual sampling, and has the defects of low efficiency, high cost, hysteresis and the like.

The mechanical property prediction of strip steel is a technology for predicting the mechanical property of a finished product by utilizing the chemical components of the strip steel and the production process parameters, and the quality judgment is carried out by adopting a product property prediction value, so that the manual sampling on site can be reduced, and the steel judging efficiency is improved. The existing mechanical property prediction models, such as a metallurgical mechanism model, a neural network model and the like, have higher prediction accuracy when the model is built, but the model accuracy is reduced along with the change of the production process due to the adaptability problem of the model; although the mechanism model has better accuracy in prediction, the model has higher cost for establishing and limited application range, and cannot be popularized in large-scale production with complex variety structures.

Disclosure of Invention

The invention provides a band steel mechanical property forecasting method based on rule optimization, which aims to solve the technical problems of poor generalization capability and low accuracy of the existing band steel mechanical property forecasting method when the number of samples is small.

In order to solve the technical problems, the invention provides the following technical scheme:

on one hand, the invention provides a band steel mechanical property forecasting method based on rule optimization, which comprises the following steps:

collecting characteristic variables and mechanical properties of the strip steel, forming a sample library, and calculating the contribution degree of each characteristic variable to the mechanical properties; wherein the characteristic variables comprise the chemical components of the strip steel and the technological parameters of the strip steel;

sequencing the feature variables from high to low according to the respective contribution degrees, selecting the feature variables with the total contribution degrees being the preset number of preset values as rule composition features, and calculating the standard deviation of each rule composition feature;

determining a standard deviation threshold value of each rule composition characteristic based on the calculated standard deviation, and combining each rule composition characteristic with the standard deviation threshold value of each rule composition characteristic to obtain a plurality of different rules;

calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording;

and predicting the mechanical properties of the strip steel to be predicted by using the optimal rule to obtain a prediction result.

Optionally, the chemical composition comprises carbon, silicon, manganese, phosphorus, sulfur, nitrogen, chromium, copper, molybdenum, niobium, nickel, titanium, and vanadium;

the process parameters comprise slab thickness, heating furnace inlet temperature, heating furnace outlet temperature, heating time, rough rolling outlet temperature, intermediate slab thickness, finish rolling outlet temperature and strip steel thickness.

The mechanical properties include yield strength, tensile strength, and elongation.

Further, the calculating the contribution degree of each characteristic variable to the mechanical property includes:

dividing the data in the sample library into a training set and a testing set; model training is carried out by utilizing characteristic variables and mechanical properties in the training set, and a random forest mechanical property regression model is obtained;

predicting the mechanical properties in the test set by using the random forest mechanical property regression model, and comparing the prediction result with the real mechanical properties in the test set to obtain an initial regression scoring coefficient;

selecting one of the characteristic variables in the test set for random replacement and disturbing the sequence each time to obtain a new test set, respectively predicting the mechanical properties in the new test set by using the random forest mechanical property regression model after obtaining the new test set each time through the characteristic variables in the new test set, and comparing the prediction result with the real mechanical properties to obtain regression scoring coefficients after replacement of each characteristic variable;

and taking the absolute value of the difference between the regression scoring coefficient after each characteristic variable is replaced and the initial regression scoring coefficient and dividing the absolute value by the initial regression scoring coefficient to obtain a numerical value which represents the contribution degree of the replaced characteristic variable.

Further, the determining a standard deviation threshold for each rule constituent feature based on the calculated standard deviation includes:

based on the calculated standard deviation, k times the standard deviation of each rule constituent feature is taken as its corresponding standard deviation threshold.

Further, the combining the rule composition features with the standard deviation threshold of the rule composition features to obtain a plurality of different rules includes:

randomly arranging and combining the rule composition characteristics and standard deviation threshold values of the rule composition characteristics to obtain a plurality of different rules; each rule respectively comprises all rule composition characteristics and any standard deviation threshold value of the rule composition characteristics corresponding to the rule composition characteristics.

Further, the calculating the predicted hit rate of the mechanical property of the strip steel under each rule comprises the following steps:

dividing the data in the sample library into a training set and a testing set; and respectively matching all samples corresponding to each sample in the training set by using each rule, and calculating the predicted hit rate under the rule.

