Due to the great effect of historical data and weather factors on the output prediction of PV,a method of combining date feature similarity and shape similarity that predicted the sub-periods of photovoltaic power was proposed.Firstly,Euclidean distance was used by the method to divide the meteorological type.Then two different selection algorithm of similar days were used to select a similar day at different periods,and based on its meteorological data and historical power value of the corresponding period,BP neutral network was utilized to predict the power output of the corresponding periods in forecasting day.The result show that the prediction accuracy of this method is improved significantly.%由于历史数据和天气因素对光伏出力预测的影响较大,提出了一种日特征相似度与形状相似度相结合的方法,分时段地预测光伏发电功率.该方法首先采用欧式距离法对气象类型进行细分,然后在不同时间段中分别利用两种相似日选取算法选取历史相似日,再利用其对应时段的历史功率值及气象数据,采用BP神经网络对预测日相应时段的功率进行预测,结果表明该方法的预测精度有明显提高.
展开▼