Production process control method of glass fiber geogrid based on machine learning

Abstract

The present invention relates to the field of regulation technology, and in particular to a method for regulating the production process of glass fiber geogrid based on machine learning. The method comprises: obtaining a yield stage variation characterization value and a strengthening stage variation characterization value of a test sample; obtaining a stress-strain curve mutation characterization value corresponding to the test sample according to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, obtaining a direction for regulating the production process of the glass fiber geogrid according to the stress-strain curve mutation characterization value, and regulating the production process of the glass fiber geogrid according to the regulated direction. The present invention can determine the direction of regulation more accurately and reliably, and subsequently can regulate the production process of the glass fiber geogrid more reliably and efficiently according to the determined regulation direction.

Description

Technical Field

[0001] The present invention relates to the field of control technology, and in particular to a production process control method for glass fiber geogrid based on machine learning. Background Art

[0002] The high tensile strength of glass fiber geogrid can not only significantly improve the mechanical properties of pavement and roadbed, but also perform well in deformation control, weather resistance, durability and other aspects. Therefore, glass fiber geogrid has become an indispensable material in highway construction. For example, glass fiber geogrid is widely used in the reinforcement of asphalt pavement, cement pavement and roadbed, railway roadbed, dam slope protection, airport runway and other engineering projects.

[0003] The material ratio and production temperature of glass fiber geogrids produced by current manufacturers are usually default values ​​or empirical values, and the default values ​​and empirical values ​​are the material ratio and production temperature determined by relevant personnel through experiments, tests, etc. The production temperature refers to the temperature in the production stage, such as the temperature specified in the curing stage, the temperature specified in the heat treatment stage, etc. However, due to the differences in the quality and source of the raw materials used in the production of glass fiber geogrids, the differences in raw materials may cause the glass fiber geogrids produced according to the default values ​​or empirical values ​​to have poor tensile properties or poor tensile strength properties. The reason is that the material ratio and production temperature are crucial to the tensile properties of the glass fiber geogrid or the material ratio and production temperature will cause changes in the chemical composition, crystal structure, etc. of the glass fiber geogrid, and the changes in the chemical composition and crystal structure of the glass fiber geogrid will affect the mechanical properties and mechanical properties of the material, thereby causing changes in the tensile properties or tensile strength properties of the glass fiber geogrid. At present, in order to make the produced glass fiber geogrid have high tensile strength properties, and to ensure smooth production and improve economic benefits, the production process of the glass fiber geogrid is usually regulated when using the same batch of raw materials for the production of glass fiber geogrid. The production process refers to the material ratio and production temperature when producing glass fiber geogrid, that is, it is crucial to regulate the production process of the glass fiber geogrid.

[0004] In the prior art, a tensile performance test is usually performed on a portion of the glass fiber geogrid cut from the produced glass fiber geogrid, and then the test results are directly used to determine whether the production process in the subsequent production process needs to be regulated. However, this existing method cannot accurately determine the specific direction of regulation during regulation, which will result in the inability to quickly, efficiently or effectively regulate the production process of the glass fiber geogrid in the subsequent production process. It will also waste a lot of time in the regulation stage, resulting in low production efficiency or damage to the economic interests of the manufacturer. Summary of the invention

[0005] In order to solve the above problems, the present invention provides a production process control method of glass fiber geogrid based on machine learning, and the technical scheme adopted is as follows:

[0006] An embodiment of the present invention provides a method for controlling the production process of a glass fiber geogrid based on machine learning, comprising the following steps:

[0007] Obtaining test samples from the produced glass fiber geogrids, and performing a tensile property test on each test sample to obtain a stress-strain curve corresponding to each test sample;

[0008] According to the derivative value of each data point on the stress-strain curve, the derivative value of the adjacent data point of each data point and the peak data point on the stress-strain curve, the yield stage starting point, the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve are obtained;

[0009] Obtaining a yield stage variation characterization value of the test sample according to a yield stage starting point and a yield stage ending point on the stress-strain curve;

[0010] According to the end point of the yield stage, the end point of the strengthening stage and the fracture point on the stress-strain curve, a strengthening stage variation characterization value of the test sample is obtained;

[0011] According to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, the stress-strain curve mutation characterization value corresponding to the test sample is obtained, and according to the stress-strain curve mutation characterization value, the direction for regulating the production process of the glass fiber geogrid is obtained, and the production process of the glass fiber geogrid is regulated according to the regulation direction.

