Operating Performance Assessment of Plant-Wide Processes Based on Hierarchical Multiblock Performance-Relevant Kernel Independent Component Analysis

Abstract

The operating performance assessment of plant-wide processes is very important for improving product quality and comprehensive economic benefits. In this study, a new operating performance assessment method based on hierarchical multiblock performance-relevant kernel independent component analysis (HMPRKICA) is proposed for nonlinear and non-Gaussian plant-wide processes. The proposed approach divides the process variables into several sub-processes according to the physical structure and incorporates sub-process information into high level for operating performance assessment of plant-wide processes. With the new method, the difficulty of process analysis is reduced and the interpretability of the assessment results are improved. Once process operating performance is nonoptimal, the responsible sub-process is first located by the sub-process assessment result, and then the variable contribution rates are calculated in the corresponding sub-process to narrow the search for the nonoptimal causes. Finally, a case study of gold hydrometallurgy process evaluates the efficiency of the proposed method.