Operating Performance Assessment for Dynamic Industrial Processes Based on CVA-TPLS

For industrial process, the operating performance assessment is very crucial in the aspect of improvement of process performance and comprehensive economic benefits (CEB) of enterprises actually. The static information of process data is investigated properly, while the data with dynamic characteristics are seldom exposed in traditional operating performance assessment methods. Hence, a novel operating performance assessment method combining canonical variate analysis and total projection to latent structure (CVA-TPLS) is proposed here for the industrial processes with dynamic characteristics. Canonical variate analysis (CVA) can extract the temporally strongly correlated variations from the process data, and total projection to latent structure (TPLS) can explore the performance-relevant features to reflect the process operating performance. Through combining CVA and TPLS, the proposed CVA-TPLS makes full use of variations with both dynamic and performance-relevant characteristics to assess the process operating performance. At last, the proposed method is applied to a numerical instance and the gold hydrometallurgrical industrial process, and the efficiency and feasibility is demonstrated.