模糊神经网络在电力系统边际电价预测中的应用

APPLICATION OF FUZZY NEURAL NETWORK IN POWER SYSTEM MARGINAL PRICE FORECASTING

文章针对短期电力系统边际电价预测研究和应用中存在的用多元回归等传统方法建模困难、用ANN方法学习速度慢和易陷入局部极小点等问题,利用模糊神经网络具有接受和处理模糊数据、自适应地以任意精度逼近映射函数、不要求明确的数学描述等优点,建立了基于模糊神经网络的系统边际电价预测模型。通过具体实例测算及现场运用,证明了该方法为提高电力市场中边际电价预测精度、制定和实施科学合理的发电企业报价策略提供了可靠的支持。

To solve the problems occurred in the research and application of power system short-term marginal price, e.g., low learning speed while artificial neural network (ANN) is used, easy to fall into local minimum point and difficult to establish model by traditional methods such as the multiple regression and so on, a marginal price forecasting model based on fuzzy neural network is established because the fuzzy neural network possesses following advantages: it can accept and process fuzzy data, the mapping function can be approximated with arbitrary accuracy and does not need the determinate mathematical description is not needed. The calculation results of practical examples and the experience of on-site application show that using this method the forecasting accuracy of marginal price in electricity market can be improved and reliable support to the reasonable bidding strategy can be provided for power generation companies.