KPI-related operating performance assessment based on distributed ImRMR-KOCTA for hot strip mill process

Abstract

With the development of modern industrial processes toward integration and complexity, industrial process operation monitoring is of great significance to ensure the plant safety, product quality, and operating efficiency. However, the inherent nonlinear, dynamic, and plant-wide characteristics make it difficult to evaluate the operating performance accurately. To handle this issue, a key performance indicator-related operating performance assessment method based on distributed improved minimal redundancy maximal relevance and kernel output-relevant common trend analysis (ImRMR-KOCTA) is proposed in this paper. First, by replacing mutual information with maximal information coefficient, the minimal redundancy maximal relevance is improved to describe the interdependencies between process variables and key performance indicators, and the correlated variables are retained in each subsystem. Second, based on kernel functions and outputrelevant common trend analysis, the assessment model is developed for describing the nonlinearity and dynamicity in each subsystem. Then, operating performance level is determined by Bayesian inference and predefined rules. Finally, a validation on a hot strip mill process is given to verify the effectiveness of the proposed method.