Research On Equipment Defect Prediction Based On The Time Series

Equipment defects trend predict in electric production can extracting useful underlying rules from its time series. By analyzing the current and history defect data of electric equipment, we can predict the trend of future defect. So the service units can make preparation in advance and providethe decision for the administrators, it has a strong significance of theory and practice. Summery to the content of research and results of this paper as flows:Firstly, this paper apply ARMA and ANN method to prediction defect of equipment, it indicate that the validity of this method because of good result.Secondly, this paper combines two methods to predict, and propose the flowing views:If the number of nerve network in input layer is equal to the order in AR mode during the course of forecasting it can improve precision of the forecast. It is also confirmed in cases. So, it overcome the traditional defects that is to decide the number nerve unite by experience or trial method.