Strip steel head thickness difference process parameter optimization method based on principal component analysis controller

Abstract

The invention provides a strip steel head thickness difference process parameter optimization method based on a principal component analysis controller, and belongs to the technical field of steel rolling. The method comprises the following steps: firstly, acquiring a head thickness difference parameter, and selecting first 20% of parameters through a random forest algorithm; then extracting features by using a principal component analysis method, reserving principal components with the characteristic value contribution rate of 85%, and detecting the head thickness difference condition by using T2 and SPE statistics; determining a head thickness difference reason by drawing each parameter contribution diagram; and finally, optimizing process parameters by using a principal component analysis controller, and giving an adjustment amount. According to the method, importance sequencing and feature extraction are carried out on process parameters influencing the head thickness difference ofthe hot-rolled strip steel, then the head thickness difference is detected and diagnosed through a control chart and a contribution chart, and finally the process parameters are optimized through a principal component analysis controller, so that the head thickness difference reason can be quickly analyzed; and process parameters are optimized to enable the head thickness difference to return toa reasonable interval.

Description

technical field [0001] The invention relates to the technical field of steel rolling, in particular to a process parameter optimization method for strip head thickness difference based on principal component analysis controller. Background technique [0002] In the hot rolling production process, the thickness quality of the strip affects the quality of the strip in the subsequent process. The overall thickness accuracy of the strip is mainly restricted by the thickness of the strip head, and the thickness quality of the head is mainly determined by the accuracy of the thickness setting model. Most of the key parameters in the head thickness setting model are assumed or simplified. After the model calculation is completed, an adaptive calculation model will be introduced to further correct the preset model and improve the accuracy of the preset model. [0003] The thickness setting model is established on the basis of mathematical mechanism model and statistical model, and i...

AI Technical Summary

Problems solved by technology

After modeling a large amount of historical production data using the principal component analysis algorithm, it was found that the parameters of the original thickness setting model were inaccurate or the correlation between multiple parameters was inaccurate, resulting in excessive thickness deviation of the head

[0005] There are many existing literatures on improving thickness accuracy, but they mainly use statistical methods to improve the accuracy of the setting model for the problem of inaccurate given single variable, including improving self-learning coefficients such as deformation resistance and temperature, Finish rolling entrance temperature, tension compensation and other methods, and then improve the accuracy of rolling force prediction, in order to achieve the purpose of improving the quality of head thickness

The paper "Causes and Control Countermeasures of Thickness of the First Steel Head of Hot-rolled Non-Oriented Silicon Steel" (Henan Metallurgy, 2019) mainly introduces the adjustment of the pyrometer at the entrance of the finish rolling, and uses the least square method to solve the deformation resistance correction related to temperature The incremental relationship between the three influencing factors of the coefficient and the rolling force and the establishment of self-learning TRD for each steel type, etc., but these methods only consider the effect of a single variable on improving the thickness accuracy

Invention patent "Comprehensive Control Method for Improving Finished Strip Rolling Force and Thickness Accuracy Through Tension Compensation" (Application No. 201310102345.3) is a comprehensive control method for improving finishing strip rolling force and thickness accuracy through tension compensation, but this method only Taking into account the improvement of the setting accuracy of the rolling force model and the roll gap model of the hot-rolled strip under three different tension states, thereby improving the thickness accuracy

The methods in the above-mentioned documents and patents fail to consider the correlation between the variables. The inaccurate setting of the strip head thickness deviation is too large, which can easily lead to the adjustment of one parameter and cause changes in all parameters, which in turn leads to the thickness of the head. The situation with large deviation

Images