基于智能家电的短期电力负荷预测与削峰填谷优化

通过可远程控制的联网智能家电，提出对城市群电力负荷的短期预测与削峰填谷优化. 分析某家电企业的智能家电集群运行产生的海量数据，建立城市群智能家电电力负荷预测模型，主要采用3种模型加权组合预测的方式，利用负荷数据中的趋势性、周期性、相关性、节假日特征及外部变量进行智能家电集群电力负荷的短期预测，单月内每日平均相对误差为4%~6%. 通过合理选择特征，该模型可以在不同家电间通用，依据家电类型分类预测后的结果可加和成为用电总负荷. 针对使用方式与用户习惯，提出智能家电电力负荷削峰填谷的控制策略，根据发电成本数据给出预期效益，说明基于智能家电负荷预测的用电调控能够有效降低电力部门发电成本、用户用电成本与电网负荷波动性.

Short term load forecasting and peak shaving optimization based on intelligent home appliance

Short term power load prediction model and peak shaving optimization in city scale were presented with remotely controlled online intelligent home appliances. Mass operational data from intelligent appliance were used. The prediction model was constructed, which ensembles three models, comprising trend, seasonality, autocorrelation, holiday effect and other factors to full the extent. The model can predict with the average daily relative error between 4%~6% per month. The model can be used on other types of home appliance and sum up to total power load through carefully selected features. Solutions for peak shaving were presented according to operational method and user preference. Returns of intelligent home appliances were estimated with power generating cost data. Power control strategy based on power prediction with intelligent appliance can effectively lower electricity generating cost, user cost and network load volatility.