Decision tree-based method for extracting key characteristic variables of finish rolling temperature control process

Abstract

The invention relates to a decision tree-based method for extracting key characteristic variables of a finish rolling temperature control process. The method comprises the following steps: finish rolling process data are pretreated, process data taking band steel numbers as indexes are converted into decision attribute vectors taking finish rolling temperature as a target attribute, and nonlinear discretization is performed on the finish rolling temperature; a decision tree method is used for calculating the information gain rate of each discrete or continuous value decision attribute; and the effect of each attribute on the finish rolling temperature is determined according to the information gain rate, the decision attribute vectors are reordered according to the values of information gain rates, and key characteristic variables with decisive effect on the finish rolling temperature are extracted in accordance with the finish rolling process mechanism and model precision requirements. The method can determine which key variables have a decisive effect on the finish rolling temperature according to the actual field process data, thereby laying a foundation for establishing a finish rolling temperature forecast model and a finish rolling temperature correction model.

Description

technical field [0001] The invention relates to a method for extracting key characteristic variables in the process of establishing a temperature model of hot continuous rolling and finish rolling of strip steel, in particular to a method for extracting key characteristic variables in the temperature control process of finish rolling based on a decision tree. Background technique [0002] In the production of hot-rolled strip steel, the control accuracy of the finish rolling temperature of the finishing rolling mill has a direct impact on the microstructure and properties of the final product. The temperature control of the full-length final rolling of the finished strip has always been an important research topic in hot rolling production, and it is also one of the difficulties. First, there are many process variables that affect the finish rolling temperature; second, there are few effective temperature measurement points in the finish rolling area, and only two points at ...

AI Technical Summary

Problems solved by technology

First, there are many process variables that affect the finish rolling temperature; second, there are few effective temperature measurement points in the finish rolling area, and only two points at the exit of rough rolling and the exit of finish rolling are the most reliable for temperature measurement; third, the original The accuracy of the temperature-related process model is limited, and it cannot meet the requirements of real-time control; Fourth, the common means used to control the temperature of the full-length finish rolling include controlling the running acceleration of the strip and the water volume and water pressure of spraying between stands. The influence of finish rolling temperature has a large hysteresis, and it is difficult to achieve effective control with conventional model-free control methods

Images