Method and device for predicting mechanical property of hot-rolled seamless steel pipe based on integrated learning

Through the integrated learning method combined with multiple prediction models and particle swarm algorithms, automatic detection and high-precision prediction of the mechanical properties of hot-rolled seamless steel pipes are realized, and subjectivity and inconsistency problems of traditional manual sampling are solved, and detection efficiency and accuracy are improved. sex.

CN119005022APending Publication Date: 2024-11-22CHENGDE JIANLONG SPECIAL STEEL

Patent Information

Application Number
CN202411487773.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-24
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Traditional manual sampling methods have strong subjectivity, inconsistency and errors in evaluating the mechanical properties of hot-rolled seamless steel pipes, making it difficult to fully cover all production batches and products, resulting in the risk of missed inspection and misjudgment, and frequent inspections have increased production line shutdowns. Time and resource consumption.

Method used

Using an integrated learning method, the input features are constructed by obtaining the chemical composition, material parameters and production process parameters of hot-rolled seamless steel pipes, and the mechanical performance prediction is carried out in combination with BP neural network, support vector machine and fast learning network model, and the particle swarm algorithm is used. Determine the model weight coefficient to achieve automatic detection and high-precision prediction.

Benefits of technology

It improves the efficiency and accuracy of hot-rolled seamless steel pipe inspection, reduces production costs, replaces traditional manual sampling, and achieves a comprehensive and accurate reflection of the actual performance status of each steel pipe.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005022A_ABST
    Figure CN119005022A_ABST
Patent Text Reader

Abstract

The invention provides a hot-rolled seamless steel pipe mechanical property prediction method and device based on integrated learning, and relates to the technical field of metallurgy. By combining the chemical components, the material parameters and the production process parameters of the hot-rolled seamless steel pipe, constructing the input characteristics, and combining various prediction models such as the BP neural network model, the support vector machine model and the fast learning network model to predict the mechanical property parameters, the automatic detection of the mechanical property of the hot-rolled seamless steel pipe is realized. And integrated learning of various network models is realized through weight coefficients, the advantages of various prediction models are integrated, and compared with a traditional manual sampling inspection mode, the detection efficiency and accuracy of the hot-rolled seamless steel pipe are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of metallurgy, and in particular to an integrated learning method and device for predicting the mechanical properties of a hot-rolled seamless steel pipe. Background Art

[0002] As a key basic material, the mechanical properties of hot-rolled seamless steel pipes are one of the main indicators of steel pipe quality. The stability and prediction accuracy of the mechanical properties of hot-rolled seamless steel pipes are directly related to the production safety and product quality of downstream industries.

[0003] However, currently, steel companies mainly rely on traditional manual sampling methods to evaluate the mechanical properties of seamless steel pipes. Traditional manual sampling methods rely on the experience and skill level of operators, are highly subjective, and are prone to inconsistency and errors in results; and the sample size of random inspections is limited, making it difficult to fully cover all production batches and products, increasing the risk of missed inspections and misjudgments. Finally, frequent inspections and testing processes not only increase the downtime of the production line, but also consume a lot of resources and manpower, making it difficult to fully and accurately reflect the actual performance of each steel pipe. Summary of the invention

[0004] The present invention provides a method and device for predicting the mechanical properties of a hot-rolled seamless steel pipe based on integrated learning, which can realize automatic detection of the mechanical properties of the hot-rolled seamless steel pipe and improve the detection efficiency of the hot-rolled seamless steel pipe.

[0005] In the first aspect, the present invention provides an integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes, the method comprising: obtaining the chemical composition, material parameters and production process parameters of the hot-rolled seamless steel pipes during the rolling process; standardizing the chemical composition, material parameters and production process parameters to generate input features; based on the input features and various prediction models, obtaining the output features of various prediction models; the prediction models include BP neural network models, support vector machine models and fast learning network models; the output features include predicted mechanical property parameters of the hot-rolled seamless steel pipes; based on the output features of various prediction models and the weight coefficients of various prediction models, determining the mechanical property parameters of the hot-rolled seamless steel pipes.

