摘要
根据某风场数据,先对风场数据进行修正,剔除错误数据,以避免累计误差的出现,提高预测精度。在神经网络的输入变量中不仅考虑了风速风向,还加入了跟大气运动形成风能的温度、重力常数和海拔。通过对神经网络法的粒子群优化算法(PSO)优化和惯性权重的调整来预测风速,通过神经网络训练该方法能够提高预测的准确性,能够改善风电并网的稳定运行和电网调度的调整。
According to the data of a wind field, the wind field data is corrected firstly, and then the wrong data is eliminated in order to avoid the emergence of the accumulated error and to improve the prediction precision. In the input variables of neural network, it not only considers the wind speed and direction, but also the wind temperature, gravitational constant and elevation formed with atmospheric movement. Through optimizing the particle swarm optimization （PSO） of neural network method and adjusting the inertia weight, it can forecast the wind speed, and through the neural network training method it can improve the prediction accuracy, which can improve the stable operation with wind power integration and the adjustment of power grid scheduling.
出处
《四川电力技术》
2016年第6期18-22,共5页
Sichuan Electric Power Technology
基金
教育部创新团队(IRT1285)
自治区重点实验室(2016D03021)
国家自然科学基金(2013211A006)