[0012] Beneficial effects: The present invention first obtains a test sample, and performs a tensile property test on each test sample to obtain a stress-strain curve corresponding to each test sample; then, according to the derivative values ​​of each data point on the stress-strain curve, the derivative values ​​of the adjacent data points of each data point, and the peak data point on the stress-strain curve, the yield stage starting point, the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve are obtained, and according to the yield stage starting point and the yield stage ending point on the stress-strain curve, the yield stage variation characterization value of the test sample is obtained, and according to the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve, the strengthening stage variation characterization value of the test sample is obtained; finally, according to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, the stress-strain curve mutation characterization value corresponding to the test sample is obtained, according to the stress-strain curve mutation characterization value, the direction for regulating the production process of the glass fiber geogrid is obtained, and the production process of the glass fiber geogrid is regulated according to the regulation direction. The present invention can determine the control direction more accurately and reliably through the sudden change characterization value of the stress-strain curve, and subsequently the production process of the glass fiber geogrid can be controlled more quickly, efficiently and effectively according to the determined control direction. This control method can not only make the produced glass fiber geogrid have high tensile strength performance, but also have higher economic benefits and production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 The present invention is a flow chart of a method for controlling the production process of glass fiber geogrid based on machine learning. DETAILED DESCRIPTION

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the embodiments of the present invention.

[0016] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0017] This embodiment provides a method for controlling the production process of glass fiber geogrid based on machine learning, which is described in detail as follows:

[0018] like Figure 1 As shown, the production process control method of glass fiber geogrid based on machine learning includes the following steps:

[0019] Step S001, obtaining test samples from produced glass fiber geogrids, and performing a tensile property test on each test sample to obtain a stress-strain curve corresponding to each test sample.

[0020] The raw materials used in the production process control process targeted by this embodiment belong to the same batch, that is, in this embodiment, whether it is the finished glass fiber geogrid and test samples, or the products produced after the subsequent production process control, the raw materials used all belong to the same batch; and the glass fiber geogrid produced in this embodiment or the glass fiber geogrid to be produced subsequently are of the same type and specifications; in addition, in this embodiment, the production process mainly refers to the material ratio and production temperature when producing the glass fiber geogrid, and the production temperature mainly refers to the temperature in the production stage, such as the temperature specified in the curing stage and the temperature specified in the heat treatment stage.

[0021] In order to make the produced glass fiber geogrid have high tensile strength performance, and to ensure smooth production and improve economic benefits, this embodiment needs to select or cut out samples for subsequent tensile performance testing from the produced glass fiber geogrid, and perform tensile performance testing on the obtained test samples to obtain a stress-strain curve, and then analyze the yield stage, strengthening stage, etc. on the stress-strain curve to determine the yield stage variation characterization value and the strengthening stage variation characterization value, and then conduct a comprehensive analysis of the obtained yield stage variation characterization value and the strengthening stage variation characterization value to obtain the stress-strain curve mutation characterization value. Since the stress-strain curve mutation characterization value can reflect whether the tensile properties of the corresponding test sample have undergone a mutation and the reasons for the mutation, finally, the purpose of reliably and effectively regulating the production process is achieved based on the stress-strain curve mutation characterization value.

[0022] Therefore, based on the above analysis, it can be known that the test samples need to be obtained first, and the process of obtaining the test samples in this embodiment is: obtain a batch of raw materials for producing glass fiber geogrids, a production line for producing glass fiber geogrids, or a manufacturer for producing glass fiber geogrids, and record them as the target batch of raw materials, target production line, or target manufacturer, and then the target production line or target manufacturer first uses the target batch of raw materials to produce some glass fiber geogrid products in the early production period, and these produced glass fiber geogrids are all recorded as glass fiber geogrids to be cut, and then cut on the glass fiber geogrid to be cut, and the cut glass fiber geogrids are all recorded as test samples; and in the early production period, the material ratio and production temperature of the target production line or target manufacturer when producing glass fiber geogrids are default values ​​or empirical values. In addition, in this embodiment, the size and shape of each test sample are required to be consistent; and as other implementation methods, the size and shape of the test samples may also be required to be inconsistent.

[0023] In specific applications, the implementer needs to set the pre-production time period and the number of test samples cut according to actual conditions. For example, in this embodiment, the pre-production time period can be set to the time period from the initial production moment to the current production moment of the target production line or the target manufacturer as the pre-production time period, and this embodiment requires that the time length corresponding to the pre-production time period is shorter, but requires that the glass fiber geogrid finished product has been produced in the pre-production time period, and the initial production moment refers to the moment when the target production line or the target manufacturer first uses the raw material for production, and this embodiment can also set the number of test samples cut to 5 or 10.

[0024] Therefore, this embodiment can obtain all test samples through the above process, and after obtaining the test samples, a tensile property test is performed on each test sample, and a stress-strain curve corresponding to each test sample is obtained. In this embodiment, the experimental or test conditions of all test samples are required to be the same; in addition, for ease of understanding, this embodiment takes the process of obtaining the stress-strain curve corresponding to any test sample as an example for description, that is, the process of obtaining the stress-strain curve corresponding to the test sample is:

[0025] First, a tensile performance test is performed on the test sample, and the deformation of the test sample under various tensile forces in the time period from the test to the end of the test is statistically obtained; for example, if there is a tensile force a0, then the deformation of the test sample under the tensile force a0 refers to the deformation of the test sample when the tensile force applied to the test sample is a0; and the detailed process of performing the tensile performance test is a well-known technology, so this embodiment will not be described in detail.

[0026] In addition, the end of the test in this embodiment means that the corresponding test sample is broken, that is, the breakage of the corresponding test sample is the condition for the end of the test.

