Fault Diagnosis of Chemical Processes Based on k-NN Distance Contribution Analysis Method

Abstract

In modern chemical processes, varieties of fault detection and diagnosis methods have been used for ensuring process safety and product quality widely. As an important branch, fault detection and diagnosis methods based on data-driven are effective in large-scale chemical processes. However, they do not often show superior performance owing to the self-limitations and the characteristics of process data, such as nonlinearity, non-Gaussian, and multi-operating mode. To cope with these issues, k-NN (k-Nearest Neighbor) fault detection method and its extension have been developed in recent years. Nevertheless, these methods are used for fault detection mainly, few papers can be found about fault diagnosis. In this paper, a novel abnormal variables identification method is proposed, this method uses k-NN distance contribution analysis theory to evaluate which variables are most likely to be abnormal, meanwhile, the feasibility of this method is verified by contribution decomposition theory. The proposed search strategy can guarantee that all abnormal variables are found in each sample. The reliability and validity of the proposed method are verified by a numerical example and the Continuous Stirred Tank Reactor system.