State Recognition Method of Heat Treatment Process Based on PLR

Abstract

Temperature can change the chemical composition and internal structure of metals during heat treatment processing, which is the decisive factor affecting the quality of processing devices. Aiming at the trend change of temperature in the process of heat treatment, a state recognition method based on piecewise linear representation of time series is proposed. The method first extracts the turning point according to the angle change of the time series temperature data, then divides the trend segment of the original data by turning point, then uses bottom-Up algorithm to merge the trend segment to obtain the segmented linear representation, trains the SVM classification by the statistical characteristics of the segmented data, and finally obtains the processing state of the heat treatment process. Taking the actual data of a unit heat treatment workshop site as an example, the effectiveness of this method is verified by experimenting with the proposed method.