[0006] In a possible implementation, the mechanical properties parameters of the hot-rolled seamless steel pipe are determined based on the output features of various prediction models and the weight coefficients of various prediction models, which also includes: obtaining the chemical composition, material parameters and production process parameters of multiple hot-rolled seamless steel pipes with known mechanical properties parameters, as well as the mechanical properties parameters of multiple hot-rolled seamless steel pipes; generating multiple input features based on the chemical composition, material parameters and production process parameters of multiple hot-rolled seamless steel pipes; generating multiple output features based on the mechanical properties parameters of multiple hot-rolled seamless steel pipes; generating training samples based on the multiple input features and the multiple output features; based on the training samples, respectively training the preset new neural network models to obtain various prediction models.

[0007] In a possible implementation, the mechanical properties parameters of the hot-rolled seamless steel pipe are determined based on the output characteristics of various prediction models and the weight coefficients of various prediction models. The method also includes: testing various prediction models based on the test set in the training sample to obtain the test results of various prediction models; based on the test results and the output characteristics in the test set, using the particle swarm algorithm to determine the weight coefficients of various prediction models.

[0008] In a possible implementation, based on the test results and the output features in the test set, a particle swarm algorithm is used to determine the weight coefficients of various prediction models, including: step 1, based on the test results and the output features in the test set, a particle swarm is generated, and the particle swarm includes multiple sets of weight coefficients corresponding to each test sample; and the particle swarm parameters are initialized. The particle swarm parameters include the number of particle populations, the maximum number of iterations, the individual learning factor, the social learning factor, the initial position of the randomly generated particles, the initial speed and the speed variation range. Step 2, calculate the individual extreme value corresponding to each test sample in the particle swarm, and the population extreme value of multiple test samples; step 3, based on the individual extreme value and population extreme value of the particle swarm, update the speed and position vector of each particle in the current iteration process; step 4, if the number of iterations is greater than the maximum number of iterations, the optimal solution is output; if the number of iterations is less than or equal to the maximum number of iterations, repeat steps 2, 3 and 4 until the iteration is exited; step 5, based on the optimal solution, determine the weight coefficients of various prediction models.

[0009] In a possible implementation, the individual extreme value corresponding to each test sample in the particle swarm and the population extreme value of multiple test samples are calculated, including: calculating the root mean square error of multiple particles corresponding to each test sample; determining the fitness of each particle based on the root mean square error of the multiple particles; determining the local optimal fitness of each test sample based on the fitness of each particle; determining the particle corresponding to the local optimal fitness of each test sample as the individual extreme value; determining the global optimal fitness of multiple test samples based on the local optimal fitness of each test sample; and determining the particle corresponding to the global optimal fitness as the population extreme value.

[0010] In a possible implementation, after determining the mechanical properties parameters of the hot-rolled seamless steel pipe based on the output characteristics of various prediction models and the weight coefficients of various prediction models, it also includes: determining the evaluation value of each mechanical properties parameter based on the mechanical properties parameters of the hot-rolled seamless steel pipe and preset mechanical properties indicators; the mechanical properties indicators are the reliability intervals of each mechanical properties parameter; based on the evaluation values ​​of each mechanical properties parameter, determining the interval coverage of each mechanical properties parameter; based on the interval coverage of each mechanical properties parameter, evaluating the predicted mechanical properties parameters and determining the evaluation results.

[0011] In a possible implementation, based on the evaluation value of each mechanical property parameter, determining the interval coverage of each mechanical property parameter includes: determining the interval coverage based on the following formula; ; Where PICP is the interval coverage, n is the number of mechanical property parameters, is the evaluation value of the i-th mechanical property parameter. If the i-th mechanical property parameter is in the interval ,but , if the i-th mechanical property parameter is not in the interval ,but ; is the benchmark value of mechanical performance parameters, is the allowable error of the predicted value.

[0012] In one possible implementation, the chemical composition includes at least one of the following: manganese, chromium, molybdenum, nickel, copper, vanadium, niobium, titanium, carbon, and silicon; the material parameters include wall thickness and outer diameter; the production process parameters include heating furnace outlet temperature, heating time, piercing machine inlet temperature, rolling mill inlet temperature, sizing machine inlet temperature, and sizing machine outlet temperature; the mechanical property parameters include yield strength, tensile strength, and elongation.

