应用最小二乘支持向量机预测煤的发热量

为解决现有的煤发热量预测神经网络法的过学习与局部极小点问题,通过对煤热量数据的分析,在统计学习理论和结构风险最小化准则的基础上,建立了基于最小二乘支持向量机(LS-SVM)的煤发热量预测数学模型。在算例分析中与BP神经网络、RBF神经网络预测法进行对比,发现该方法比BP和RBF神经网络具有更高的预测精度,且具有收敛速度快、泛化能力强等优点,为燃煤发热量的预测提供了一种有效的方法。

Forecasting Calorific Value of Coal with the Method of least-Squares Support Vector Machine

In order to overcome the over-fitting problem and the local minor problem of the artificial neural network(ANN) method in prediction of calorific value of coal,a new mathematical model according to the Least-Squares Support Vector Machine(LS-SVM) theory was developed.The model was based on the analysis of the characters of coal calorific value data,the statistical theory(SLT),and the empirical risk minimization(ERM) principle.The experimental results indicated that compared to the back propagation(BP) and Radial Basis Function(RBF) neural network method,this model was faster in computation and had better generalization performance,which provides a highly effective method for the prediction of coal calorific value.