SBR-Extended Kalman Filter model-based fault diagnosis and signal reconstruction for the papermaking wastewater treatment process

Abstract

In the study, for an existing paper mill, to realize the fault diagnosis and further fault signal reconstruction for the papermaking SBR wastewater treatment process, a novel model-based SBR-EKF (Extended Kalman Filter) fault diagnosis model has been proposed. Combining the SBR process model and EKF method, using the field 120 sets of normal data of dissolved oxygen (DO) and level (L) from a paper mill, the model-based SBR-EKF fault diagnosis model for the papermaking SBR wastewater treatment process was established, and the weighted sum of squared residuals thresholds (WSSR0) was determined off-line. Subsequently, based on the normal data, four common types of faults, fixed bias, drift bias, total failure, and precision degradation, were generated and applied to the developed SBR-EKF model for on-line monitoring. The fault diagnostic results show that, by comparing the calculated WSSR with the obtained WSSR0, the developed model-based SBR-EKF model demonstrated acceptable fault detection rates for DO and L. Moreover, using the filtered value of the SBR-EKF model, the effective signal reconstructions for L and DO were realized. These investigation results reveal the effectiveness of the proposed SBR-EKF fault diagnosis model, achieving fault diagnosis with acceptable precision and reconstructed signal.