Further, each rule is utilized to respectively match all samples corresponding to each sample in the test set in the training set, including:

under the current rule, taking one sample in the test set, matching the sample with characteristic variable in the positive and negative standard deviation threshold range in the training set according to the rule composition characteristic and the corresponding standard deviation threshold in the current rule, and taking the average value of the mechanical properties of all the matched samples as the prediction result of the corresponding sample in the test set.

Further, the calculating the predicted hit rate under the rule includes:

and (3) making differences between mechanical property prediction results and true values of all test set samples obtained under the current rule, and if the differences are within a preset error range, determining the differences as prediction hits, so as to calculate the prediction hit rate of the current rule.

Further, predicting the mechanical properties of the strip steel to be predicted by utilizing the optimal rule, including:

acquiring characteristic variables of strip steel to be predicted; according to rule composition characteristics and corresponding standard deviation threshold values in the current rule, matching samples of characteristic variables in a positive standard deviation threshold value and a negative standard deviation threshold value range in a sample library, and taking the average value of the mechanical properties of all the matched samples as the predicted mechanical property of the strip steel to be predicted.

In yet another aspect, the present invention also provides an electronic device including a processor and a memory; wherein the memory stores at least one instruction that is loaded and executed by the processor to implement the above-described method.

In yet another aspect, the present invention also provides a computer readable storage medium having at least one instruction stored therein, the instruction being loaded and executed by a processor to implement the above method.

The technical scheme provided by the invention has the beneficial effects that at least:

the invention calculates the contribution degree of the characteristic variable to the mechanical property by utilizing the chemical components, the technological parameters and the mechanical property of the existing strip steel in the sample library; taking the characteristic of a certain contribution degree as a rule characteristic and calculating the standard deviation of the rule characteristic; combining the characteristic variable and the standard deviation threshold to obtain a plurality of different rules; calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording; and predicting the mechanical properties of the strip steel by utilizing an optimal rule. On one hand, the optimal rule is updated continuously along with the production progress and the process change, so that the method can be well adapted to the field production change and has strong generalization capability; on the other hand, by checking the feature matching condition of each sample, the correction of the data acquisition system can be guided by checking whether the unmatched features reflect the problem of the on-site data acquisition system when acquiring the feature parameter. Therefore, the mechanical property of the strip steel can be accurately predicted, the prediction generalization capability is improved, the on-site production can be guided, and the abnormal problem of the on-site data acquisition system is reflected.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a method for forecasting mechanical properties of strip steel based on rule optimization provided by an embodiment of the invention;

fig. 2 is a diagram of a prediction result of a method for predicting mechanical properties of strip steel according to an embodiment of the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings.

First embodiment

The embodiment provides a band steel mechanical property forecasting method based on rule optimization, which can be realized by electronic equipment, and the execution flow of the method is shown in figure 1, and comprises the following steps:

s1, collecting characteristic variables and mechanical properties of strip steel, forming a sample library, and calculating the contribution degree of each characteristic variable to the mechanical properties; wherein, the characteristic variables comprise the chemical components of the strip steel and the technological parameters of the strip steel;

specifically, in this embodiment, the chemical components of the strip steel include: carbon, silicon, manganese, phosphorus, sulfur, nitrogen, chromium, copper, molybdenum, niobium, nickel, titanium, and vanadium; the technological parameters of the strip steel include: slab thickness, furnace inlet temperature, furnace outlet temperature, heating time, rough rolling outlet temperature, intermediate slab thickness, finish rolling outlet temperature and strip thickness. The mechanical properties of the strip steel comprise: yield strength, tensile strength and elongation.

The calculation of the contribution degree of each characteristic variable to the mechanical property comprises the following steps:

a1, dividing all samples in a sample library into a training set (TrainData) and a test set (TestData) according to a certain proportion, and performing model training by using chemical components, process parameters and mechanical property data in the training set to obtain a random forest mechanical property regression model;

a2, predicting the mechanical properties of the strip steel according to the chemical components and the technological parameters of the strip steel in the test set sample by using the random forest mechanical property regression model obtained in the A1. Comparing and calculating a prediction result with the real mechanical property data in the test set, and taking an R2_score index as an initial regression scoring coefficient beta;

a3, marking chemical components and process parameters in training set as [ alpha ] ₁ ,α ₂ ,α ₃ ,…α _n ]The features alpha are each separated using a "shuffle" function _i Each sample in i epsilon {1,2,..n } is randomly replaced to obtain n new test sets, a random forest model obtained by A1 is used to obtain a mechanical property predicted value of the new test set sample, and the mechanical property predicted value is compared with a true value to calculate, so as to obtain a regression scoring coefficient beta after replacement _i ，i∈{1，2，...n}。

