一种改进的风电功率超短期组合预测

An Improved Supper Short-term Combination Forecast of Wind Power

为提高风电功率超短期预测精度,针对目前的风电功率超短期组合预测算法都是将各子预测算法的权重设为固定值,导致风电功率超短期预测结果精度不高的问题,提出一种改进的风电功率超短期组合预测算法。该算法包含BP神经网络、天气预报、实测功率外推法等子预测算法,结合实际运行情况判断各子预测算法的执行结果,并根据执行结果动态改变各子预测算法的权重,以保持较高的预测精度。实际应用效果表明:该算法预测精度较高,运行效果较好,4h内的预测均方根误差在10%以内。

In order to improve the supper short-term forecasting precision of wind power, and aiming at the problem that weights of sub-prediction algorithms are set as fixed values in the wind power supper short-tei^n combination forecasting algorithm which caused lower precision of supper short-term forecasting results, a modified wind power supper short-term combination forecasting algorithm is proposed. The proposed algorithm included sub-prediction algorithms such as BP neural network, weather forecast, and measured power extrapolation prediction algorithm and so on, and cannot only judge execution results by combining actual operation situation and but also change dynamically weights of sub-prediction algorithms according to execution results to keep higher forecasting precision. Effects of actual application show that the algorithm has higher forecasting precision, better operation effect and 4b root mean square error of prediction within 10%.