Stability Evaluation based Non-steady Variable Identification for Online Fault Prognosis

In the present work, an online fault prognosis strategy is developed for proactive abnormality management. For online prognosis, the fault degradation process should be well revealed. To well capture the evolution process, the proposed approach includes three components, First, the stability factor is defined to identify those significant faulty variables that show degradation process. Second, the fault variations departing from normal status are extracted by performing a modified Fisher discriminant analysis (MFDA) on the selected variables in normal and fault data. These critical variations are deemed to be evolving with time and thus responsible to the future process failure. Third, the significant variations are captured to track the fault evolution process for fault prognosis by developing a vector auto-regression model to reveal how soon the process failure will happen. By the above modeling strategy, uninformative fault effects that do not present degradation are excluded so that the true fault degradation process can be focused on for online fault prognosis. The proposed method is verified by both numerical and experimental data.