A method and system for online parallel hot-rolled strip parameter optimization

Abstract

This invention provides an online parallel parameter optimization method and system for hot-rolled strip. By adding a parameter optimization terminal to the existing model architecture and utilizing existing production data and specially designed online parallel parameter optimization software, real-time production data (L1, L2, L3) can be acquired without affecting existing production. The software enables online parallel data output, and through comparative analysis with the online production data using a specific optimization algorithm, parameter optimization experiments for any function of interest can be conducted, reducing the risk of parameter modification in the secondary model of hot-rolled strip to near zero. Furthermore, the program modification is minimal, and the parameter optimization accuracy is higher. This method is used to optimize the parameters of existing hot-rolled strip secondary models, and is particularly suitable for improving the setting accuracy of secondary models without affecting on-site hot-rolling production.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for hot rolling, and in particular to an online parallel hot-rolled strip parameter optimization method and system. Background Technology

[0002] In my country, most suppliers of secondary molds for hot rolling mills are foreign companies. Their technology not only includes American components but also contains numerous "black boxes" (unreliable, often opaque, technical aspects). This has resulted in long-term stagnation in the capacity improvement of secondary mold production lines for most hot rolling mills. In some cases, a distorted technological advancement approach has emerged, where new production lines are invested in instead of existing ones, leading domestic steel companies into a vicious cycle of increasing reliance on imported technology. How to effectively utilize existing secondary molds is a problem that every steel company must face. Hot-rolled strip secondary molds operate 24 / 7 after being put into use. Currently, the mainstream methods for optimizing the parameters of hot-rolled strip secondary molds include the following:

[0003] Relying entirely on foreign parties for parameter optimization work leads to problems such as time constraints due to the provider of the secondary model, large project funding requirements, and the need to re-hire external units for optimization after each product upgrade. In particular, the funds required for modification and upgrading of secondary models provided by foreign companies are even greater.

[0004] We obtain optimized parameter schemes for other similar production lines through technical exchanges. However, these schemes only provide the "how" but not the "why". Due to the wide specifications of hot rolling production lines and the vast differences in process configurations among various production lines, the obtained parameters are mostly not fully applicable to the corresponding production lines. The secondary model program is extremely complex, making it difficult to judge the chain reaction after parameter optimization. This can lead to risks such as steel accumulation and strip breakage. The economic loss from a single such risk is no less than 500,000 yuan.

[0005] Using offline simulation platforms or digital twin systems for parameter optimization simulation analysis and guidance for on-site production, such as the invention patent (a method for simulating the secondary process of hot strip rolling), involves using an Oracle database to modify the dynamic data of the strip and using the dynamic data stored in a CSV file for simulation. The disadvantages of this method are that it requires a high level of understanding of the program, necessitates the replacement of the middleware part of the original program, is technically difficult, and the simulated data is somewhat inaccurate compared with the on-site production data, making it difficult to fully reflect the real situation on-site, and resulting in simulation analysis errors. Summary of the Invention

[0006] To address the technical problems raised in the background section, this invention provides an online parallel parameter optimization method and system for hot-rolled strip. By adding a parameter optimization terminal to the existing model architecture and utilizing existing production data and specially designed online parallel parameter optimization software, real-time production data (L1, L2, and L3) can be acquired without affecting existing production. The software enables online parallel data output, and through comparative analysis with the online production data using a specific optimization algorithm, parameter optimization experiments for any desired function can be conducted, reducing the risk of parameter modification in the secondary model of hot-rolled strip to near zero. Furthermore, the program modification is minimal, and the parameter optimization accuracy is higher.

[0007] To achieve the above objectives, the present invention employs the following technical solution:

[0008] An online parallel parameter optimization method for hot-rolled strip includes the following steps:

[0009] Step 1: Determine the functional modules to be optimized in the L2 level two model control system based on the operation status of the L3 level production management system;

[0010] Step 2: Review the optimization parameters and programs of the modules of the L2 level two model control system to be optimized. The review refers to identifying which parameters and programs in the level two model affect the model calculation results and which parameters and programs can be adjusted through the analysis of the level two model, including the configuration parameters in the configuration file.

