Prediction of circulating water loss based on support vector machine and neural network

Abstract

Based on the operational data of the circulating water system in a thermal power plant, BP neural network and support vector machine regression were used to establish the prediction model of circulation water system evaporation and wind blow loss. The trail method was used to improve the BP neural network prediction model, and for the prediction model of support vector machine regression, the kernel function and the corresponding parameters were selected through optimization. The results showed that the mean square error of the simulation results of the two models were 0.071 and 0.070 respectively in summer and 0.046 and 0.047 respectively in winter, this meets the prediction requirements of project and demonstrates high prediction accuracy. With the evaluation index of neural network model, the simulation and prediction results of the two models were compared and analysed. The results showed that the simulation results of two model were basically the same, but the support vector machine model training sample time is shorter, the convergence speed is faster, and the overall network model performance is better.