Fault Prediction for Power Plant Equipment Based on Support Vector Regression

To provide effective fault prediction on power plant equipment, a method of fault prediction based on support vector regression is proposed in this paper. First, we calculate the correlation coefficient to select proper features to form the feature vector, Then we use the grid search method to optimize the two important parameters of support vector regression, Finally, we establish the prediction model with the feature vector and the optimized parameters obtained above to predict expected values of corresponding data items of the equipment. The analysis of one day data of coal mill of a measurement point A1 shows that, compared with the method without optimization, the root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of this method have significantly reduced. This result indicates that: the model established by this method is able to predict the value of measurement points more accurately with superior generalization ability, and can be applied in the field of fault prediction.