[0013] In the second aspect, an embodiment of the present invention provides an integrated learning device for predicting the mechanical properties of hot-rolled seamless steel pipes, the prediction device comprising: a communication module for acquiring the chemical composition, material parameters and production process parameters of the hot-rolled seamless steel pipes during the rolling process; a processing module for standardizing the chemical composition, material parameters and production process parameters to generate input features; based on the input features and various prediction models, output features of various prediction models are obtained; the prediction models include a BP neural network model, a support vector machine model and a fast learning network model; the output features include predicted mechanical property parameters of the hot-rolled seamless steel pipe; based on the output features of various prediction models and the weight coefficients of various prediction models, the mechanical property parameters of the hot-rolled seamless steel pipe are determined.

[0014] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a memory and a processor, the memory storing a computer program, the processor being used to call and run the computer program stored in the memory to perform the steps of the method described in the first aspect and any possible implementation method of the first aspect.

[0015] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and wherein when the computer program is executed by a processor, the steps of the method described in the first aspect and any possible implementation method of the first aspect are implemented.

[0016] The present invention provides an integrated learning method and device for predicting the mechanical properties of hot-rolled seamless steel pipes. The present invention combines the chemical composition, material parameters and production process parameters of hot-rolled seamless steel pipes to construct input features, and combines various prediction models such as BP neural network model, support vector machine model and fast learning network model to predict mechanical property parameters, thereby realizing automatic detection of the mechanical properties of hot-rolled seamless steel pipes. The integrated learning of various network models is realized through weight coefficients, and the advantages of various prediction models are combined. Compared with the traditional manual sampling method, the present invention improves the detection efficiency and accuracy of hot-rolled seamless steel pipes. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be 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 labor.

[0018] Figure 1 It is a schematic flow chart of a method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning provided by an embodiment of the present invention; Figure 2It is a schematic diagram of the elongation prediction result of a hot-rolled seamless steel pipe provided by an embodiment of the present invention; Figure 3 is a schematic structural diagram of a hot-rolled seamless steel pipe mechanical property prediction device using integrated learning provided by an embodiment of the present invention; Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0020] In the description of the present invention, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, "at least one" and "plurality" refer to two or more. The words "first", "second", etc. do not limit the quantity and execution order, and the words "first", "second", etc. do not limit them to be different.

[0021] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way for easy understanding.

[0022] In addition, the terms "including" and "having" and any variations thereof mentioned in the description of this application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but may optionally include other steps or modules that are not listed, or may optionally include other steps or modules that are inherent to these processes, methods, products or devices.

[0023] In order to make the purpose, technical solutions and advantages of the present invention more clear, the following will be described through specific embodiments in conjunction with the accompanying drawings of the present invention.

[0024] The production process of hot-rolled seamless steel pipe mainly includes the following steps: Preparation and inspection of tube billets: Select qualified steel billets or solid tube billets, carry out necessary pretreatment and inspection to ensure the quality of raw materials. Heating of tube billets: Send the tube billets into the heating furnace and heat them to an appropriate temperature to improve their plasticity and reduce deformation resistance. Perforation: In the perforator, the heated tube billets are inserted into the hollow thick-walled capillary to form a preliminary hollow tube. Rolling: The perforated tube billets are rolled by the tube rolling mill to further reduce the wall thickness and adjust the tube diameter to form a semi-finished tube. Reheating of steel pipes: According to the process requirements, the semi-finished tubes are heated again to meet the temperature requirements of subsequent processes. Sizing (reducing): The rolled steel pipes are precisely adjusted by sizing machines or reducing machines to meet the specified size and shape requirements. Heat treatment: The steel pipes are subjected to heat treatments such as annealing and normalizing to improve their organizational structure and mechanical properties. Straightening of finished pipes: The heat-treated steel pipes are straightened to eliminate defects such as bending and twisting. Finishing: The steel pipes are subjected to finishing treatments such as cutting heads, cutting tails, and surface treatment to improve the appearance quality and surface finish of the products. Inspection and warehousing: The finished steel pipes are subjected to strict quality inspections, including non-destructive testing, physical and chemical analysis, and dimensional measurement, and are put into storage after passing the inspections.

