A Novel Degradation Prediction for Analog Circuits using Particle Filter

With the increasing demand of high reliability and safety of modern electric devices, failure prediction becomes more and more important since it is efficient to increase reliability and reduce downtime cost. A novel prediction method for analog circuits is proposed in this paper. Firstly, output waveforms in time domain of the initial state and the degradation states are extracted, then particle filter algorithm is implemented to estimate the changes of the waveforms according to the principles of noise estimation based on Grey Theory to obtain more reasonable fault indicators from more complete information. Thereafter, a novel degradation prediction model for analog circuits is constructed according to the newly obtained fault indicators. To validate the proposed degradation prediction method, the experiments are implemented on high-voltage power circuit board. The experimental results show that the method can predict the degradation trend and the information will be useful for the reliability design of the analog circuits.