[0011] Step 3: Review the inputs and outputs of the existing two-level model control system. This review involves analyzing the main program of the model that needs to be optimized, identifying the main program calculation model, the configuration files that need to be read, and checking the input and output variables.

[0012] Step 4: Establish an online parallel parameter optimization terminal. The main software model of this online parallel parameter optimization terminal is consistent with the secondary model of the existing L2 level secondary model control system.

[0013] The software model of the online parallel parameter optimization terminal is modified by programming on the basis of the existing secondary model. Specifically, it is changed from single input and single output to multiple inputs of different parameters, parallel calculation of multiple results, and simultaneous output of multiple results to different file directories.

[0014] The software model of the online parallel parameter optimization terminal only accepts data from L1 level basic automation and L2 level secondary model control system. The output data is not sent to L1 level basic automation, does not participate in the control of L2 level basic automation, and is not uploaded to L3 level production management system.

[0015] Step 5: Transfer the data from the existing L1 level basic automation and L2 level secondary model control systems to the main body of the software model of the parallel optimization parameter terminal via the data bus;

[0016] Step 6: In daily production, simultaneously start the parallel parameter optimization terminal. Execute various optimization schemes to be verified on the secondary model parameters to be optimized in the parallel parameter optimization terminal, modify the program and parameters, and complete the calculation work; specifically including the following:

[0017] X optimization schemes are designed for the parameters, and the modified X parameter optimization schemes are placed in X different configuration directory files. Then, the modified multi-input parallel computing model to be optimized reads online data and runs in the software model of the online parallel parameter optimization terminal. After completing n rolling plans, the X different calculation results are output.

[0018] Step 7: After the online parallel parameter optimization terminal has been running for a set time, obtain the optimized parameter calculation results, compare them with the existing secondary model calculation results of the comparison sample, and optimize them through the optimization algorithm. Repeat the optimization multiple times to form the final optimization scheme.

[0019] Step 8: Run the final optimized scheme on the online parallel parameter optimization terminal for a set time. After the results are completely correct, gradually transfer the calculation results of the online parallel parameter optimization terminal to the secondary model of the existing L2 level secondary model control system in a weighted manner using λ-weighting, and finally complete the parameter optimization work.

[0020] Furthermore, in step 6, x kinds of technical solutions to be optimized are designed through regularization experiments and other methods. The impact of the x kinds of optimization on the existing secondary model is evaluated to form a sample library for use by optimization algorithms such as machine learning.

[0021] Furthermore, in step 7, the sample library formed in step 6 is optimized using machine learning and other optimization algorithms to obtain the optimal solution that minimizes the error with the measured results.

[0022] Furthermore, step 8 also includes: gradually increasing the weight of λ-weighting and tracking the stable operation of the existing system in real time until λ-weighting reaches 1, that is, completely replacing the optimized parameter scheme into the secondary model of the existing L2 level secondary model control system.

[0023] The present invention also provides a system for the online parallel hot-rolled strip parameter optimization method, comprising: an L1 level basic automation device, an L2 level secondary model control system, and an L3 level production management system; and an online parallel parameter optimization terminal; wherein the online parallel parameter optimization terminal uses a PC as the hardware carrier and is connected in parallel with the L1 level basic automation and L2 level secondary model control systems via a data bus.

[0024] The online parallel parameter optimization terminal contains a software model of the online parallel hot-rolled strip parameter optimization method. It accepts data from L1 level basic automation and L2 level secondary model control systems. It does not send the output data to L1 level basic automation, does not participate in the control of L2 level basic automation, and does not upload it to L3 level production management system.

[0025] Compared with the prior art, the beneficial effects of the present invention are:

[0026] 1) The present invention provides an online parallel hot-rolled strip parameter optimization method and system, which is used to optimize the setting parameters of the existing hot-rolled strip secondary model. It is particularly suitable for improving the setting accuracy of the secondary model without affecting the hot rolling on-site production, especially for secondary modules such as hot rolling secondary thermal expansion and wear that are greatly affected by the setting parameters.

[0027] 2) The platform can be built by technical personnel who have the ability to perform daily maintenance of the secondary model and data communication, and who require minimal program modifications.

