摘要
提出了一种交替梯度算法对径向基函数(RBF)神经网络的训练方法进行改进,并将之运用于电力系统短期负荷预测。交替梯度算法通过优化输出层权值和优化RBF函数的中心与标准偏差值来实现。改进的算法与传统梯度下降算法相比,具有更快的收敛速度和更高的预测精度。所构建的负荷预测模型综合考虑了气象、日类型等影响负荷变化的因素,并在预测形式上做了巧妙处理。预测结果表明改进的RBF网络算法具有一定的实用性。
This paper proposes one kind of alternant gradient algorithm for improving the training of RBF neural network, which is applied to short-term electric load. This algorithm came true by optimum output layer coefficient and center and standard deviation of optimum RBF function. Compared to the traditional gradient drop algorithm, the improvement algorithm has quicker convergence rate and higher forecasting precision. The forecasting model considers many influencing factors such as weather, date-type, and so on, and deals with forecast forms very tactfully. We can see that the forecasting model has certain usability from the result of forecasting.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2008年第23期45-48,共4页
Power System Protection and Control
关键词
短期负荷预测
交替梯度算法
人工神经网络
径向基函数
实用性
short-term load forecasting
alternant gradient algorithm
artificial neural network
radial basis function (RBF)
usability