[0025] The present invention establishes a model that can predict the microstructure changes and final mechanical properties at all stages. In fact, the prediction of mechanical properties can not only play the role of organizational performance prediction in actual production, but also play the role of improving product quality and developing new steel grades and new processes. The present invention integrates multiple artificial intelligence algorithms and establishes accuracy evaluation rules to achieve high-precision prediction of the mechanical properties of hot-rolled seamless steel pipes, replacing manual sampling and reducing production costs. like Figure 1 As shown, an embodiment of the present invention provides an integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes. The method includes steps S101-S104.

[0026] S101. Obtain the chemical composition, material parameters and production process parameters of the hot-rolled seamless steel pipe during the rolling process.

[0027] In some embodiments, the chemical composition includes at least one of the following: manganese Mn, chromium Cr, molybdenum Mo, nickel Ni, copper Cu, vanadium V, niobium Nb, titanium Ti, carbon C, and silicon Si.

[0028] In some embodiments, the material parameters include wall thickness and outer diameter.

[0029] In some embodiments, the production process parameters include heating furnace outlet temperature, heating time, piercing mill inlet temperature, rolling mill inlet temperature, sizing mill inlet temperature, and sizing mill outlet temperature.

[0030] In some embodiments, the mechanical property parameters include yield strength, tensile strength, and elongation.

[0031] Exemplarily, the production process data of hot rolled seamless steel pipe is shown in Table 1.

[0032] Table 1

[0033] S102, standardize the chemical composition, material parameters and production process parameters to generate input features.

[0034] As a possible implementation manner, the embodiment of the present invention can perform data cleaning and normalization processing on the collected data to obtain input features.

[0035] Exemplarily, the embodiment of the present invention may delete samples with missing values ​​in the acquired steel pipe production process parameters.

[0036] Exemplarily, the embodiment of the present invention can construct a confidence interval according to the Pauta criterion, mark the data exceeding the confidence interval boundary as outliers and delete them, thereby improving the representativeness of the sample and completing data cleaning.

[0037] The confidence interval can be expressed as ; ; ; in, represents the i-th sample in the data set, represents the mean of the data set, Represents standard deviation.

[0038] For example, the embodiments of the present invention can perform data conversion on the data with high sample representativeness retained after data cleaning according to the standardization method, remove the restrictions on data units and magnitudes, and complete data normalization. The normalization formula can be expressed as ;in, Represents normalized data.

[0039] S103. Based on the input features and various prediction models, output features of various prediction models are obtained.

[0040] In some embodiments, the prediction model includes a BP neural network model, a support vector machine model, and a fast learning network model.

[0041] In some embodiments, the output features include predicted mechanical property parameters of the hot-rolled seamless steel tube, wherein the mechanical property parameters include yield strength, tensile strength and elongation.

[0042] Exemplarily, the embodiments of the present invention can process data of multiple hot-rolled seamless steel pipes with known mechanical property parameters to obtain training samples, and based on the training samples, train the BP neural network, support vector machine (SVM) and fast learning network (FLN) to establish single models for predicting the mechanical properties of steel pipes.

[0043] S104. Determine the mechanical property parameters of the hot-rolled seamless steel pipe based on the output characteristics of various prediction models and the weight coefficients of various prediction models.

[0044] As a possible implementation method, the embodiment of the present invention can determine the mechanical property parameters of the hot-rolled seamless steel pipe based on the following formula.

[0045] ; in, are the mechanical properties parameters of hot rolled seamless steel pipes, is the output feature of the BP neural network model, is the output feature of the support vector machine model, In order to quickly learn the output features of the network model, is the weight coefficient of the BP neural network model, is the weight coefficient of the support vector machine model, To quickly learn the weight coefficients of the network model.

[0046] Exemplarily, the weight coefficients are a1=0.33, a2=0.28, and a3=0.39.

[0047] The present invention provides an integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes. By combining the chemical composition, material parameters and production process parameters of the hot-rolled seamless steel pipes, input features are constructed, and various prediction models such as BP neural network model, support vector machine model and fast learning network model are combined to predict the mechanical property parameters, so as to realize automatic detection of the mechanical properties of hot-rolled seamless steel pipes. The integrated learning of various network models is realized through weight coefficients, and the advantages of various prediction models are combined. Compared with the traditional manual sampling method, the present invention improves the detection efficiency and accuracy of hot-rolled seamless steel pipes.

