【摘要】 短期负荷预测是电力系统运行和分析的基础,对机组组合、经济调度以及安全校核等具有重要意义。针对地区负荷在小样本情况下预测精度不高的问题,在对某地区负荷数据进行分析并剔除异常数据之后,建立了基于支持向量机回归(support vector regression,SVR)的短期负荷预测模型。为了提高模型的预测性能,采用细菌觅食算法(bacteria foraging optimization algorithm,BFOA)对SVR的参数进行优化,并将温度、湿度和降雨量等气象信息引入预测模型。考虑到负荷与时间点的耦合关系,对每日96个时间点分别进行预测。同时,根据工作日和周末2种不同属性分别建立了基于SVR的负荷预测模型。仿真结果表明,所建立的短期负荷预测模型能够在小样本的情况下以较快的速度获得较高的预测精度。更多还原

【Abstract】 Short-term load forecasting is the basis of operation and analysis for power system,which is of great significance to unit combination,economic dispatch and safety check. Aimed at the low accuracy problem in area load forecasting of small samples,this paper establishes a short-term load forecasting model based on support vector regression( SVR) after analyzing the load data of a certain area and removing the abnormal data. In order to improve the prediction performance of the model,we optimize the parameters of SVR by bacteria foraging optimization algorithm( BFOA),and introduce the meteorological information such as temperature,humidity and rainfall into the prediction model. Considering the coupling relationship between the load and the time points,we predict the load of daily 96 time points respectively. At the same time,we establish two load forecasting models based on SVR according to different attributes of working day and weekend. The simulation results show that the proposed short-term load forecasting model can obtain high prediction accuracy at a faster speed in the case of small samples.更多还原