A Quality-Relevant Fault Diagnosis Scheme Aided by Enhanced Dynamic Just-in-Time Learning for Nonlinear Industrial Systems

This article studies the quality-relevant fault diagnosis issue in the nonlinear industrial processes based on the proposed dynamic just-in-time learning scheme. In contrast to the existing methods that utilize all the offline data, the designed framework constructs the local model online based on the enhanced dynamic time-warping technique to handle the dynamic property and the nonlinear feature in data. In addition, the developed fault diagnosis scheme monitors the fault in the lower dimensional manifold space and indicates the variations of the quality indicators. Lastly, the effectiveness of the proposed method is illustrated through a simulation scenario and an industrial scenario.