基于神经网络的浙江发电市场清算电价预测方法  

FORECASTING MARKET-CLEARING PRICE IN ZHEJIANG GENERATION MARKET USING NEURAL NETWORK

作　　者：刘正国 付祥明 高伏英[2] 王立群[3] 郑苗荣 王维明 胡余富 陈武军 杨国富 

机构地区：[1]宁波四维科技有限公司,浙江省宁波市315600 [2]华东电力调度局,上海市200002 [3]浙江电力调度通信中心,浙江省杭州市310000 [4]镇海发电有限责任公司,浙江省宁波市313000

出　　处：《电力系统自动化》2002年第9期49-52,共4页Automation of Electric Power Systems

摘　　要：市场清算电价 ( MCP)预测是电力市场决策的基础。文中以浙江省电力市场为背景 ,采用 BP神经网络的方法 ,建立了提前一天的浙江电网发电市场清算电价短期预测模型 ,并针对不同时段分别建模。此外 ,根据浙江发电市场的特点 ,考虑市场外机组出力的影响 ,从供需出发 ,提出了竞价空间的概念 ,并据此设计了神经网络的结构。在线预测结果表明 ,该方法具有良好的预测效果。Forecasting the market- clearing price MCP is the basis of decision making for each participant in electricity market.This paper presents a successful application of back- propagation neural network to forecast MCP in Zhejiang day- ahead generation m arket.By this method,different models are designed for the different settlem ent periods respectively. Especially,according to Zhejiang circumstance,a new concept of Bidding Space based on the relationship between supply and demand,which reflects the existences of generators'output outside the com petitive m arket,is used in m odeling.Preliminary application to the real power system shows the efficiency and accuracy of the proposed m ethod.

关 键 词：神经网络 浙江 发电市场 清算 电价 预测 

分 类 号：F426.61[经济管理—产业经济]