一种自主多层神经网风能预测算法

A Self-organized Multilayer Neural Network Wind Power Prediction Algorithm

当前以风电为代表的可再生能源技术在世界范围内得到了快速发展。由于风电的不稳定性,自然风力时强时弱会引起风电的电能质量、电压、电网稳定性等问题,最终影响整个电网的稳定性。进行风能预测可以预估风电的发电出力程度,对于整个电网的稳定十分重要。基于此,提出一种自主多层神经网风能预测算法,通过自组织算法来动态地决定风能预测神经网模型的层数、中间节点个数、传输函数类型。试验表明通过本算法获得的模型具有更高的预测精度。

At present, the renewable energy technologies represented by wind power have been developing rapidly all over the world. Due to the instability of wind power, natural wind power will cause the problem of power quality, volt-age and grid stability, which will affect the stability of the whole power grid. Wind power prediction can predict the extent of wind power generation, the stability of the entire power grid is very important. Based on this, a self- orga-nized multi-layer neural network wind power prediction algorithm （ S-MLP） was proposed, the number of layers, the number of intermediate nodes, and the type of transfer function of the neural network model could be dynamically de-termined by the self-organizing algorithm. The experimental results show that the model obtained by this algorithm has higher prediction accuracy.