[0027] Then, according to the deformation of the test sample under the action of each tensile force in the time period from the test of the test sample to the end of the test, a set of vectors to be analyzed corresponding to the test sample is constructed, and the vectors to be analyzed in the set of vectors to be analyzed corresponding to the test sample are composed of the tensile force and the deformation of the test sample under the corresponding tensile force; for example, if there is a tensile force a0 acting on the test sample in the time period from the test of the test sample to the end of the test, and the deformation of the test sample under the action of the tensile force a0 is a1, then there is a vector to be analyzed composed of a0 and a1 in the set of vectors to be analyzed.

[0028] Afterwards, according to each vector to be analyzed in the set of vectors to be analyzed corresponding to the test sample, the stress and strain corresponding to each vector to be analyzed are obtained, and then the stress and strain corresponding to all the vectors to be analyzed are input into the preset model, and the curve output by the preset model is recorded as the stress-strain curve corresponding to the test sample, and the vector to be analyzed at this time belongs to the set of vectors to be analyzed corresponding to the test sample, and the ordinate of the data point on the stress-strain curve is stress, and the abscissa is strain; in addition, since stress-strain is a nonlinear relationship, in order to obtain a more accurate stress-strain curve, this embodiment uses the model in the ‌Origin software to achieve the purpose of obtaining the stress-strain curve, that is, the preset model in this embodiment is the model in the ‌Origin software, and ‌Origin is a professional data analysis software that can provide a powerful nonlinear fitting function; and since the process of obtaining the stress-strain curve based on the model in the ‌Origin software is a well-known technology, this embodiment will not be described in detail; and as other implementation methods, other methods can also be used to obtain the stress-strain curve, such as polynomial regression and SVR algorithm.

[0029] In addition, in this embodiment, the process of obtaining the stress and strain corresponding to each vector to be analyzed in the set of vectors to be analyzed corresponding to the test sample is also a well-known technology. For example, for the bth vector to be analyzed in the set of vectors to be analyzed corresponding to the test sample, the stress corresponding to the bth vector to be analyzed refers to the ratio of the tension value in the bth vector to be analyzed to the original cross-sectional area, and the strain corresponding to the bth vector to be analyzed refers to the ratio of the deformation value in the bth vector to be analyzed to the original length, and the original cross-sectional area here refers to the cross-sectional area of ​​the test sample before the tensile test, and the original length refers to the length of the test sample before the tensile test.

[0030] Therefore, this embodiment can obtain the stress-strain curve corresponding to each test sample through the above process; and in this embodiment, each stress-strain curve is segmented using the preset strain value, and the segmentation position on the stress-strain curve is used as the data point on the corresponding stress-strain curve, and this embodiment requires that the value of the preset strain value is small, such as 0.01 or 0.05 in this embodiment; in addition, the starting position and the ending position on the stress-strain curve are also data points.

[0031] Step S002, according to the derivative values ​​of each data point on the stress-strain curve, the derivative values ​​of the adjacent data points of each data point and the peak data point on the stress-strain curve, obtain the yield stage starting point, yield stage ending point, strengthening stage ending point and fracture point on the stress-strain curve; according to the yield stage starting point and yield stage ending point on the stress-strain curve, obtain the yield stage variation characterization value of the test sample; according to the yield stage ending point, strengthening stage ending point and fracture point on the stress-strain curve, obtain the strengthening stage variation characterization value of the test sample.

[0032] Based on the analysis in the above step S001, it can be known that this embodiment mainly achieves the purpose of reliable and effective regulation of the production process by obtaining the yield stage variation characterization value and the strengthening stage variation characterization value. The reason why the production process can be reliably and effectively regulated based on the mutation characterization value in this embodiment is that the material ratio and production temperature will have a significant impact on the tensile properties of the glass fiber geogrid, and the changes in the tensile properties caused by the material ratio and production temperature will cause the corresponding stress-strain curve to also change significantly, especially in the yield stage and strengthening stage of the curve. The performance is extremely obvious, and the yield stage refers to the process in which the test sample changes from elastic deformation to plastic deformation after being subjected to tension, and the strengthening stage is the stage when the test sample continues to apply tension after yielding. These two stages contain a variety of mechanical properties of the test sample, including elastic modulus, yield strength, ductility, etc. Therefore, by analyzing the curve change characteristics of the yield and strengthening stages in the stress-strain curve, it is possible to more reliably determine whether the tensile properties of the corresponding test sample have undergone a mutation and the cause of the mutation, thereby achieving the purpose of reliable and effective regulation of the production process.

