An adaptive Kalman filter-based estimation method for online oxygen flow measurement in PEMFCs with mismatch detection

Accurate airflow monitoring is critical for optimizing the performance of PEMFCs in dynamic environments. However, existing sensor techniques face significant limitations due to safety risks and leakage concerns, making direct measurement impractical. Furthermore, the complex interactions among system parameters challenge traditional observer techniques, as parameter mismatches often compromise estimation accuracy. To address these issues, this paper proposes a novel state estimation method to achieve robust online oxygen flow estimation across diverse scenarios. The proposed method uses a square root cubature Kalman filter to fuse the predictive model with limited sensor signals, enabling precise estimation of unmeasurable states in the cathode channel. To deal with model uncertainties, a Mahalanobis distance-based metric is introduced to assess the occurrence of mismatches, while a cascade classifier identifies specific parameters that influence estimation performance. Subsequently, the corresponding observer combined with an augmented mechanism is activated to correct the estimated oxygen flow, considering the influence of mismatched parameters. Additionally, an event-triggered mechanism is employed to minimize unnecessary computational requirements. Simulation results demonstrate that the proposed method significantly outperforms traditional estimation methods, improving estimation accuracy and reducing the mean absolute error of oxygen flow estimation by over 31 %, 70 %, and 83 % when the three uncertain parameters are mismatched, respectively. This method represents a significant advancement in monitoring unmeasurable states, further driving the application of advanced estimation technologies.