基于负荷特性聚类的样本自适应神经网络台区短期负荷预测

 介绍了批量处理时间序列数据情况下，基于台区负荷特性聚类的样本自适应反向传播神经（BP）神经网络预测短期电力负荷的方法，通过对历史数据的预处理、初始聚类中心的设置以及最优聚类数目的确定，建立典型日负荷曲线的聚类预测模型。基于历史数据的聚类结果及待预测日的温度、湿度、气压、风速、星期等相关参数，使用BP神经网络算法得出待预测日负荷曲线预测结果。通过实例验证，基于台区负荷特性聚类的样本自适应神经网络短期负荷预测能够得到较为准确的预测结果。

A neural network for short term load forecasting based on Sample self adapted of load characteristics clustering

This paper introduces the methods and the steps of predicting the power load by the BP neural network with cluster optimization in batch processing time series. Through the preconditioning of historical data, the setting of the initial clustering center and the determination of the optimal number of clusters, a clustering prediction model of the load curve is established based on the clustering results of the historical data and the relevant parameters such as the temperature, the humidity, the air pressure, the wind speed and the time (the current week). The results show that with the clustering algorithm, the related factors and the BP network adaptive rate can be comprehensively considered, while the training speed is improved, to obtain more accurate prediction results.