[0048] The present invention collects the production process data of hot-rolled seamless steel pipes; performs data cleaning and normalization processing on the collected data; uses BP neural network, support vector machine (SVM) and fast learning network (FLN) to respectively establish single models for predicting the mechanical properties of hot-rolled seamless steel pipes; constructs an integrated model for predicting the mechanical properties of hot-rolled seamless steel pipes by adding weight coefficients to the single model through particle swarm algorithm (PSO); creates mechanical property accuracy evaluation indicators to evaluate the prediction results of the integrated model, and realizes high-precision prediction of the mechanical properties of hot-rolled seamless steel pipes. This method uses the idea of ​​integrated learning to construct an integrated model for predicting the mechanical properties of hot-rolled seamless steel pipes by combining multiple weak learning models. Compared with a single model, the integrated model makes up for the limitations of a single data model in the search and learning process, solves the shortcomings of a single model with large prediction deviation and low stability, and provides an important reference for improving the product quality of hot-rolled seamless steel pipes and improving the mechanical properties of steel pipes.

[0049] Optionally, the integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes provided in an embodiment of the present invention further includes steps S201-S205 before step S104.

[0050] S201. Obtain chemical composition, material parameters, and production process parameters of multiple hot-rolled seamless steel pipes with known mechanical property parameters, as well as mechanical property parameters of multiple hot-rolled seamless steel pipes.

[0051] S202. Generate multiple input features based on the chemical composition, material parameters and production process parameters of multiple hot-rolled seamless steel pipes.

[0052] S203. Generate multiple output features based on the mechanical property parameters of multiple hot-rolled seamless steel pipes.

[0053] S204: Generate training samples based on multiple input features and multiple output features.

[0054] S205. Based on the training samples, the preset new neural network models are trained respectively to obtain various prediction models.

[0055] In this way, before the mechanical properties are tested, the embodiment of the present invention obtains various prediction models through neural network training, which facilitates the prediction of the mechanical properties of the hot-rolled seamless steel pipe during the rolling process.

[0056] Optionally, the integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes provided in an embodiment of the present invention further includes steps S301-S305 before step S104.

[0057] S301. Based on the test set in the training sample, various prediction models are tested to obtain test results of various prediction models.

[0058] S302: Based on the test results and the output features in the test set, a particle swarm algorithm is used to determine the weight coefficients of various prediction models.

[0059] As a possible implementation manner, step S302 may be specifically implemented as the following steps.

[0060] Step 1: Generate a particle swarm based on the test results and the output features in the test set. In some embodiments, the particle swarm includes multiple sets of weight coefficients corresponding to each test sample, and the particle swarm parameters are initialized. The particle swarm parameters include the particle population number N, the maximum number of iterations, the individual learning factor c1, the social learning factor c2, the initial position x0 of the randomly generated particles, the initial velocity v0 and the velocity variation range.

[0061] For example, the particle population number N=30, the maximum number of iterations is 30, the individual learning factor c1=2, the social learning factor c2=2, the initial position x0=0 and the initial velocity v0=0 of the randomly generated particles, and the velocity variation range is [-10,10].

[0062] Step 2: Calculate the individual extreme value corresponding to each test sample in the particle swarm, as well as the population extreme value of multiple test samples.

[0063] As a possible implementation method, the embodiment of the present invention can use the root mean square error coefficient RMSE as the fitness function of the particle swarm algorithm to calculate the fitness of each particle, and use the position corresponding to the optimal fitness value of each particle in the iteration process as the individual optimal position pbest, and use the individual position corresponding to the optimal fitness value in the particle swarm iteration process as the population optimal position gbest.

[0064] Exemplarily, step 2 can be specifically implemented as steps A1-A6.

[0065] A1. Calculate the root mean square error of multiple particles corresponding to each test sample.

[0066] A2. Determine the fitness of each particle based on the root mean square error of multiple particles.

[0067] A3. Based on the fitness of each particle, determine the local optimal fitness of each test sample.