A4 beta obtained by A3 _i Respectively differencing with beta obtained by A2 and dividing by beta, using the obtained numerical value to represent the characteristic contribution degree of characteristic variable to mechanical property,the larger the result is, the maximum contribution degree of the feature which replaces the feature value and corresponds to the regression scoring coefficient to the mechanical property is indicated.

S2, sorting the characteristic variables according to the respective contribution degree from high to low, selecting the characteristic variables with the total contribution degree being the preset number of preset values as rule composition characteristics, and calculating the standard deviation of each rule composition characteristic;

specifically, in this embodiment, the steps described above are:

after the contribution degree of each feature is calculated in the previous step, the first x features with the total contribution degree of n are selected from high to low to serve as rule composition features to be optimized, and standard deviations of the features are calculated by using an std function.

S3, determining a standard deviation threshold value of each rule composition characteristic based on the calculated standard deviation, and combining each rule composition characteristic with the standard deviation threshold value of each rule composition characteristic to obtain a plurality of different rules;

specifically, in this embodiment, the steps described above are:

selecting k times (such as 0.5, 0.7, 1, 1.2 and 1.5 times) of the standard deviation obtained by the previous calculation as a rule optimizing threshold value, forming all rules with x features, wherein each rule comprises the x features and a threshold value corresponding to the features, carrying out random permutation and combination on the threshold values to obtain all rules, and if L optimizing threshold values exist in each rule, obtaining the quantity of all rules as L ^x 。

S4, calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording;

specifically, in this embodiment, the steps described above are: dividing data in a sample library into a training set and a testing set; and respectively matching all samples corresponding to each sample in the training set by utilizing each rule, calculating the predicted hit rate under the rule, and taking the rule with the highest hit rate as the optimal rule and recording.

Further, the matching of each sample in the test set with each rule is specifically: under the current rule, taking one sample in the test set, matching the sample with characteristic variables in the positive and negative threshold ranges according to the rule composition characteristics and the corresponding standard deviation threshold in the training set in the current rule, and taking the average value of the mechanical properties of all the matched samples as the prediction result of the sample in the test set.

Calculating the predicted hit rate under the rule, taking the rule with the highest hit rate as the optimal rule, and recording the optimal rule as follows: calculating the mechanical property prediction results of all the test set samples according to the method of the last step and making differences with the true values, if the differences are within the set error range, namely, predicting hit rates of all the test set samples, calculating the prediction hit rate of all the test set samples, wherein the rule with the highest hit rate is the optimal rule, and recording the optimal rule into a database for direct taking of subsequent predictions.

S5, predicting the mechanical properties of the strip steel to be predicted by utilizing the optimal rule to obtain a prediction result.

Specifically, in this embodiment, the steps specifically include: the method comprises the steps of obtaining chemical components and technological parameters of strip steel to be predicted, selecting an optimal rule, matching the characteristics of the optimal rule and a corresponding threshold value in a sample library, and taking an average value of mechanical properties of all samples obtained by matching as the predicted mechanical property of the strip steel to be predicted.

Next, the effects of the method of the present embodiment will be described with reference to specific application examples. The method comprises the following specific steps:

step 1, collecting characteristic variables and mechanical properties of strip steel to form a sample library, and calculating the contribution degree of each characteristic variable to the mechanical properties; wherein, the characteristic variables comprise the chemical components of the strip steel and the technological parameters of the strip steel;

specifically, in the present embodiment, carbon, chromium, copper, manganese, molybdenum, nitrogen, niobium, nickel, phosphorus, sulfur, silicon, titanium, vanadium are each represented by C, CR, CU, MN, MO, N, NB, NI, P, S, SI, TI, V; the SLAB thickness, STRIP thickness, rough rolling outlet temperature, intermediate SLAB thickness, finish rolling outlet temperature, heating TIME, heating furnace inlet temperature and heating furnace outlet temperature are respectively expressed by SLAB_THK, STRIP_ THK, RMXT, RMXT _ THK, FMXT, FCE _TIME, FCE_ENT_TEMP, FCE_EXT_TEMP and CT; yield strength, tensile strength and elongation are denoted by YS, TS and EL, respectively.

