基于优先选择的风电功率超短期预测

Ultra-Short Term Wind Power Forecast Based on Priority Selection

提出一种基于优先选择的风电功率超短期预测算法,该算法结合BP神经网络、天气预报、实测功率外推法等多种预测算法,并能动态选择输入数据最优的预测算法进行预测。优先选择法的各个子算法分别采集不同的输入数据进行预测,这种做法能有效地规避单一输入数据无效时,整个预测失败的情况。同时,结合在线建模,能动态调整各个子算法的执行顺序。经过实际运行后,该算法运行效果较好,4h内的预测均方根误差在10%以内。

Proposes a priority selection algorithm, which combined with neural network, weather forecast, the measured power extrapolation algorithm, can dynamically select the prediction algorithm which having the best input data. Each sub-algorithm of priority selection collects different in put data to predict, this approach avoids the prediction failure effectively, when a single data input invalid. At the same time, combines with the online modeling, the order of execution of each algorithm can be dynamic adjusted. In practical operation, it acquires a good effective- ness, the RMS error of prediction less than 10% in 4h.