Abstract
After analysing the influencing factors on the life of automotive electromagnetic relays, this paper determines six performance degradation parameters as the input parameters for automobile electromagnetic relays, including contact resistance, pick-up time, super-path time, bounce time, arc time and release time. Then, three prediction models were proposed based on radial basis function (RBF), backpropagation (BP) network and wavelets neural network (WNN), considering the nonstationary nature of electrical performance parameters of relay. The models successfully predicted the original nonstationary parameter runoff series. Through comparative analysis, it is seen that the WNN-based life prediction method is the most accurate and suitable approach for relay life prediction.
