Root Cause Diagnosis of Plant Wide Oscillations based Fuzzy Kernel Granger Causality

Plant-wide oscillations are common phenomenon in large-scale industrial processes, which once generated in one unit and can spread along flows in the whole plant, then cause product quality degradation or safety problems; thus, it is important to diagnose the root cause. Plant-wide oscillations are nonlinear, and the kernel Granger causality is proposed in some literature for root cause diagnosis to solve the problem that the conventional Granger cannot deal with nonlinear signal. In this paper, a fuzzy kernel granger causality method is proposed to save operation time. By application to a small nonlinear numerical and industrial process, this method shows the effectiveness. In the root cause diagnosis of industrial cases, the method also successfully detected the correct root point.