Observer-Based Fault Detection Approach Using Fuzzy Adaptive Poles Placement System With Real-Time Implementation

The conventional methods of observer poles placement in sensor fault detection usually adopt the trial-and-error methods. These methods cannot achieve global optimal performance because of their fixed poles placement and it leads to an observer with constant parameters, which could be reducing the system performance. Therefore, this paper proposes a fuzzy-based observer tuning method to optimize and adapt the selection of poles locations to determine the optimal gains of the observer, and it is experimentally applied to a composite sensor fault detection. Fuzzy logic is a promising method that could overcome the trial-and-error method challenges by introducing better adaptation and system robustness. The proposed observer structure includes adaptive tuning corresponding to an unknown input. Utilizing self-tuning for the observer correction stage, the gain is going to be updated online using the proposed fuzzy adaptive poles placement (FAPP) system. This paper validated the system simulation by implementing fault detection algorithms by using a real-time embedded observer-based system. The experimental results demonstrate the effectiveness of the proposed fuzzy-based observer schemes at detecting sensor faults in the Brushless DC (BLDC) motors, with significantly better performance than conventional counterparts' methodologies. The experiments indicate that the average estimation error is 0.146, which less by 43.8% than was obtained for high levels of noise and disturbances compared with the traditional Luenberger observer approach.