[0068] A4. The particle corresponding to the local optimal fitness of each test sample is determined as the individual extreme value.

[0069] A5. Based on the local optimal fitness of each test sample, determine the global optimal fitness of multiple test samples.

[0070] A6. The particle corresponding to the global optimal fitness is determined as the population extreme value.

[0071] Step 3: Based on the individual extreme value and population extreme value of the particle swarm, the velocity and position vector of each particle in the current iteration process are updated.

[0072] Exemplarily, the embodiment of the present invention may be updated based on the following formula.

[0073] ; ; in, is the position of the particle, is the velocity of the particle, A random number between 0 and 1.

[0074] Step 4: If the number of iterations is greater than the maximum number of iterations, the optimal solution is output; if the number of iterations is less than or equal to the maximum number of iterations, the number of iterations is increased by one, and steps 2, 3 and 4 are repeated until the iteration is exited.

[0075] Step 5: Based on the optimal solution, determine the weight coefficients of various prediction models.

[0076] In this way, the embodiment of the present invention can obtain the optimal solution through the particle swarm algorithm, that is, obtain a more accurate weight coefficient for the prediction result, thereby improving the accuracy of the mechanical property detection of the hot-rolled seamless steel pipe.

[0077] Optionally, the integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes provided in an embodiment of the present invention further includes steps S401-S403 after step S104.

[0078] S401. Based on the mechanical property parameters of the hot-rolled seamless steel pipe and the preset mechanical property indicators, determine the evaluation value of each mechanical property parameter.

[0079] In some embodiments, the mechanical performance index is a reliability interval of each mechanical performance parameter.

[0080] For example, the embodiment of the present invention can construct a mechanical performance interval. Set the reliable interval to ,in, represents the allowable error of the predicted value, It is the benchmark value of mechanical performance parameters.

[0081] For example, if the mechanical property parameter is in the range , then the evaluation value of the mechanical performance parameter , if the mechanical properties parameters are not in the range , then the evaluation value of the mechanical performance parameter .

[0082] For example, it is set as follows according to the needs of the production site: =5%× Taking the elongation index, the most representative index among mechanical properties, as an example, some prediction results of the test set are as follows: Figure 2 As shown, pre represents the model prediction value and up represents , low represents .

[0083] S402: Determine the interval coverage of each mechanical property parameter based on the evaluation value of each mechanical property parameter.

[0084] Exemplarily, the embodiment of the present invention may determine the interval coverage based on the following formula.

[0085] ; Where PICP is the interval coverage, n is the number of mechanical property parameters, is the evaluation value of the i-th mechanical property parameter. If the i-th mechanical property parameter is in the interval ,but , if the i-th mechanical property parameter is not in the interval ,but .

[0086] S403. Based on the interval coverage of each mechanical property parameter, the predicted mechanical property parameters are evaluated to determine the evaluation results.

[0087] Exemplarily, the embodiment of the present invention can use the interval coverage rate PICP as the accuracy evaluation index of the model to evaluate the mechanical property prediction results of the model, as shown in Table 2, where the PICP of the yield strength reaches 95%, and the minimum elongation of PICP reaches more than 90%, which fully proves that the method of the present invention has high prediction accuracy.

[0088] Table 2

[0089] In this way, the embodiment of the present invention can calculate and evaluate the predicted mechanical performance indicators after the integrated learning model is predicted, obtain evaluation results, and verify the accuracy of the predicted results of the mechanical properties of the hot-rolled seamless steel pipe.

[0090] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0091] The following is an embodiment of the device of the present invention. For details not described in detail therein, reference may be made to the corresponding method embodiment described above.

[0092] Figure 3 The schematic diagram of the structure of a hot-rolled seamless steel pipe mechanical property prediction device based on integrated learning provided by an embodiment of the present invention is shown. The prediction device 500 includes a communication module 501 and a processing module 502 .

[0093] The communication module 501 is used to obtain the chemical composition, material parameters and production process parameters of the hot-rolled seamless steel pipe during the rolling process.

