An Adaptive Weighted Envelope Based Piecewise Fitting Method for Molding Process Data

Data preprocessing and fitting are important and challenging problems, which are used in system control and modeling in industrial processes nowadays. In the paper, a piecewise fitting method for molding process data based on the adaptive weighted envelope is proposed. Due to the characteristics of the multi-period molding process, the data is first segmented by different application conditions in the molding process. Next, the upper and lower envelopes of different data segments are calculated, the adaptive weights of which are designed to synthesize smooth curves by adding a sliding window. Then, two ways of polynomial fitting and Gaussian fitting are adopted to fit the data according to the characteristics of different subsection data. Finally, compared with the moving average filtering algorithm, the proposed method has a smaller absolute percentage error (APE) and lower root mean square error (RMSE) value of the fitting curve, which demonstrates the superiority of the proposed method.