CN-116305885-B - Method, device and storage medium for predicting or controlling iron loss of silicon steel

CN116305885BCN 116305885 BCN116305885 BCN 116305885BCN-116305885-B

Abstract

The invention provides a method, a device and a storage medium for predicting or controlling silicon steel core loss, wherein the predicting method comprises the following steps of S1, determining a technological parameter variable set for predicting the silicon steel core loss in the whole process of silicon steel coil production, specifically, determining the technological parameter variable set by adopting PLS in combination with recursive variable elimination, S2, training a neural network based on a selected variable set and an obtained historical data set to obtain a silicon steel core loss predicting model, S3, obtaining an actual technological parameter value of a current finished process and a historical median of technological parameters of a follow-up unfinished process in the silicon steel production process based on the selected variable set, and S4, inputting data obtained in the step S3 into the silicon steel core loss predicting model to conduct core loss prediction. By utilizing the technical scheme, the on-line forecasting of the iron loss of the silicon steel can be more accurately realized.

Inventors

CAI QUANFU
HE LIHONG
YAO WENDA
WANG ZHIJUN

Assignees

中冶南方工程技术有限公司

Dates

Publication Date: 20260505
Application Date: 20230303

Claims (8)

1. A method for predicting iron loss of silicon steel during the production of silicon steel, comprising: s1, determining a technological parameter variable set for predicting the iron loss of the silicon steel coil in the whole production process of the silicon steel coil, wherein the method comprises the following steps: S11, selecting the whole-flow process parameter variable of the steel coil as an initial variable set, obtaining a corresponding whole-flow process parameter variable value and a corresponding iron loss value of each finished steel coil, and establishing a historical data set by utilizing the obtained whole-flow process parameter variable value and the iron loss value; S12, establishing a regression model by adopting a partial least squares PLS method based on a current variable set, and calculating root mean square error of the regression model under the current variable set through cross verification; s13, removing the variable with the minimum absolute value of the regression coefficient in the variable set formed by the whole process technological parameter variables according to the regression model; S14, judging whether the number of the residual variables is equal to the preset minimum variable number, if so, turning to a step S15, otherwise, returning to the step S12; S15, comparing root mean square errors obtained by the regression model through cross verification under different variable numbers, and selecting a variable set corresponding to the minimum root mean square error as a process parameter variable set for predicting the iron loss of the silicon steel; s2, training a neural network based on the selected variable set for predicting the iron loss of the silicon steel and the historical data set to obtain a prediction model of the iron loss of the silicon steel; s3, obtaining an actual value of a technological parameter of a current finished process and a historical median of a technological parameter of a follow-up unfinished process in the silicon steel production process based on the selected variable set for predicting the iron loss of the silicon steel; S4, inputting the obtained actual values of the technological parameters and the historic median into a prediction model of the iron loss of the silicon steel to perform iron loss prediction.
2. The method of claim 1, wherein the neural network is a three-layer BP neural network.
3. The method of claim 1, wherein the full-process parameters include at least chemical content in the steelmaking process, heating temperature and thickness in the hot rolling process, tension and temperature in the normalizing pickling line process, thickness in the rolling mill process, tension in the continuous annealing line, temperature and current.
4.A method for controlling iron loss of silicon steel during the production of silicon steel, comprising: Predicting the iron loss of a silicon steel during the production of the silicon steel by using the method of any one of claims 1 to 3; And optimizing the technological parameters of the follow-up unfinished working procedures according to the predicted iron loss of the silicon steel.
5. The method of claim 4, wherein optimizing the process parameters of the subsequent unfinished process based on the predicted iron loss of silicon steel comprises: and optimizing the technological parameters of the follow-up unfinished working procedure according to the actual values of the technological parameters of the current finished working procedure and the upper and lower limits of the technological parameters of the follow-up unfinished working procedure by adopting a particle swarm optimization algorithm.
6. An apparatus for predicting iron loss in silicon steel, comprising a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the method of any one of claims 1 to 3.
7. An apparatus for controlling iron loss in silicon steel, comprising a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the method of any one of claims 4 to 5.
8. A computer readable storage medium, characterized in that at least one program is stored in the storage medium, the at least one program being executed by a processor to implement the method of any one of claims 1 to 5.

