Uncertainty Management in Lebesgue-Sampling-Based Diagnosis and Prognosis for Lithium-Ion Battery

Lebesgue-sampling-based fault diagnosis and prognosis (LS-FDP) is developed with the advantage of less computation requirement and smaller uncertainty accumulation. Same as other diagnostic and prognostic approaches, the accuracy and precision of LS-FDP are significantly influenced by the parameters and uncertainties in the diagnostic and prognostic models. To improve performance of LS-FDP, this paper introduces an online model parameter adaptation scheme, which is realized by a recursive least square method with a forgetting factor. In addition, uncertainty of remaining useful life (RUL) prediction is managed by adjusting the model noises through a short-term prediction and correction loop. To verify the proposed parameter adaptation and noise adjustment methods, they are designed and implemented in a particle-filtering-based LS-FDP algorithm with applications to Li-ion batteries. Experimental results show that the proposed approach has significant improvement on both battery capacity estimation and RUL prediction.