中长期负荷预测的GMDH多结构自动搜索模型  

An Improved Medium and Long Term Load Forecasting Model Based on GMDH Variable-structure Auto-searching Algorithm

作　　者：林佳[1] 程浩忠[1] 顾洁[1] 杨宗麟[2] 王峥[2] 

机构地区：[1]上海交通大学电气工程系,上海200240 [2]华东电网有限公司发展计划部,上海200002

出　　处：《电力系统及其自动化学报》2010年第5期41-45,共5页Proceedings of the CSU-EPSA

基　　金：国家重点基础研究发展计划(973)资助项目(2009CB219703)

摘　　要：针对我国当前经济、政策变动的大背景,提出了采用数据分组处理方法GMDH(group method of data handling)结合多结构突变理论,实现时序突变点自动搜索建模,建立了中长期负荷预测的GMDH多结构自动搜索模型。该模型能够客观准确地搜索时间序列中的所有突变点,并充分利用突变点信息修正由于经济环境和突发事件引起的预测偏差,大大提高了传统时序外推预测模型的精度。华东地区的实际算例结果表明了该模型的有效性。An improved medium and long term load forecasting model based on GMDH（group method of data handling） multi-layer iterative algorithm and variable-structure analysis is proposed for nowadays＇ China with a various turns of events.With all the discontinuity points in time series precisely and automatically recognized,the established model combines the variable-structure analysis to correct the deviation caused by economic environmental changes or emergencies.The model improves the forecasting accuracy.The results of Eastern China＇s industrial power consumption verify its feasibility.

关 键 词：中长期电力负荷预测 数据分组处理 多结构突变 自动搜索算法 华东地区 

分 类 号：TM715[电气工程—电力系统及自动化]