Adaptive and robust prediction for the remaining useful life of electrolytic capacitors.

In integrated avionics systems, ensuring the high reliability and lengthening the life cycle of the avionics circuits become more and more important. This paper proposes an adaptive and robust prediction method to estimate the state of health and predict the remaining useful life (RUL) of electrolytic capacitors, which is one of the most significant components in avionics circuits. Based on an accelerated aging experiment performed by NASA, the degradation mechanism of electrolytic capacitors is analyzed. According to the capacitance loss data, a combination of the Verhulst model and the exponential model is adopted as the empirical model, and the unscented Kalman filter is applied to generate the proposal distribution of the particle filter to track the degradation path. Regarding the particle impoverishment, a particle swarm optimization algorithm is adopted to optimize the residual resampling step to improve the prediction accuracy. Also, adaptively adjusting the number of particles is introduced to make the algorithm more computationally efficient. Compared with the conventional particle filter algorithms, the experiment on the electrolytic capacitors degradation data indicates that the proposed novel method is able to provide a higher accuracy for the remaining useful life evaluation.