[0094] The processing module 502 is used to standardize the chemical composition, material parameters and production process parameters to generate input features; based on the input features and various prediction models, the output features of various prediction models are obtained; the prediction models include BP neural network models, support vector machine models and fast learning network models; the output features include predicted mechanical property parameters of hot-rolled seamless steel pipes; based on the output features of various prediction models and the weight coefficients of various prediction models, the mechanical property parameters of the hot-rolled seamless steel pipe are determined.

[0095] Figure 4 Schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 4 As shown, the electronic device 600 includes: a processor 601, a memory 602, and a computer program 603 stored in the memory 602 and executable on the processor 601. When the processor 601 executes the computer program 603, the steps in the above-mentioned method embodiments are implemented, for example Figure 1 Alternatively, when the processor 601 executes the computer program 603, the functions of each module / unit in the above-mentioned device embodiments are realized, for example, Figure 3 The functions of the communication module 501 and the processing module 502 are shown.

[0096] Exemplarily, the computer program 603 may be divided into one or more modules / units, which are stored in the memory 602 and executed by the processor 601 to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program 603 in the electronic device 600. For example, the computer program 603 may be divided into Figure 3 A communication module 501 and a processing module 502 are shown.

[0097] The processor 601 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0098] The memory 602 may be an internal storage unit of the electronic device 600, such as a hard disk or memory of the electronic device 600. The memory 602 may also be an external storage device of the electronic device 600, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 600. Further, the memory 602 may also include both an internal storage unit of the electronic device 600 and an external storage device. The memory 602 is used to store the computer program and other programs and data required by the terminal. The memory 602 may also be used to temporarily store data that has been output or is to be output.

[0099] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those 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 invention, and should all be included in the protection scope of the present invention.

Claims

1. An integrated learning method for predicting the mechanical properties of hot-rolled seamless steel pipes, characterized in that: include: Obtain the chemical composition, material parameters and production process parameters of hot rolled seamless steel pipes during rolling; Standardizing the chemical composition, material parameters and production process parameters to generate input features; Based on the input features and various prediction models, output features of various prediction models are obtained; the prediction models include BP neural network models, support vector machine models and fast learning network models; the output features include predicted mechanical property parameters of the hot-rolled seamless steel pipe; Based on the output characteristics of the various prediction models and the weight coefficients of the various prediction models, the mechanical property parameters of the hot-rolled seamless steel pipe are determined.

2. The method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning according to claim 1, characterized in that: The method of determining the mechanical property parameters of the hot-rolled seamless steel pipe based on the output characteristics of the various prediction models and the weight coefficients of the various prediction models also includes: Obtaining chemical compositions, material parameters and production process parameters of multiple hot-rolled seamless steel pipes with known mechanical property parameters, as well as mechanical property parameters of multiple hot-rolled seamless steel pipes; Generate multiple input features based on chemical compositions, material parameters, and production process parameters of multiple hot-rolled seamless steel pipes; Based on multiple mechanical property parameters of hot-rolled seamless steel pipes, multiple output features are generated; Generate training samples based on the multiple input features and the multiple output features; Based on the training samples, the preset new neural network models are trained respectively to obtain the various prediction models.

3. The method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning according to claim 2, characterized in that: The method of determining the mechanical property parameters of the hot-rolled seamless steel pipe based on the output characteristics of the various prediction models and the weight coefficients of the various prediction models also includes: Based on the test set in the training sample, the various prediction models are tested to obtain test results of the various prediction models; Based on the test results and the output features in the test set, a particle swarm algorithm is used to determine the weight coefficients of the various prediction models.

4. The method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning according to claim 3 is characterized in that: Based on the test results and the output features in the test set, the particle swarm algorithm is used to determine the weight coefficients of the various prediction models, including: Step 1: Based on the test results and the output features in the test set, a particle swarm is generated, wherein the particle swarm includes multiple sets of weight coefficients corresponding to each test sample; and the particle swarm parameters are initialized; the particle swarm parameters include the number of particle populations, the maximum number of iterations, the individual learning factor, the social learning factor, the initial position of randomly generated particles, the initial velocity and the velocity variation range; Step 2: Calculate the individual extreme value corresponding to each test sample in the particle swarm, as well as the population extreme value of multiple test samples; Step 3: Based on the individual extreme value and population extreme value of the particle swarm, the velocity and position vector of each particle in the current iteration process are updated; Step 4: If the number of iterations is greater than the maximum number of iterations, the optimal solution is output; if the number of iterations is less than or equal to the maximum number of iterations, steps 2, 3 and 4 are repeated until the iteration is exited; Step 5: Based on the optimal solution, determine the weight coefficients of various prediction models.