[0028] 3) The model parameter optimization risk is low. Since the online parallel parameter optimization terminal adopts a scheme that only allows data to be fed in, there is no interference with the existing model. Therefore, it will not have any adverse effects on the existing model.

[0029] 4) The operation of the online parallel parameter optimization terminal can also deepen the system maintenance personnel's understanding of the program. After a deeper understanding, the impact of the output of the online parallel parameter optimization terminal on the existing model can be gradually released in a weighted manner, and more system modules, such as self-learning modules, can be optimized.

[0030] 5) High parameter optimization accuracy: Since the model of the online parallel parameter optimization terminal is completely consistent with the existing model, high accuracy of parameter optimization can be guaranteed.

[0031] 6) Parameter optimization of multiple models can be carried out simultaneously, and the related impacts can be analyzed. In the weighted transfer stage after the parameter optimization scheme of one model is perfected, parameter optimization of other models can be carried out, and even parameter optimization of multiple mutually influential models can be carried out simultaneously. Attached Figure Description

[0032] Figure 1 This is a flowchart of the online parallel hot-rolled strip parameter optimization method of the present invention;

[0033] Figure 2 This is a diagram of the online parallel hot-rolled strip parameter optimization system of the present invention;

[0034] Figure 3 The diagram shows the optimization results of different parameters in an embodiment of the present invention. Detailed Implementation

[0035] The specific embodiments provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0036] like Figure 1 As shown, the present invention proposes an online parallel hot-rolled strip parameter optimization method, which includes the following steps:

[0037] Step 1: Determine the functional modules to be optimized in the L2 level two model control system based on the operation status of the L3 level production management system.

[0038] Step 2: Review the optimization parameters of the modules of the L2 level two-stage model control system to be optimized. This review refers to identifying which parameters in the level two model affect the model calculation results and which parameters can be adjusted through analysis of the level two model. Generally, these are the configuration parameters in the configuration file.

[0039] For example, many models for hot-rolled plates are open-loop, such as thermal expansion models and wear models, and cannot form a closed loop. The actual values ​​of their calculation results need to be obtained through indirect measurement, and then compared with the model calculation results for correction.

[0040] Step 3: Review the inputs and outputs of the existing two-level model control system. This review involves analyzing the main program of the model that needs optimization, identifying the main program calculation model, the configuration files that need to be read, and checking the input and output variables.

[0041] Step 4: Establish an online parallel parameter optimization terminal. The main software model of this online parallel parameter optimization terminal is consistent with the secondary model of the existing L2 level secondary model control system.

[0042] The software model of the online parallel parameter optimization terminal is modified by programming based on the existing secondary model. Specifically, it is changed from single input and single output to multiple inputs of different parameters, parallel calculation of multiple results, and simultaneous output of multiple results to different file directories.

[0043] This refers to the model calculation code in the secondary program that needs optimization. For example, the calculation module of the thermal expansion model, except for the output, has the same source code as the code running on site.

[0044] The software model of the online parallel parameter optimization terminal only accepts data from L1 level basic automation and L2 level secondary model control systems. The output data is not sent to L1 level basic automation, does not participate in the control of L2 level basic automation, and is not uploaded to L3 level production management system.

[0045] Step 5: Transfer the data from the existing L1 level basic automation and L2 level secondary model control systems to the main body of the software model of the parallel optimization parameter terminal via the data bus;

[0046] Step 6: In daily production, simultaneously start the parallel parameter optimization terminal. Modify the program and parameters of various optimization schemes for the secondary model parameters to be optimized in the parallel parameter optimization terminal, and complete the calculation work; specifically including the following:

[0047] X optimization schemes are designed for the parameters, and the modified X parameter optimization schemes are placed in X different configuration directory files. Then, the modified multi-input parallel computing model to be optimized reads online data and runs in the software model of the online parallel parameter optimization terminal. After completing n rolling plans, the X different calculation results are output.

[0048] Step 7: After the online parallel parameter optimization terminal has been running for a set time (usually 2-4 weeks, with a maximum of 8 weeks), obtain the optimized parameter calculation results, compare them with the calculation results of the existing secondary model of the comparison sample, and optimize them through the optimization algorithm. Repeat the optimization multiple times to form the final optimization scheme.

