Improved particle filter remaining life prediction method for switch mode power supply using variance of case temperature

Switch mode power supply plays a crucial role in industrial systems as an energy source. Its remaining life prediction is vital to help in schedule maintenance to avoid the unexpected failures of industrial systems. Due to the high cost of monitoring the degradation characterization parameter and the complexity of the degradation rule, predicting remaining life for the switch mode power supply is challenging. This study used a novel degradation characterization parameter, i.e., the variance of the case temperature, to characterize switch mode power supply degradation. Multi-stage state equations and degradation thresholds of each stage were established based on the change in the variance of the case temperature. Additionally, the state equations and degradation thresholds were introduced into the traditional particle filter method to form an improved particle filter, which is a useful life prediction method. The effect of the proposed method was evaluated and compared with other methods. The results show that over the life of a switch mode power supply, there are three change stages in the variance of the case temperature: the constant tendency stage, linear increase stage, and exponential increase stage. The degradation thresholds of these stages were 20 %, 70 %, and 200 %, respectively. The predicted curve of the variance of the case temperature was in good agreement with the test curve, and the prediction interval at a 90 % confidence level was narrow. Compared with other algorithms, the improved particle filter method could maintain the prediction error within 5 %, and was more suitable for predicting the remaining life of a switch mode power supply with multi-stage degradation characteristics.