Enhanced time series forecasting by means of dynamics boosting for industrial process monitoring

Time series forecasting represents a critical factor, mainly in the industrial sector, in order to assure the proper operation of the manufacturing processes. In this work, a classical ANFIS forecasting scheme is enhanced by the proposal of a dynamics boosting strategy. First, the objective signal is decomposed by means of the Empirical Mode to highlight the main characteristics functions. Next, the dynamics of the functions in regard to the performance of the ANFIS is analyzed. Thus, the functions are separated into different sets. Then, the forecasting is faced with the employment of multiple ANFIS models focused on different dynamics modes. The performance of the proposed system is validated experimentally. According to the obtained results, the proposed approach outperforms the classical methods and represents a reliable and feasible methodology suitable to multiple applications.