基于最小二乘法-加权马尔科夫链模型的母线负荷预测  

Bus Load Forecasting Based on Least Square Method and Weighted Markov Chain Model

作　　者：杜雅楠 齐敬先 施建华 DU Ya'nan;QI Jingxian;SHI Jianhua(NARI Group Corporation(State Grid Electric Power Research Institute),Nanjing 211106)

机构地区：[1]南瑞集团有限公司(国网电力科学研究院有限公司),南京211106

出　　处：《计算机与数字工程》2022年第8期1858-1863,共6页Computer & Digital Engineering

基　　金：国家电网公司总部科技项目(编号:B520609170001)资助。

摘　　要：母线负荷预测精准结果可为电网安全运行提供预警策略、为电网运行方式的变更计划编排、为无功优化提供调度方案等。论文针对现有技术预测精度低、模型训练复杂等问题提出了一种基于最小二乘法及加权马尔科夫链模型的母线负荷预测,该模型基于最小二乘法实现对负荷曲线的基础拟合,然后采用加权马尔科夫链模型对其预测精度提升并防止最小二乘法的过拟合,同时采用部分指数加权法在减少转移概率矩阵生成频次的基础上提升中长期母线负荷的预测准确率,最终实现对预测结果的优化提升。利用所述方法对新疆电网某母线的负荷数据进行预测,结果表明该方法建模速度快且具有较高的准确率。The accurate result of bus load forecasting can provide early warning strategy for the safe operation of power grid,change plan arrangement for power grid operation mode,and dispatching scheme for reactive power optimization. This paper proposes a bus load forecasting method based on the least square method and the weighted Markov chain model to solve the problems of low prediction accuracy and complex model training. The model realizes the basic fitting of the load curve based on the least square method,and then uses the weighted Markov chain model to improve the prediction accuracy and prevent the over fitting of the least square method. At the same time,based on reducing the frequency of generating transfer probability matrix,the partial index weighting method is used to improve the prediction accuracy of medium and long-term bus load,and finally to optimize the prediction results. The method is used to predict the load data of a bus in Xinjiang power grid. The results show that the method is fast and has high accuracy.

关 键 词：最小二乘法 加权马尔科夫链 母线负荷 曲线拟合 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]