Description

Method, device and storage medium for predicting or controlling iron loss of silicon steel Technical Field The invention relates to the field of silicon steel production control, in particular to a method and a device for predicting or controlling iron loss of silicon steel and a storage medium. Background Silicon steel has the characteristics of high magnetic permeability, low coercive force, large resistivity and the like, and is mainly used as a magnetic material in motors, transformers, electric appliances and electrical instruments. Iron core consumption, abbreviated as iron loss, is the most important quality index of silicon steel, and directly determines the usability of cold rolled silicon steel. The low iron loss of the silicon steel can save a large amount of electric energy, prolong the working and running time of the motor and the transformer and simplify a cooling system. Therefore, under the condition of ensuring safe operation of production, the iron loss should be reduced as much as possible, thereby realizing the optimized operation of the whole silicon steel production process. The silicon steel production process is complex, and the influence factors of the iron loss of the silicon steel are numerous, wherein the influence factors comprise chemical components such as C, si, mn, P, S, al, N and the like, hot rolling parameters such as finish rolling inlet temperature, finish rolling outlet temperature, coiling temperature, hot rolling thickness and the like, annealing parameters, furnace section temperatures, cooling section parameters, furnace atmosphere parameters, drying furnace temperature parameters and the like. At present, researches on the iron loss of the silicon steel are mainly focused on technological mechanisms, the researches are qualitative analyses, the researched technological parameters are fewer, the researches on the prediction of the iron loss of the silicon steel by utilizing a statistical modeling method are fewer, a mechanism model or a data model of the whole process technological parameters and the iron loss of the silicon steel production is not established at present, and a related optimization control model for controlling the iron loss in production is not established. Disclosure of Invention The embodiment of the invention provides a method, a device and a storage medium for predicting or controlling iron loss of silicon steel, so as to perform online prediction or control optimization on the iron loss in the silicon steel production process. To achieve the above object, in one aspect, there is provided a method for predicting iron loss of silicon steel in a silicon steel production process, the method comprising: s1, determining a technological parameter variable set for predicting the iron loss of the silicon steel coil in the whole production process of the silicon steel coil, wherein the method comprises the following steps: S11, selecting the whole-flow process parameter variable of the steel coil as an initial variable set, obtaining a corresponding whole-flow process parameter variable value and a corresponding iron loss value of each finished steel coil, and establishing a historical data set by utilizing the obtained whole-flow process parameter variable value and the iron loss value; S12, establishing a regression model by adopting a partial least squares PLS method based on the current variable set, and calculating the root mean square error of the regression model under the current variable set through cross verification; s13, aiming at the regression model, removing the variable with the minimum absolute value of the regression coefficient in the variable set formed by the process parameter variables of the whole process; S14, judging whether the number of the residual variables is equal to the preset minimum variable number, if so, turning to a step S15, otherwise, returning to the step S12; S15, comparing root mean square errors obtained by the regression model through cross verification under different variable numbers, and selecting a variable set corresponding to the minimum root mean square error as a process parameter variable set for predicting the iron loss of the silicon steel; S2, training a neural network based on the selected variable set and the historical data set for predicting the iron loss of the silicon steel to obtain a prediction model of the iron loss of the silicon steel; s3, obtaining an actual value of a technological parameter of a current finished process and a historical median of a technological parameter of a follow-up unfinished process in the silicon steel production process based on the selected variable set for predicting the iron loss of the silicon steel; S4, inputting the obtained actual values of the process parameters and the historic median into a silicon steel iron loss prediction model to perform iron loss prediction. Preferably, the method, wherein the neural network is a three-layer BP neural network.