Sparse representation and modified density peak clustering-based state identification for multimode processes

Abstract

Abstract Industrial processes with high-dimensional data are generally operated with mixed normal/faulty states in different modes, making it difficult to automatically and accurately identify the faults. In this paper, a state identification framework is proposed for multimode processes. First, a key variable selection approach is presented based on sparse representation to eliminate redundant variables. Then, modified density peak clustering is proposed to identify different states, in which a distance measurement with a time factor is constructed to select all the possible cluster centers. Then, the sum of squared errors-based approach is developed to determine the optimal cluster centers automatically. Further, considering that the mode attributes may be mixed with the fault attributes, a two-step ‘coarse-to-fine identification’ strategy is designed to precisely identify the modes and the faults in each mode. Finally, three cases including a numerical simulation, Tennessee Eastman benchmark process and an actual semiconductor manufacturing process are presented to show the feasibility of the proposed method.