Energy anomaly detection in tire curing by using data integration and forecasting techniques

This study proposed a method of energy anomaly detection by using data integration and forecasting techniques to improve energy efficiency in tire curing. Proposed method integrates energy consumption with different factors (environments, equipments, operators, tire blanks and tire types). Artificial neural network model and Support Vector Machine model were used to forecast normal interval for energy efficiency ratio; instances dropping out of this interval indicate potential anomaly affairs. Compared with traditional method, proposed method is robust against environment changes, highly correlated to curing process and can discover curing energy anomalies (leakage of steam or nitrogen, idling, and improper curing parameters configuration) effectively.