基于GRNN神经网络的短期电力负荷预测

Short-Term Power Load Prediction Based on GRNN Neural Network

使用安徽省某地区月电力负荷原始数据,在深入研究风速、温度和湿度对电力负荷的影响的基础上,利用广义回归神经网络(GRNN)建立了短期电力负荷的模型,通过GRNN神经网络的非线性映射能力预测出日负荷数据,研究结果表明:平均相对误差小于2.2%,最大相对误差小于4%,,优于同等条件下建立的多层前馈神经网络误差反向传播(BP)电力负荷预测模型,可以作为对电力负荷合理规划和准确调度的重要依据。

Using the monthly electricity load data from an area in Anhui Province, and based on an in-depth study of the impact of wind speed, temperature, and humidity on power load,a generalized regressionneural network （GRNN） was used to establish a short-term power load model through the use of GRNN neuralnetworks. The linear mapping capability predicts the daily load data. The results show that the average relativeerror is less than 2.2% and the maximum relative error is less than 4%. It is superior to the backpropagation（BP） power load of multilayer feedforward neural networks established under the same conditions. Theprediction model can be used as an important basis for reasonable planning and accurate scheduling of powerloads.