摘要
针对小水电站年发电量序列的特点,将最小二乘支持向量机(least squares support vector machine,LS-SVM)回归模型引入年发电量预测领域,并给出了相应的过程和算法。与常规基于人工神经网络(artificial neural net-works,ANN)的智能预测方法比较,该模型优点是明显的:①将神经网络迭代学习问题转化为直接求解多元线性方程;②整个训练过程中有且仅有一个全局极值点,确定了预测的稳定性。最后,一个实际的预测例子表明:该模型实现容易、预测准确,适用于小水电站预测。
Through analysis of the specialty of annual power generation series of small hydropower stattons, a regression mooel based on the least squares support vector machine (LS-SVM) was introduced into the field of annual energy production forecasting in this paper. The design steps and learning algorithm were also provided. Compared with conventional intelligent forecasting methods based on artificial neural networks (ANN), the proposed method has the advantages of: OThe LS-SVM model changes the iteration problem of neural network into a solution problem of a set of linear equations;(2)The global optimal solution can be uniquely obtained because there is only one global extremum in the training process. So the result of forecasting is stable. The application example proves that the proposed method is easy to realize with accurate prediction, and can be applied to small hydropower stations.
出处
《中国农村水利水电》
北大核心
2007年第2期93-95,98,共4页
China Rural Water and Hydropower
关键词
最小二乘支持向量机(LS-SVM)
小水电站
年发电量
时间序列
预测
least squares support vector machine ( LS- SVM)
small hydropower stations
annual power generation
time series
forecasting