The process of calculating the contribution degree of each characteristic variable to the mechanical property is as follows:

existing samples were taken at 8:2 into a training set and a test set, wherein the training set uses an ExtraTreesRegror algorithm in a sklearn library to set model parameters (such as max_depth (maximum tree depth) =14, max_leaf_nodes (maximum leaf node number) =148, min_interior_depth (minimum uncertainty of leaf node splitting) =1.31) to obtain a training model.

Carrying out mechanical property prediction on samples in a test set by using the training model, comparing the result with the actual mechanical property of the test set, and calculating an initial regression scoring coefficient by using an r2 score function in a sklearn library;

and then, using a 'shuffle' function in a sklearn library to respectively and immediately replace the chemical components and the values of the technological parameters in the test set, replacing one column of features each time to obtain a plurality of new test sets, and calculating regression scoring coefficients of each new sample set replacing one column of features by using the method.

And (3) subtracting the new regression scoring coefficient from the initial regression scoring coefficient and dividing the new regression scoring coefficient by the initial regression scoring coefficient to obtain the importance, namely the contribution degree, of the result characterization feature to the mechanical property.

Step 2, taking the characteristic of a certain contribution degree as a rule characteristic and calculating the standard deviation of the rule characteristic;

specifically, in this embodiment, the existing steel strip data in the sample library only includes samples with steel grade Q345B and slab steel grade mbry 34536, and after the contribution degree is obtained in the first step, the selected characteristics are respectively: yield strength [ STRIP_ THK, RMXT, CT, FMXT, FCE _EXT_TEMP ], tensile strength [ C, STRIP_THK, CT, FMXT, P ], elongation [ STRIP_THK, RMXT, S, CT, FMXT ]. Calculating their standard deviation and multiplying by the corresponding coefficient yields the threshold range to be selected as shown in table 1 below:

table 1 threshold ranges for eigenvalues to be selected

Features (e.g. a character)	0.5σ	0.7σ	σ	1.2σ	1.5σ
						STRIP_THK	0.6524	0.9133	1.3047	1.5656	1.9571
RMXT	10.4631	14.6483	20.9261	25.1113	31.3892
						CT	10.1648	14.2307	20.3296	24.3955	30.4944
FCE_EXT_TEMP	14.7319	20.6247	29.4638	35.3566	44.1957
						C	0.031	0.0434	0.062	0.0744	0.093
P	0.0055	0.0076	0.0109	0.0131	0.0164
						S	0.0014	0.0019	0.0027	0.0032	0.0041
FMXT	11.8320	16.5647	23.6639	28.3967	35.4959
						SLAB_THK	0.9708	1.3591	1.9416	2.3299	2.9124

Step 3, combining the characteristic variable and the standard deviation threshold to obtain a plurality of different rules;

specifically, in the present embodiment, one of the rules of yield strength is: there are 3125 rules for [ STRIP_THK,0.6524], [ RMXT,10.4631], [ CT,10.1648], [ FMXT,11.8320], [ FCE_EXT_TEMP,14.7319], yield strength, tensile strength and elongation, respectively.

Step 4, calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording;

specifically, in this embodiment, the implementation procedure of the above steps is as follows:

dividing the data set into a training set and a testing set, taking one rule obtained by combining in the step 3, and matching the corresponding sample of each sample in the training set by using the rule. For example, the first sample characteristic parameters in the test set are: [ STRIP_THK,15.56], [ RMXT,1128.8], [ CT,680.612], [ FMXT,911.69], [ FCE_EXT_TEMP,1153.48], the sample characteristic parameters matched in the training set using the example rule in step 3 are in the range of: the average of the yield strengths of all training set samples with characteristic variables in the range of [ STRIP_THK, 15.56-0.6524-15.56+0.6524 ], [ RMXT, 1128.8-10.4631-1128.8+10.4631 ], [ CT, 680.612-10.1648-680.612-10.1648 ], [ FMXT, 911.69-11.8320-911.69+11.8320 ], [ FCE_EXT_TEMP, 1153.48-14.7319-1153.48+14.7319 ] is taken as the predicted yield strength of the first sample in the test set and is worse than the actual yield strength thereof, and if the error is less than the set error limit of 30MPa, the predicted hit is considered.

