基于遗传算法和神经网络的分时段风速预测方法

The Wind Speed Prediction Based on Genetic Algorithm and Neural Network with Different Periods

为提高风速预测的准确性,提出一种分时段GA-BP(遗传算法优化BP神经网络)的风速预测方法,以遗传算法来优化BP神经网络,并将原始数据进行分时段处理,改善训练样本的相似程度。基于matlab进行了仿真验证,结果表明:遗传算法优化BP神经网络使其预测结果的平均相对误差降低,准确性提升;原始数据分时段处理后,预测准确性进一步提升。

In order to improve the accuracy of wind speed prediction, a wind speed prediction method based on GA-BP is proposed. The proposed method divides original data into different periods of time to improve training sample＇s similarity. Simulation results show the absolute error of GA-BP decreased and its accuracy promoted. Furthermore, the accuracy of the prediction becomes more accurate after original data processing. Therefore, the proposed method improves the performance of wind speed prediction, and it provides more accurate gist for wind power grid scheduling control.