[0049] Step 8: Run the final optimized scheme on the online parallel parameter optimization terminal for a set time. After the results are completely correct, gradually transfer the calculation results of the online parallel parameter optimization terminal to the secondary model of the existing L2 level secondary model control system in a weighted manner using λ-weighting, and finally complete the parameter optimization work.

[0050] Furthermore, in step 6, x kinds of technical solutions to be optimized are designed through regularization experiments and other methods. The impact of the x kinds of optimization on the existing secondary model is evaluated to form a sample library for use by optimization algorithms such as machine learning.

[0051] Furthermore, in step 7, the sample library formed in step 6 is optimized using machine learning and other optimization algorithms to obtain the optimal solution that minimizes the error with the measured results.

[0052] Furthermore, step 8 also includes: gradually increasing the weight of λ-weighting and tracking the stable operation of the existing system in real time until λ-weighting reaches 1, that is, completely replacing the optimized parameter scheme into the secondary model of the existing L2 level secondary model control system.

[0053] like Figure 2 As shown, the present invention also provides a system for the online parallel hot-rolled strip parameter optimization method, comprising: an L1 level basic automation device, an L2 level secondary model control system, and an L3 level production management system; and an online parallel parameter optimization terminal; wherein the online parallel parameter optimization terminal uses a PC as the hardware carrier and is connected in parallel with the L1 level basic automation and L2 level secondary model control systems via a data bus.

[0054] The online parallel parameter optimization terminal contains a software model of the online parallel hot-rolled strip parameter optimization method. It accepts data from L1 level basic automation and L2 level secondary model control systems. It does not send the output data to L1 level basic automation, does not participate in the control of L2 level basic automation, and does not upload it to L3 level production management system. Specific implementation examples:

[0056] The following example illustrates the parameter optimization method and system for a secondary thermal expansion model of a hot-rolled strip production line:

[0057] I. Adding new hardware, see reference Figure 2 As shown, one additional PC and one switch are required (depending on the network communication method on site).

[0058] II. New Software (Refer to) Figure 1 As shown, the existing secondary model and all operating environments need to be copied to the new PC, and the communication interface between the secondary model and L1 needs to be sorted out. A new parallel parameter interface needs to be created to import the relevant parameters of L1 and L2 into the secondary model of the new PC in real time, forming the model of the online parallel parameter optimization terminal.

[0059] Third, for the parameters that need optimization in the thermal expansion model of the existing secondary model, such as the roll thermal conductivity multiplier, the strip-to-work roll heat transfer multiplier, the roll core thermal conductivity, and the roll shell thermal conductivity, x optimization schemes are designed. The modified x parameter optimization schemes are placed in x different configuration directory files. Then, the modified multi-input parallel calculation is used to optimize the model, read online data, and run the model in the online parallel parameter optimization terminal. After completing n rolling schedules, x different calculation results are output. The output of the model in the online parallel parameter optimization terminal is compared with the output of the existing secondary model.

[0060] Figure 3The optimization results for four different parameters were shown. Four different optimization schemes were set at once, and the calculations for the four different schemes were completed simultaneously. Finally, the results were compared with the measured results to select the optimal technical optimization scheme.

[0061] Fourth, through optimization schemes such as machine learning, the impact of x optimizations on the existing model is evaluated to find the optimal solution.

[0062] Fifth, place the optimized scheme parameters into the model of the online parallel parameter optimization terminal, and output the output of the online parallel parameter optimization terminal to the existing secondary system in a λ-weighted manner.

[0063] 6. Gradually increase the weight of λ and track the stable operation of the existing system in real time until λ is weighted to 1, at which point the optimized parameter scheme can be completely replaced in the existing secondary model.

[0064] The results show that the online parallel hot-rolled strip parameter optimization method and system of the present invention is used to optimize the setting parameters of the existing hot-rolled strip secondary model. It is particularly suitable for improving the setting accuracy of the secondary model without affecting the hot rolling on-site production, especially for secondary modules such as hot rolling secondary thermal expansion and wear, which are greatly affected by the setting parameters.