[0033] Therefore, it can be seen that the present embodiment needs to first obtain the yield stage variation characterization value and the strengthening stage variation characterization value, and before obtaining the yield stage variation characterization value and the strengthening stage variation characterization value, it is necessary to first determine the yield stage starting point, the yield stage end point, the strengthening stage end point and the fracture point on the stress-strain curve. Therefore, the present embodiment will then obtain the yield stage starting point, the yield stage end point, the strengthening stage end point and the fracture point on the stress-strain curve, and the specific acquisition process is:

[0034] First, the derivative values ​​of each data point on the stress-strain curve and all peak data points on the stress-strain curve are obtained, wherein the peak data points include trough data points and peak data points, and then the yield stage starting point, yield stage ending point, strengthening stage ending point and fracture point on the stress-strain curve are obtained according to the derivative values ​​of each data point on the stress-strain curve, the derivative values ​​of the adjacent data points of each data point and the peak data points on the stress-strain curve; and for ease of understanding, this embodiment takes the process of obtaining the yield stage starting point, yield stage ending point, strengthening stage ending point and fracture point on any stress-strain curve C as an example for description, specifically:

[0035] First, the first derivative difference of each data point on the stress-strain curve C is obtained, and then the data points on the stress-strain curve C are traversed from left to right, and the data point whose first derivative difference is not equal to the preset derivative difference threshold is taken as the starting point of the yield stage on the stress-strain curve C, and the first derivative difference of the cth data point on the stress-strain curve C is the difference between the derivative value of the cth data point and the derivative value of the c+1th data point on the stress-strain curve C, where c is not equal to the total number of data points on the stress-strain curve C.

[0036] Then, the maximum value point on the stress-strain curve C is obtained and recorded as the end point of the strengthening stage on the stress-strain curve C, and the maximum value point refers to the point on the stress-strain curve C with the largest ordinate value; then all peak data points between the yield starting point and the end point of the strengthening stage on the stress-strain curve C are obtained, and are all recorded as data points to be analyzed, and then the data point to be analyzed closest to the end point of the strengthening stage on the stress-strain curve C is obtained, and used as the end point of the yield stage on the stress-strain curve C, and the end point of the yield stage on the stress-strain curve C refers to the data point to be analyzed with the smallest difference in the horizontal coordinate value with the end point of the strengthening stage on the stress-strain curve C; and it should be noted that when the starting point of the yield stage is reached, the microstructure of the glass fiber geogrid will cause the curve to have slight fluctuations, resulting in multiple local peaks and valleys, and after the end of the yield stage, the curve stops fluctuating, and the stress and strain gradually increase until the tensile strength limit value, that is, the maximum value point, is reached.

[0037] Then, on the stress-strain curve C, the data point corresponding to the maximum horizontal coordinate value is obtained and recorded as the fracture point on the stress-strain curve C. The fracture point refers to the point corresponding to the fracture of the test sample corresponding to the stress-strain curve C. The process of obtaining the peak data point and the derivative value of the data point on the curve is a well-known technology, so it will not be described in detail; in addition, in a specific application, the implementer needs to set a preset derivative difference threshold according to the actual situation, such as the preset derivative difference threshold is 0 in this embodiment.

[0038] After obtaining the yield stage variation characterization value and the strengthening stage variation characterization value, the present embodiment obtains the yield stage variation characterization value of the test sample according to the yield stage strain difference value, the yield stage starting point on the stress-strain curve, and the first mean and the second mean. The yield stage variation characterization value is an important parameter for subsequently determining the stress-strain curve mutation characterization value. Therefore, in the present embodiment, the process of obtaining the yield stage variation characterization value is as follows:

[0039] First, according to the yield stage starting point and the yield stage ending point on the stress-strain curve corresponding to each test sample, the yield stage strain difference value corresponding to each test sample is obtained, and the yield stage strain difference value corresponding to the test sample refers to the absolute value of the difference between the horizontal coordinate of the yield stage starting point and the yield stage ending point on the stress-strain curve corresponding to the test sample; for example, for any test sample, the yield stage strain difference value corresponding to the test sample refers to the absolute value of the difference between the horizontal coordinate of the yield stage starting point on curve R and the horizontal coordinate of the yield stage ending point on curve R, and the curve R refers to the stress-strain curve corresponding to the test sample; then the mean of the yield stage strain difference values ​​corresponding to all test samples is obtained and recorded as the first mean; then the second set is obtained, and the mean of the second set is recorded as the second mean, and the second set is composed of the vertical coordinate values ​​of the yield stage starting point on the stress-strain curve corresponding to all test samples. Then, according to the yield stage starting point and the yield stage ending point on the stress-strain curve, as well as the first mean and the second mean, the yield stage variation characterization value of the test sample is obtained.

[0040] In this embodiment, according to the yield stage strain difference value, the yield stage starting point on the stress-strain curve, and the first mean and the second mean, the specific process of obtaining the yield stage variation characterization value of the test sample is as follows: for any test sample F, first obtain the absolute value of the difference between the yield stage strain difference value corresponding to the test sample F and the first mean, and record it as the first characterization value; then obtain the absolute value of the difference between the ordinate value of the yield stage starting point on the stress-strain curve corresponding to the test sample F and the second mean, and record it as the second characterization value; then obtain the normalized value after adding the first characterization value and the second characterization value, and record it as the yield stage variation characterization value of the test sample F; and the specific expression for obtaining the yield stage variation characterization value of the test sample F is:

[0041]

[0042] in, is the yield stage variation characterization value of the test sample F, tanh() is the hyperbolic tangent function, and its main function in this embodiment is to achieve the purpose of normalization. is the strain difference value of the yield stage corresponding to the test sample F, S1 is the first mean value, is the ordinate value of the starting point of the yield stage on the stress-strain curve corresponding to the test sample F, and S2 is the second mean value.

