A Hierarchical Quality-Related Fault Diagnosis Method for Nonlinear Industrial Systems

This article focuses on the quality-related fault diagnosis problem in industrial systems in terms of the proposed hierarchical data-driven (HDD) method. As opposed to the existing approaches, the proposed approach makes use of the ensemble learning method to avoid selecting the optimal kernel parameter, saving computation costs in the training process. Further, a novel quality-related fault diagnosis scheme that incorporates the quality-related fault detector and the weighted fault reconstitution method is designed to monitor the fault and reveal the trends of the quality state. In the end, both a simulation case and an industrial case are applied to demonstrate the effectiveness of the proposed method.