[0065] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0066] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0067] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0068] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0069] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0070] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An online parallel hot-rolled strip parameter optimization method, characterized in that, Includes the following steps: Step 1: Determine the functional modules to be optimized in the L2 level two model control system based on the operation status of the L3 level production management system; Step 2: Review the optimization parameters and programs of the modules of the L2 level two model control system to be optimized. The review refers to identifying which parameters and programs in the level two model affect the model calculation results and which parameters and programs can be adjusted through the analysis of the level two model, including the configuration parameters in the configuration file. Step 3: Review the inputs and outputs of the existing two-level model control system. This review involves analyzing the main program of the model that needs to be optimized, identifying the main program calculation model, the configuration files that need to be read, and checking the input and output variables. Step 4: Establish an online parallel parameter optimization terminal. The main software model of this online parallel parameter optimization terminal is consistent with the secondary model of the existing L2 level secondary model control system. The software model of the online parallel parameter optimization terminal is modified by programming on the basis of the existing secondary model. Specifically, it is changed from single input and single output to multiple inputs of different parameters, parallel calculation of multiple results, and simultaneous output of multiple results to different file directories. The software model of the online parallel parameter optimization terminal only accepts data from L1 level basic automation and L2 level secondary model control system. The output data is not sent to L1 level basic automation, does not participate in the control of L2 level basic automation, and is not uploaded to L3 level production management system. Step 5: Transfer the data from the existing L1-level basic automation and L2-level secondary model control systems to the main body of the software model in the online parallel optimization parameter terminal via the data bus; Step 6: In daily production, simultaneously start the parallel parameter optimization terminal. Execute various optimization schemes to be verified on the secondary model parameters and program to be optimized in the parallel parameter optimization terminal, modify the program and parameters, and complete the calculation work; specifically including the following: X optimization schemes are designed for the parameters and program, and the modified X parameter and program optimization schemes are placed in X different configuration directory files. Then, the modified multi-input parallel computing model to be optimized reads online data and runs in the software model of the online parallel parameter optimization terminal. After completing n rolling plans, the X different calculation results are output. Step 7: After the online parallel parameter optimization terminal has run for a set time, obtain the calculation results of the optimized parameters, compare them with the calculation results of the existing secondary model of the comparison sample, and optimize them through the optimization algorithm. Repeat the optimization multiple times to form the final optimization scheme. Step 8: Run the final optimized scheme on the online parallel parameter optimization terminal for a set time. After the results are completely correct, gradually transfer the calculation results of the online parallel parameter optimization terminal to the secondary model of the existing L2 level secondary model control system in a weighted manner using λ-weighting, and finally complete the parameter optimization work.

2. The online parallel hot-rolled strip parameter optimization method according to claim 1, characterized in that, In step 6, x optimization schemes, including regularized testing, are used to evaluate the impact of these x optimization schemes on the existing secondary model, thereby forming a sample library for optimization algorithms, including machine learning.

3. The online parallel hot-rolled strip parameter optimization method according to claim 2, characterized in that, In step 7, the sample library formed in step 6 is optimized using optimization algorithms, including machine learning, to obtain the optimal solution that minimizes the error with the measured results.

4. The online parallel hot-rolled strip parameter optimization method according to claim 1, characterized in that, Step 8 further includes: gradually increasing the weight of λ-weighting and tracking the stable operation of the existing system in real time until λ-weighting reaches 1, that is, completely replacing the optimized parameter scheme into the secondary model of the existing L2 level secondary model control system.

5. A system for the online parallel hot-rolled strip parameter optimization method according to claim 1, comprising: The system comprises an L1-level basic automation device, an L2-level secondary model control system, and an L3-level production management system; characterized in that it further includes an online parallel parameter optimization terminal; the online parallel parameter optimization terminal uses a PC as its hardware carrier and is connected in parallel with the L1-level basic automation and L2-level secondary model control systems via a data bus. The online parallel parameter optimization terminal contains a software model of the online parallel hot-rolled strip parameter optimization method. It accepts data from L1 level basic automation and L2 level secondary model control systems. It does not send the output data to L1 level basic automation, does not participate in the control of L2 level basic automation, and does not upload it to L3 level production management system.

Images