基于小波包分解和改进Elman神经网络的风电场风速和风电功率预测

发布时间：2018-03-23 07:24

本文选题：风力发电　切入点：风电场　出处：《电工技术学报》2017年21期 　论文类型：期刊论文

【摘要】：准确预测风电场风速和风电功率对做好风电场运行维护、合理安排开停机计划以及确保电力系统的安全稳定运行具有重要意义。提出了基于小波包分解和改进Elman神经网络的新型风电场风速和风电功率预测方法并给出了具体应用步骤。首先利用小波包分解理论对经过初步处理的历史风速数据进行分解处理,根据相关性剔除随机数据,保留最优分解树;随后提出带扰动的PSO训练算法用以提高Elman神经网络的训练速度,并解决PSO算法易陷入局部最优解的问题;最后利用不同结构的Elman神经网络寻找最优分解树不同频段下的风速规律进而获得风速和风电功率预测结果。南方某风电场算例表明该方法具有更高的预测精度,能够正确反映风速和风电功率规律。
[Abstract]:Accurately forecast wind speed and wind power of wind farm to do a good job in the operation and maintenance of wind farm, It is of great significance to arrange the on-off plan reasonably and to ensure the safe and stable operation of power system. A new wind speed and wind power prediction method based on wavelet packet decomposition and improved Elman neural network is presented. First of all, using wavelet packet decomposition theory to decompose the historical wind speed data after preliminary processing, According to the correlation, the random data is eliminated, the optimal decomposition tree is preserved, and then the PSO training algorithm with disturbance is proposed to improve the training speed of Elman neural network, and to solve the problem that PSO algorithm is prone to fall into local optimal solution. Finally, the Elman neural network with different structures is used to find out the wind speed law under different frequency bands of the optimal decomposition tree, and the results of wind speed and wind power prediction are obtained. An example of a wind farm in southern China shows that the proposed method has higher prediction accuracy. Can correctly reflect wind speed and wind power law.
【作者单位】：哈尔滨工业大学电气工程及自动化学院;国家电网公司交流建设分公司;
【基金】：国家科技支撑计划课题资助项目(2015BAA01B00)
【分类号】：TM614

【相似文献】