摘要
市场清算电价(MCP)预测是电力市场决策的基础。文中以浙江电力市场为背景,对一些预测技术作了介绍;用神经网络、时间序列以及基于小波分解的时间序列预测方法对浙江电力市场MCP作了预测。预测结果表明时间序列方法和基于小波分解的时序方法在一周的MCP预测过程中精度衰减较快,但是基于小波分解的时序方法在下一日的MCP预测中还是有较好的精度;神经网络方法预测精度衰减较慢,预测效果相对比较稳定。
Forecasting the market-clearing price (MCP) is the basis of decision making for each participant in electricity market. In this paper,some short-term forecasting methods are introduced. The ANN method, time series method and time series method with wavelet decomposition are employed to forecast MCP in Zhejiang electricity market. The experimental results show that the precision of both time series method and wavelet-decomposition-based time series method attenuates rapidly as time goes by, but the next-day forecasting with the latter one has good precision. The ANN method has a stable performance.
出处
《继电器》
CSCD
北大核心
2004年第11期58-61,共4页
Relay
基金
宁波市重点博士基金(2003A61006)
关键词
电力工业
电力市场
MCP
电价预测
时间序列预测
浙江
electric power market
market clearing price
time series
ANN
wavelet transform
price forecasting
