基于数据状态驱动的电力用户智能负荷预测方法研究

Research on Intelligent Load Forecasting Method for Electric Power Users Based on Data State Drive

提出了基于状态分析的变输入结构SVM预测模型。通过辨识负荷数据中辨识关键的负荷特征,通过存储负荷特征而非原始负荷数据,实现对用户负荷数据的精准高效处理。基于状态分析的变输入结构SVM模型,根据预测日各时段状态预测结果搜索历史相近时段的同状态负荷,并将其作为模型输入因素进行预测,有效克服了用户用电漂移效应给负荷预测造成的干扰,提高了预测精度。

This paper proposes a variable input structure SVM prediction model based on state analysis.By identifying the key load characteristics in the load data and storing the load characteristics instead of the original load data,accurate and efficient processing of user load data is realized.The variable input structure SVM model based on state analysis searches for the same state load in similar historical periods according to the results of state prediction in each period of the forecast day,and uses it as the input factor of the model to predict.It effectively overcomes the interference of user power drift effect on load forecasting and improves the forecasting accuracy.