Research on methods of forecasting unburned carbon content in the fly ash from coal-fired power plant

This paper proposed a new algorithm technologies for forecasting the unburned carbon content in the fly ash from coal-fired utility boilers by combination improved artificial bee colony algorithm with support vector machine ABC-SVM, for comparative purpose, back propagation neural network (BP) was also presented, comparing the pros and cons of both in the field of the predictive ability. Applied to a 1000MW coal-fired utility boiler, the ABC-SVM model which had been trained forecasted the unburned carbon in the fly ash in the test samples set, and got the mean square root error and the mean relative error of 1.25%, and 1.79%, respectively, which are 33.75% and 46.63% of BP neural network. These results show that ABC-SVM method is more accurate than the BP neural network, and can satisfy the forecasting demand well.