[0043] And when The larger the value of The larger the value of The larger the value of is, the greater the probability that there is a sudden change or abnormal change in the yield stage of the stress-strain curve corresponding to the test sample F due to the material ratio and production temperature during production. The smaller the value, the smaller the probability that there is a sudden change or abnormal change in the yield stage on the stress-strain curve corresponding to the test sample F due to the material ratio and production temperature during production; and it should be noted that the change in the yield stage on the stress-strain curve corresponding to the test sample F can reflect the tensile properties of the corresponding sample. The main reason is that the yield stage refers to the process in which the test sample changes from elastic deformation to plastic deformation after being subjected to tension, which is related to the bearing capacity.

[0044] Therefore, the yield stage variation characterization value of each test sample can be obtained through the above process. After the yield stage variation characterization value is obtained, the strengthening stage variation characterization value of the test sample is obtained according to the yield stage end point, the strengthening stage end point and the fracture point on the stress-strain curve. The strengthening stage variation characterization value is an important parameter for subsequently determining the stress-strain curve mutation characterization value. Therefore, in this embodiment, the process of obtaining the strengthening stage variation characterization value is as follows:

[0045] First, the third set is obtained, and the mean of the third set is recorded as the third mean, and the third set is constructed by the ordinate values ​​of all strengthening stage end points in the strengthening stage end point set, and the strengthening stage end point set is constructed by the strengthening stage end points on the stress-strain curve corresponding to all test samples; then the strengthening stage curvature corresponding to each test sample is obtained, and the mean of the strengthening stage curvature corresponding to all test samples is recorded as the fourth mean; then the fifth set is obtained, and the mean of the fifth set is recorded as the fifth mean, and the fifth set is constructed by the ordinate values ​​of all breaking points in the breaking point set, and the breaking point set is constructed by the breaking points on the stress-strain curve corresponding to all test samples. Then, according to the yield stage end point, strengthening stage end point and breaking point on the stress-strain curve, as well as the third mean, the fourth mean and the fifth mean, the strengthening stage variation characterization value of the test sample is obtained.

[0046] And in this embodiment, the yield stage end point, strengthening stage end point and fracture point on the stress-strain curve, as well as the third mean, fourth mean and fifth mean, the specific process of obtaining the strengthening stage variation characterization value of the test sample is as follows: for the test sample F, first obtain the ratio of the ordinate value of the strengthening stage end point on the stress-strain curve corresponding to the test sample F to the third mean, and record it as the first ratio; then obtain the ratio of the strengthening stage curvature corresponding to the test sample F to the fourth mean, and record it as the second ratio; then obtain the ratio of the ordinate value of the fracture point on the stress-strain curve corresponding to the test sample F to the fifth mean, and record it as the third ratio; then obtain the normalized value after adding the first ratio and the second ratio, and use it as the first index value, obtain the normalized value of the third ratio, and use it as the second index value, then perform weighted summation on the first index value and the second index value, and record the result of the weighted summation as the strengthening stage variation characterization value of the test sample F. And the specific expression for obtaining the strengthening stage variation characterization value of the test sample F is:

[0047]

[0048] in, is the variation characterization value of the strengthening stage of the test sample F, is the first weight value, is the second weight value, is the ordinate value of the end point of the strengthening stage on the stress-strain curve corresponding to the test sample F, S3 is the third mean value, is the curvature of the strengthening stage corresponding to the test sample F, S4 is the fourth mean, is the ordinate value of the breaking point on the stress-strain curve corresponding to the test sample F, S5 is the fifth mean value, is the first ratio, is the second ratio, The third ratio.

[0049] And when , as well as The larger the value of The larger the value of The larger the value of is, the greater the probability that the ultimate strength of the test sample F has changed, and the greater the probability that there is a sudden change or abnormal change in the strengthening stage of the stress-strain curve corresponding to the test sample F due to the material ratio and production temperature during production. The smaller the value, the smaller the probability that there is a sudden change or abnormal change in the strengthening stage of the stress-strain curve corresponding to the test sample F due to the material ratio and production temperature during production; and it should be noted that the change of the strengthening stage on the stress-strain curve corresponding to the test sample F can reflect the tensile properties of the corresponding sample, mainly because the strengthening stage is related to the deformation ability and ductility of the corresponding sample.

[0050] In addition, in a specific application, the implementer needs to set the first weight value and the second weight value according to the actual situation. For example, in this embodiment, the first weight value and the second weight value are set to 0.7 and 0.3 respectively.

