PLR-Based Heat Treatment Process State Identification Method

In industrial production, the heat treatment process has a variety of forms, but the most basic process is the thermal process, it includes heating, insulation and cooling three stages. According to the importance of temperature and data characteristics in heat treatment industry, the pretreatment operation of data from dimensionality reduction, interpolation, normalization and feature representation of time series data were studied, and a method of temperature state recognition of heat treatment process based on piecewise linear representation was proposed. In this method, the change point of temperature data sequences was extracted by the deviation degree change between temperature data, and the initial temperature data was divided into trend segments based on this method. Then, the trend segments were merged by the top-down algorithm to obtain the overall trend feature representation of the original data. According to the statistical characteristics of the data, the classifier is trained with RNN characteristics, and the processing state of the heat treatment process is finally obtained. Based on the actual temperature data of a heat treatment unit, experiments were carried out to verify the feasibility and effectiveness of the method.