The hit rates of all samples in the test set were calculated by the above method as hit rates of the example rule in step 3. After all rules are matched, the rule with the highest hit rate is taken as the yield strength matching rule of the sample with the steel grade of Q235B and the slab steel grade of MBRY 23501, and the yield strength matching rule is stored in a database.

And respectively calculating the rules of yield strength, tensile strength and elongation of all steel types and slab steel types according to the method, and storing the rules into a database.

And 5, predicting the mechanical properties of the strip steel by utilizing an optimal rule.

Specifically, in this embodiment, the steel grade of the strip steel to be predicted is Q235B, the SLAB steel grade is Q235B, and the corresponding yield strength rule [ slab_thk,17.3345], [ P,0.0109], [ SI,0.0217], [ CR,0.014], [ CT,18.1087] is found in the database, and the rule is used to match all samples in the sample library, and the average value of the mechanical properties of the samples is taken as the mechanical property prediction value of the strip steel to be predicted.

Based on the above, the yield strength of 184 rolls of strip steel under the steel grade and the slab steel grade is predicted by extracting the data of Q345B-MBRY 34536 in the sample library, and when the error limit is 30MPa, the predicted hit rate is 91.3%, and the result is shown in FIG. 2.

In summary, the embodiment calculates the contribution degree of the characteristic variable to the mechanical property by using the chemical components, the technological parameters and the mechanical property of the existing strip steel in the sample library; taking the characteristic of a certain contribution degree as a rule characteristic and calculating the standard deviation of the rule characteristic; combining the characteristic variable and the standard deviation threshold to obtain a plurality of different rules; calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording; and predicting the mechanical properties of the strip steel by utilizing an optimal rule. Therefore, the mechanical property of the strip steel can be accurately predicted, the prediction generalization capability is improved, the on-site production can be guided, and the abnormality of the on-site data acquisition system is reflected.

Second embodiment

The embodiment provides an electronic device, which comprises a processor and a memory; wherein the memory stores at least one instruction that is loaded and executed by the processor to implement the method of the first embodiment.

The electronic device may vary considerably in configuration or performance and may include one or more processors (central processing units, CPU) and one or more memories having at least one instruction stored therein that is loaded by the processors and performs the methods described above.

Third embodiment

The present embodiment provides a computer-readable storage medium having stored therein at least one instruction that is loaded and executed by a processor to implement the method of the first embodiment. The computer readable storage medium may be, among other things, ROM, random access memory, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. The instructions stored therein may be loaded by a processor in the terminal and perform the methods described above.

Furthermore, it should be noted that the present invention can be provided as a method, an apparatus, or a computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the invention may take the form of a computer program product on one or more computer-usable storage media having computer-usable program code embodied therein.

Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It should also be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or terminal device comprising the element.

It is finally pointed out that the above description of the preferred embodiments of the invention, it being understood that although preferred embodiments of the invention have been described, it will be obvious to those skilled in the art that, once the basic inventive concepts of the invention are known, several modifications and adaptations can be made without departing from the principles of the invention, and these modifications and adaptations are intended to be within the scope of the invention. It is therefore intended that the following claims be interpreted as including the preferred embodiment and all such alterations and modifications as fall within the scope of the embodiments of the invention.

Claims (9)

1. A band steel mechanical property forecasting method based on rule optimizing is characterized by comprising the following steps:

collecting characteristic variables and mechanical properties of the strip steel, forming a sample library, and calculating the contribution degree of each characteristic variable to the mechanical properties; wherein the characteristic variables comprise the chemical components of the strip steel and the technological parameters of the strip steel;

sequencing the feature variables from high to low according to the respective contribution degrees, selecting the feature variables with the total contribution degrees being the preset number of preset values as rule composition features, and calculating the standard deviation of each rule composition feature;

determining a standard deviation threshold value of each rule composition characteristic based on the calculated standard deviation, and combining each rule composition characteristic with the standard deviation threshold value of each rule composition characteristic to obtain a plurality of different rules;

calculating the predicted hit rate of the mechanical property of the strip steel under each rule, and taking the rule with the highest hit rate as the optimal rule and recording;