[0051] In this embodiment, the method for obtaining the strengthening stage curvature corresponding to the test sample is: for the test sample F, the curve between the end point of the yield stage and the end point of the strengthening stage on the stress-strain curve corresponding to the test sample F is recorded as the strengthening stage curve corresponding to the test sample F, and then the curvature of the strengthening stage curve corresponding to the test sample F is obtained and recorded as the strengthening stage curvature corresponding to the test sample F, and the curvature of the strengthening stage curve corresponding to the test sample F refers to the mean of the curvatures corresponding to all data points on the strengthening stage curve corresponding to the test sample F.

[0052] Therefore, this embodiment can obtain the enhancement stage variation characterization value of each test sample through the above process.

[0053] Step S003, obtaining the stress-strain curve mutation characterization value corresponding to the test sample according to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, obtaining the direction for regulating the production process of the glass fiber geogrid according to the stress-strain curve mutation characterization value, and regulating the production process of the glass fiber geogrid according to the regulation direction.

[0054] After obtaining the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, this embodiment performs a comprehensive analysis to obtain the stress-strain curve mutation characterization value corresponding to the test sample, which is specifically:

[0055] For any test sample F: perform weighted summation on the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample F, and use the result of the weighted summation as the stress-strain curve mutation characterization value corresponding to the test sample F. The specific expression for obtaining the stress-strain curve mutation characterization value corresponding to the test sample F is: ,in, and are the third weight value and the fourth weight value respectively, and in this embodiment and The value of is 0.5.

[0056] Therefore, through the above process, the stress-strain curve mutation characterization value corresponding to each test sample F can be obtained, and the larger the stress-strain curve mutation characterization value, the more likely it is that the test sample has undergone a sudden change or mutation in the tensile properties caused by the material ratio or production temperature, that is, the larger the stress-strain curve mutation characterization value, the greater the probability that the tensile properties of the test sample do not meet the specified tensile properties requirements, and the greater the probability that the tensile properties of the test sample do not meet the specified tensile properties requirements, the greater the probability that the glass fiber geogrid products produced by the target production line or target manufacturer using the target batch of raw materials in the early production period do not meet the specified tensile properties requirements, and vice versa, the greater the probability that they meet the specified tensile properties requirements.

[0057] Therefore, this embodiment will obtain the direction of regulating the production process of the glass fiber geogrid according to the stress-strain curve mutation characterization value, and regulate the production process of the glass fiber geogrid according to the regulation direction. The specific process is:

[0058] Among the stress-strain curve mutation characterization values ​​corresponding to all test samples, it is determined whether there is no stress-strain curve mutation characterization value greater than the preset mutation characterization value threshold. If so, it indicates that the tensile properties of the glass fiber geogrid produced according to the material ratio and production temperature used in the previous production period are qualified. Therefore, it is determined that there is no need to adjust the production process of the glass fiber geogrid, that is, the target production line or target manufacturer will continue to produce according to the material ratio and production temperature used in the previous production period until the target batch of raw materials is used up. Otherwise, it indicates that the tensile properties of the glass fiber geogrid produced according to the material ratio and production temperature used in the previous production period are unqualified, and the material ratio and production temperature used in the production of the glass fiber geogrid need to be adjusted. The specific adjustment process is:

[0059] First, the test sample corresponding to the largest stress-strain curve mutation characterization value is selected as the test sample to be analyzed from all stress-strain curve mutation characterization values ​​greater than the preset mutation characterization value threshold value. After obtaining the test sample to be analyzed, it is determined whether the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed are both greater than the preset judgment threshold value. If so, it is determined that the material ratio and production temperature used in the production process of glass fiber geogrid need to be adjusted at the same time, and the adjusted material ratio and production temperature are obtained, and the subsequent target production line or target manufacturer will produce according to the adjusted material ratio and production temperature. , until the target batch of raw materials is used up; and when it is determined that the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed are not both greater than the preset judgment threshold, it is necessary to continue to determine whether the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed are equal. If they are equal, it is also determined that the material ratio and production temperature for producing the glass fiber geogrid need to be adjusted at the same time, and the adjusted material ratio and production temperature are obtained, and then the subsequent target production line or target manufacturer produces according to the adjusted material ratio and production temperature until the target batch of raw materials is used up; but if it is determined that If the yield stage variation characterization value and the strengthening stage variation characterization value of the analysis test sample are not equal, then it is necessary to determine whether the yield stage variation characterization value of the test sample to be analyzed is greater than the strengthening stage variation characterization value of the test sample to be analyzed. If so, it indicates that the sudden change of the stress-strain curve is mainly caused by the production temperature. Therefore, it is determined at this time that only the production temperature used to produce the glass fiber geogrid can be adjusted, and the adjusted production temperature can be obtained. Then, the subsequent target production line or target manufacturer will continue to produce according to the material ratio used in the previous production time period and the adjusted production temperature until the target batch of raw materials is used up; If it is judged that the yield stage variation characterization value of the test sample to be analyzed is not greater than the strengthening stage variation characterization value of the test sample to be analyzed, that is, the yield stage variation characterization value of the test sample to be analyzed is less than the strengthening stage variation characterization value of the test sample to be analyzed, it indicates that the sudden change of the stress-strain curve is mainly caused by the material ratio. At this time, it is determined that only the material ratio used to produce the glass fiber geogrid needs to be adjusted, and the adjusted material ratio is obtained. Then, the subsequent target production line or target manufacturer will continue to produce according to the production temperature and the adjusted material ratio used in the previous production time period until the target batch of raw materials is used up.

