摘要
风速在时间和空间上具有波动性,对风力发电的稳定性具有较大影响。为实现对风速的准确预测,该文提出了一种基于混合神经网络的预测模型,该模型结合了克隆选择算法训练的小波神经网络,粒子群算法训练的小波神经网络,以及基于多层感知机和极限学习机的神经网络方法,以实现短期和长期风速预测。通过仿真实验验证了该模型在执行效率、均方根误差、平均绝对误差等方面的性能。
Wind speed fluctuates in time and space,which has a great impact on the stability of wind power generation.In order to achieve accurate prediction of wind speed,this paper proposes a prediction model based on hybrid neural network,which combines the wavelet neural network trained by clonal selection algorithm,the wavelet neural network trained by particle swarm optimization algorithm,and the neural network method based on multi-layer perceptron and extreme learning machine to achieve short-term and long-term wind speed prediction.The performance of the model in execution efficiency,root mean square error,average absolute error and other aspects is verified by simulation experiments.
作者
祁卓平
QI Zhuoping(Lintao County Meteorological Bureau,Dingxi 730500,China)
出处
《自动化与仪表》
2023年第10期91-94,99,共5页
Automation & Instrumentation
关键词
风速预测
神经网络
小波变换
wind speed forecast
neural networks
wavelet transform
