Study on State Prediction Method for Electronic System

There are various fault modes in many electronic systems and it is very difficult to predict these complex signals which show the fault symptom. The relations between state features and states of two electrical systems are mainly studied and the paper proposes a hybrid method which takes full advantage of individual models to predict state of electronic system: Firstly through using the correlation of state feature ,the embedding dimension is calculated to reconstruct the phase space of state feature; Secondly, according to the character of state feature ARMA-GM(1,1) model is used to predict its linear part and LSSVM with parameters optimized by PSO algorithm is applied to predict its residual part, thus the final predicted values can be gained through adding the two parts of predicted values. Two experiment results show that the forecasting performance of the proposed method is superior to other methods and offers a potential for electronic system condition prognosis.