[0060] In a specific application, the implementer needs to set the preset mutation characterization value threshold and the preset judgment threshold according to the actual situation. For example, in this embodiment, the preset mutation characterization value threshold can be set to 0.6, and the preset judgment threshold can be set to 0.8. In addition, the specific process of obtaining the adjusted material ratio or the adjusted production temperature is a well-known technology, so it will not be described in detail in this embodiment.

[0061] In summary, in this embodiment, the test samples are first obtained, and the tensile properties test is performed on each test sample to obtain the stress-strain curve corresponding to each test sample; then, according to the derivative values ​​of each data point on the stress-strain curve, the derivative values ​​of the adjacent data points of each data point, and the peak data point on the stress-strain curve, the yield stage starting point, the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve are obtained, and according to the yield stage starting point and the yield stage ending point on the stress-strain curve, the yield stage variation characterization value of the test sample is obtained, and according to the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve, the strengthening stage variation characterization value of the test sample is obtained; finally, according to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, the stress-strain curve mutation characterization value corresponding to the test sample is obtained, according to the stress-strain curve mutation characterization value, the direction of regulating the production process of the glass fiber geogrid is obtained, and the production process of the glass fiber geogrid is regulated according to the direction of regulation. This embodiment can determine the control direction more accurately and reliably through the sudden change characterization value of the stress-strain curve, and subsequently the production process of the glass fiber geogrid can be controlled more quickly and efficiently according to the determined control direction. This control method can not only make the produced glass fiber geogrid have high tensile strength performance, but also have higher economic benefits and production efficiency.

[0062] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for controlling the production process of glass fiber geogrid based on machine learning, characterized in that: The method comprises the following steps: Obtaining test samples from the produced glass fiber geogrids, and performing a tensile property test on each test sample to obtain a stress-strain curve corresponding to each test sample; According to the derivative value of each data point on the stress-strain curve, the derivative value of the adjacent data point of each data point and the peak data point on the stress-strain curve, the yield stage starting point, the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve are obtained; Obtaining a yield stage variation characterization value of the test sample according to a yield stage starting point and a yield stage ending point on the stress-strain curve; Obtaining a strengthening stage variation characterization value of the test sample according to the yield stage end point, the strengthening stage end point and the fracture point on the stress-strain curve; According to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample, the stress-strain curve mutation characterization value corresponding to the test sample is obtained, and according to the stress-strain curve mutation characterization value, the direction for regulating the production process of the glass fiber geogrid is obtained, and the production process of the glass fiber geogrid is regulated according to the regulation direction.

2. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 1, characterized in that: The method for obtaining the stress-strain curve comprises: For any test sample: perform a tensile performance test on the test sample, and obtain statistically the deformation of the test sample under each tensile force during the test of the test sample; construct a set of vectors to be analyzed corresponding to the test sample according to the deformation of the test sample under each tensile force, wherein the vectors to be analyzed in the set of vectors to be analyzed are composed of the tensile force and the deformation of the test sample under the corresponding tensile force; obtain the stress and strain corresponding to the vector to be analyzed according to the parameters in the vector to be analyzed, and input the stress and strain corresponding to all the vectors to be analyzed into a preset model, and record the curve output by the preset model as the stress-strain curve corresponding to the test sample, wherein the ordinate of the stress-strain curve is stress and the abscissa is strain.

3. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 1, characterized in that: The method for obtaining the yield stage starting point, the yield stage ending point, the strengthening stage ending point and the fracture point on the stress-strain curve comprises: For any stress-strain curve C: Obtaining the first derivative difference of each data point on the stress-strain curve C, and taking the data point on the stress-strain curve C where the first derivative difference is not equal to the preset derivative difference threshold as the starting point of the yield stage on the stress-strain curve C, the first derivative difference of the c-th data point on the curve C being the difference between the derivative value of the c-th data point and the derivative value of the c+1-th data point on the curve C; The maximum value point on the curve C is obtained and recorded as the end point of the strengthening stage on the curve C; all peak data points between the yield starting point on the curve C and the end point of the strengthening stage on the curve C are recorded as data points to be analyzed, and the data point to be analyzed closest to the end point of the strengthening stage on the curve C is taken as the end point of the yield stage on the curve C; the data point corresponding to the maximum horizontal coordinate value on the curve C is recorded as the breaking point on the curve C.

4. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 1, characterized in that: The method for obtaining the yield stage variation characterization value includes: According to the yield stage starting point and the yield stage ending point on the stress-strain curve corresponding to each test sample, the yield stage strain difference value corresponding to each test sample is obtained, and the yield stage strain difference value corresponding to the test sample refers to the absolute value of the horizontal coordinate difference between the yield stage starting point and the yield stage ending point on the stress-strain curve corresponding to the test sample; the mean of the yield stage strain difference values ​​corresponding to all the test samples is recorded as the first mean; the set constructed by the ordinate values ​​of the yield stage starting points on the stress-strain curves corresponding to all the test samples is recorded as the second set, and the mean of the second set is recorded as the second mean; The yield stage variation characterization value of the test sample is obtained according to the yield stage strain difference value, the yield stage starting point on the stress-strain curve, and the first mean value and the second mean value.

5. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 4, characterized in that: The method for obtaining the yield stage variation characterization value of the test sample according to the yield stage strain difference value, the yield stage starting point on the stress-strain curve, and the first mean and the second mean includes: For any test sample F: the absolute value of the difference between the yield stage strain difference value corresponding to the test sample F and the first mean value is recorded as the first characterization value; the absolute value of the difference between the ordinate value of the yield stage starting point on the stress-strain curve corresponding to the test sample F and the second mean value is recorded as the second characterization value; the normalized value obtained by adding the first characterization value and the second characterization value is recorded as the yield stage variation characterization value of the test sample F.

6. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 1, characterized in that: The method for obtaining the variation characterization value of the test sample during the strengthening phase includes: Recording a set constructed from the ordinate values ​​of the end points of the strengthening stage on the stress-strain curves corresponding to all the test samples as a third set, and recording the mean of the third set as a third mean; Obtaining the curvature of the strengthening stage corresponding to the test sample, and recording the average of the curvatures of the strengthening stage corresponding to all the test samples as a fourth average; Recording a set constructed from the ordinate values ​​of the breaking points on the stress-strain curves corresponding to all the test samples as a fifth set, and recording the mean of the fifth set as a fifth mean; For any test sample F: the ratio of the ordinate value of the end point of the strengthening stage on the stress-strain curve corresponding to the test sample F to the third mean is recorded as the first ratio; the ratio of the curvature of the strengthening stage corresponding to the test sample F to the fourth mean is recorded as the second ratio; the ratio of the ordinate value of the fracture point on the stress-strain curve corresponding to the test sample F to the fifth mean is recorded as the third ratio; the normalized value after adding the first ratio and the second ratio is used as the first index value, the normalized value of the third ratio is used as the second index value, the first index value and the second index value are weightedly summed, and the result of the weighted sum is recorded as the strengthening stage variation characterization value of the test sample F.

7. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 6, characterized in that: The method for obtaining the curvature of the strengthening stage corresponding to the test sample includes: For any test sample F: the curve between the end point of the yield stage and the end point of the strengthening stage on the stress-strain curve corresponding to the test sample F is recorded as the strengthening stage curve corresponding to the test sample F, and the curvature of the strengthening stage curve is obtained and recorded as the strengthening stage curvature corresponding to the test sample F.

8. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 1, characterized in that: The method for obtaining the stress-strain curve mutation characterization value corresponding to the test sample comprises: For any test sample F: perform weighted summation on the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample F, and use the result of the weighted summation as the stress-strain curve mutation characterization value corresponding to the test sample F.

9. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 1, characterized in that: According to the stress-strain curve mutation characterization value, a direction for regulating the production process of the glass fiber geogrid is obtained, and a method for regulating the production process of the glass fiber geogrid according to the regulated direction includes: Among the stress-strain curve mutation characterization values ​​corresponding to all test samples, determine whether there is no stress-strain curve mutation characterization value greater than the preset mutation characterization value threshold. If so, it is determined that the tensile properties of the produced glass fiber geogrid are qualified, and it is determined that there is no need to adjust the production process of the glass fiber geogrid. Otherwise, it is determined that the tensile properties of the produced glass fiber geogrid are unqualified, and the test sample corresponding to the largest stress-strain curve mutation characterization value is selected from all stress-strain curve mutation characterization values ​​greater than the preset mutation characterization value threshold as the test sample to be analyzed. According to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed, the direction of adjusting the production process of the glass fiber geogrid is obtained, and the production process of the glass fiber geogrid is adjusted according to the direction of adjustment.

10. The method for controlling the production process of glass fiber geogrid based on machine learning according to claim 9, characterized in that: According to the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed, the direction of regulating the production process of the glass fiber geogrid is obtained, and the method for regulating the production process of the glass fiber geogrid according to the regulated direction includes: Determine whether the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed are both greater than the preset judgment threshold value. If so, determine whether the material ratio and production temperature for producing the glass fiber geogrid are adjusted. Otherwise, continue to determine whether the yield stage variation characterization value and the strengthening stage variation characterization value of the test sample to be analyzed are equal. If they are equal, determine whether the material ratio and production temperature for producing the glass fiber geogrid are adjusted. If they are not equal, continue to determine whether the yield stage variation characterization value of the test sample to be analyzed is greater than the strengthening stage variation characterization value of the test sample to be analyzed. If so, determine that only the production temperature for producing the glass fiber geogrid is adjusted. Otherwise, determine that only the material ratio for producing the glass fiber geogrid is adjusted.

Images