Data-driven method for monitoring and diagnosis in energy system of papermaking process

This study proposes a data-driven model for monitoring and diagnosis in energy system of papermaking process based on kernel component analysis (KPCA) and the kernel slow feature analysis (KSFA). Four different abnormal patterns are designed, while the false alarm rate (FAR) and the missed detection rate (MDR) are used to evaluate the validity of the proposed model. The performance of KPCA is better than that of conventional PCA and KSFA. The online monitoring and diagnosis analysis of energy utilization are achieved based on the historical data and the proposed model. The production status and the energy consumption level across the whole process can be acquired. The results demonstrate that the proposed method is effective. It can provide the valuable reference foundation for further energy analysis and optimization in papermaking process.