基于对数平均迪氏指数-布谷鸟搜索算法-最小二乘支持向量机的区域中长期电力需求预测

Forecast of Regional Medium and Long-term Electric Power Demand Based on Logarithmic Mean Divisia Index Decomposition and Least Squares Support Vector Machine Model

中长期用电需求预测对地区电网规划与运行具有重要意义。准确地对其进行预测有助于配电网利用效率的提高。中长期用电需求与地区宏观经济形势息息相关,选用(对数平均迪氏指数)LMDI分解模型对用电增长量进行了分解。根据影响因素分解为生产效应、结构效应及强度效应,然后选用布谷鸟搜索优化的最小二乘支持向量机对各效应数据进行拟合及预测,再加总得到预测用电需求量。最后,以冀北地区为例对模型进行测算,通过与不同预测方法比较证明了基于LMDI电量分解的CS-LSSVM模型在中长期电力需求预测方面有较高的准确度。

The prediction of medium and long term electricity demand is of great significance to the planning and operation of regional power grid,and the accurate prediction of the demand for electricity will improve the efficiency of resource utilization.And medium and long term electricity demand is closely related to regional macro-economic situation.The logarithmic mean divisia(LMDI)decomposition model is used to decompose the growth of electricity consumption and decompose it into production effect,structure effect and intensity effect according to the influence factors.Then the Cuckoo Search optimization of least squares support vector machine is used to fit and predict the data of the above effect,which could be added to get the growth of electricity demand.Finally,the northern Hebei Province as an example was taken to calculate the model.After that,compared with different forecasting methods,the CS-LSSVM model based on LMDI decomposition has higher accuracy in medium and long term power demand forecasting.