predicting the mechanical properties of the strip steel to be predicted by utilizing the optimal rule to obtain a prediction result;

the calculating the contribution degree of each characteristic variable to the mechanical property comprises the following steps:

dividing the data in the sample library into a training set and a testing set; model training is carried out by utilizing characteristic variables and mechanical properties in the training set, and a random forest mechanical property regression model is obtained;

predicting the mechanical properties in the test set by using the random forest mechanical property regression model, comparing the prediction result with the real mechanical properties in the test set, and using the R2_score index as an initial regression scoring coefficient;

selecting one of the characteristic variables in the test set for random replacement and disturbing the sequence each time to obtain a new test set, respectively predicting the mechanical properties in the new test set by using the random forest mechanical property regression model after obtaining the new test set each time through the characteristic variables in the new test set, and comparing the prediction result with the real mechanical properties to obtain regression scoring coefficients after replacement of each characteristic variable;

and taking the absolute value of the difference between the regression scoring coefficient after each characteristic variable is replaced and the initial regression scoring coefficient and dividing the absolute value by the initial regression scoring coefficient to obtain a numerical value which represents the contribution degree of the replaced characteristic variable.

2. The rule-based optimizing strip steel mechanical property forecasting method of claim 1, wherein the chemical components comprise carbon, silicon, manganese, phosphorus, sulfur, nitrogen, chromium, copper, molybdenum, niobium, nickel, titanium and vanadium;

the process parameters comprise slab thickness, heating furnace inlet temperature, heating furnace outlet temperature, heating time, rough rolling outlet temperature, intermediate slab thickness, finish rolling outlet temperature and strip steel thickness.

3. The rule-based optimization method for predicting the mechanical properties of a strip steel of claim 1, wherein the mechanical properties include yield strength, tensile strength and elongation.

4. The method for predicting mechanical properties of a strip steel based on rule optimization of claim 1, wherein determining a standard deviation threshold for each rule composition feature based on the calculated standard deviation comprises:

based on the calculated standard deviation, k times the standard deviation of each rule constituent feature is taken as its corresponding standard deviation threshold.

5. The method for predicting mechanical properties of strip steel based on rule optimization as claimed in claim 4, wherein said combining each rule composition feature with a standard deviation threshold of each rule composition feature to obtain a plurality of different rules comprises:

randomly arranging and combining the rule composition characteristics and standard deviation threshold values of the rule composition characteristics to obtain a plurality of different rules; each rule respectively comprises all rule composition characteristics and any standard deviation threshold value of the rule composition characteristics corresponding to the rule composition characteristics.

6. The method for predicting mechanical properties of a strip steel based on rule optimization according to claim 1, wherein the calculating the predicted hit rate of the mechanical properties of the strip steel under each rule comprises:

dividing the data in the sample library into a training set and a testing set; and respectively matching all samples corresponding to each sample in the training set by using each rule, and calculating the predicted hit rate under the rule.

7. The method for predicting mechanical properties of strip steel based on rule optimization as claimed in claim 6, wherein the matching of all samples corresponding to each sample in the training set by each rule comprises:

under the current rule, taking one sample in the test set, matching the sample with characteristic variable in the positive and negative standard deviation threshold range in the training set according to the rule composition characteristic and the corresponding standard deviation threshold in the current rule, and taking the average value of the mechanical properties of all the matched samples as the prediction result of the corresponding sample in the test set.

8. The method for predicting mechanical properties of a strip steel based on rule optimization as claimed in claim 7, wherein said calculating the predicted hit rate under the rule comprises:

and (3) making differences between mechanical property prediction results and true values of all test set samples obtained under the current rule, and if the differences are within a preset error range, determining the differences as prediction hits, so as to calculate the prediction hit rate of the current rule.

9. The method for predicting mechanical properties of a strip steel based on rule optimization according to claim 7, wherein the predicting mechanical properties of the strip steel to be predicted by using the optimal rule comprises:

acquiring characteristic variables of strip steel to be predicted; according to rule composition characteristics and corresponding standard deviation threshold values in the current rule, matching samples of characteristic variables in a positive standard deviation threshold value and a negative standard deviation threshold value range in a sample library, and taking the average value of the mechanical properties of all the matched samples as the predicted mechanical property of the strip steel to be predicted.

Images