5. The method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning according to claim 4 is characterized in that: The calculating of the individual extreme value corresponding to each test sample in the particle swarm and the population extreme value of the multiple test samples includes: Calculate the root mean square error of multiple particles corresponding to each test sample; Determining the fitness of each particle based on the root mean square error of the plurality of particles; Based on the fitness of each particle, determining the local optimal fitness of each test sample; The particle corresponding to the local optimal fitness of each test sample is determined as the individual extreme value; Determine the global optimal fitness of multiple test samples based on the local optimal fitness of each test sample; The particle corresponding to the global optimal fitness is determined as the population extreme value.

6. The method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning according to claim 1, characterized in that: After determining the mechanical property parameters of the hot-rolled seamless steel pipe based on the output characteristics of the various prediction models and the weight coefficients of the various prediction models, the method further includes: Based on the mechanical property parameters of the hot-rolled seamless steel pipe and the preset mechanical property index, the evaluation value of each mechanical property parameter is determined; the mechanical property index is the reliability interval of each mechanical property parameter; Based on the evaluation values ​​of each mechanical property parameter, the interval coverage of each mechanical property parameter is determined; Based on the interval coverage of each mechanical property parameter, the predicted mechanical property parameters are evaluated and the evaluation results are determined.

7. The method for predicting mechanical properties of hot-rolled seamless steel pipes by integrated learning according to claim 6, characterized in that: Determining the interval coverage of each mechanical property parameter based on the evaluation value of each mechanical property parameter includes: Determine the interval coverage based on the following formula; ; Where PICP is the interval coverage, n is the number of mechanical property parameters, is the evaluation value of the i-th mechanical property parameter. If the i-th mechanical property parameter is in the interval ,but , if the i-th mechanical property parameter is not in the interval ,but ; is the benchmark value of mechanical performance parameters, is the allowable error of the predicted value.

8. The method for predicting mechanical properties of hot-rolled seamless steel pipes according to any one of claims 1 to 7, characterized in that: The chemical composition includes at least one of the following: manganese, chromium, molybdenum, nickel, copper, vanadium, niobium, titanium, carbon, and silicon; the material parameters include wall thickness and outer diameter; the production process parameters include heating furnace outlet temperature, heating time, piercing machine inlet temperature, rolling mill inlet temperature, sizing machine inlet temperature, and sizing machine outlet temperature; the mechanical property parameters include yield strength, tensile strength, and elongation.

9. An integrated learning hot-rolled seamless steel pipe mechanical properties prediction device, characterized in that: The prediction device comprises: Communication module, used to obtain the chemical composition, material parameters and production process parameters of the hot-rolled seamless steel pipe during the rolling process; A processing module is used to standardize the chemical composition, material parameters and production process parameters to generate input features; based on the input features and various prediction models, output features of various prediction models are obtained; the prediction models include BP neural network models, support vector machine models and fast learning network models; the output features include predicted mechanical property parameters of the hot-rolled seamless steel pipe; based on the output features of the various prediction models and the weight coefficients of the various prediction models, the mechanical property parameters of the hot-rolled seamless steel pipe are determined.

10. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 8.

Citation Information

Patent Citations

  • Particle swarm algorithm-based variable weight combination power load short-term prediction method

    CN108038568A

  • A method for predicting crop yield based on multi-source heterogeneous fusion of influencing factors

    CN109635994A

  • Multi-mark C-Mn steel mechanical property prediction method based on improved neural network

    CN113128124A

  • Mechanical property interval prediction method and device based on ABC extreme learning machine

    CN113392594A

  • Rolling force prediction method based on quantum particle swarm optimization and BP neural network

    CN116306220A

Cited By

  • Quality detection method and system for seamless steel tube finished product

    CN120121709A

  • Concrete quality prediction method based on artificial intelligence

    CN120354189A

  • Hot-rolled bar mechanical property prediction method and system based on deep learning

    CN121687320A