基于数据挖掘的PSO-BP短期电力负荷预测

针对海量用电数据环境下,如何提高电力负荷预测精度的问题,采用数据挖掘对电力负荷历史数据进行聚类分析以及异常检测,并利用灰色序列对异常数据进行修正。利用蚁群算法对粒子群优化-反向传播(PSO-BP)算法进行优化,以提高算法的预测精度。通过对历史负荷数据进行试验,验证该方法的预测平均误差为3.16%,低于无数据挖掘的PSO-BP算法模型以及PSO-BP算法模型的预测误差。该方法具有一定的实用性以及有效性。

Short-Term Power Load Forecasting Based on PSO-BP with Data Mining

In order to improve the accuracy of power load forecasting under the circumstance of massive electricity consumption data,data mining was proposed to perform clustering analysis and anomaly detection on historical data of power load,and the grey sequence was used to correct the existing abnormal data.The particle swarm optimization-back propagation(PSO-BP)algorithm was optimized by ant colony optimization to improve the prediction accuracy of PSO-BP.Tests on historical load data showed that the average prediction error of the proposed method was 3.16%,which was lower than the prediction error of PSO-BP algorithm model without data mining and PSO-BP algorithm model.The proposed method was practical and effective.