基于主成分分析与遗传优化BP神经网络的风电场短期功率预测研究

为降低风电场弃风率及对电网稳定性影响,对风电场短期功率进行准确预测显得十分重要。针对传统BP神经网络泛化能力差、网络收敛速度慢等问题,建立了一种基于主成分分析与遗传优化BP神经网络相结合的风电场短期功率预测模型。首先,利用主成分分析法对风电场原始气象数据进行分析,将得到的独立变量作为BP神经网络的输入;然后利用遗传算法确定了神经网络的最优初始权值和阈值的大致范围,并用L-M算法对BP网络权值和阈值进行细化训练;最后,利用中国北方某风电场实际运行数据进行验证,结果表明,所建立的预测模型合理有效,不仅可以加快BP神经网络收敛速度,减少预测误差,还可以提高风电场短期输出功率的预测精度,具有一定的工程应用价值。

Research on shortterm power forecasting of wind farm based on principal component analysis and genetic optimization of BP neural network

In order to reduce the wind abandonment rate of wind farms and its impact on the stability of power grid, it is very important to predict the shortterm power of wind farms accurately. In view of the poor generalization ability and slow convergence speed of the traditional BP neural network, a shortterm power prediction model of wind farm based on the combination of principal component analysis and genetic optimization BP neural network is established. Firstly, principal component analysis is used to analyze the original meteorological data of wind farm, and the independent variables are used as the input of BP neural network; Then, genetic algorithm is used to determine the optimal initial weight and threshold range of neural network, and L-M algorithm is used to refine the weight and threshold of BP network; Finally, the actual operation data of a wind farm in northern China are used to verify the model. The results show that the model is reasonable and effective. It can not only accelerate the convergence speed of BP neural network, reduce the prediction error, but also improve the prediction accuracy of the shortterm output power of the wind farm, which has a certain engineering application value.