基于BP神经网络优化的风电场短期功率预测研究

发布时间：2018-05-15 15:42

本文选题：风力发电 + 功率预测　；参考：《昆明理工大学》2017年硕士论文

【摘要】：风能作为一种绿色清洁的能源,以其成本低廉,便于开发利用的优势,开始从补充能源向战略替代能源转变。我国约20%的国土都具有比较丰富的风能资源,无论是发展规模还是发展水平都有很大的进步和提升,风电在我国有着巨大的发展潜力。但是,由于风能具有随机性和间歇性的特点,造成了其功率输出的不稳定,也给电力系统的正常稳定运行带来了挑战。因此,只有做好风电功率预测的工作,才能有效的管理风电场运行,保证电力系统的安全以及电能质量。基于这个背景,本文以风电场短期功率预测方法为研究内容,通过神经网络预测的手段,对风电场预测方法进行研究和探讨,论文的主要工作有如下几个方面:首先,对风电场功率预测方法进行分类,在综合比较现存各种方法后,本文决定采用BP神经网络的方法预测风电功率。在介绍了 BP神经网络原理的基础上,详细分析了影响风电场输出的因素,确定了以风速、风向正弦和余弦作为影响风电输出的最主要因素。其次,选定BP神经网络对风电功率进行预测,以某风电场的历史运行数据作为模型训练数据的来源,接着选取典型测试样本数据来验证预测的精度。结果表明,BP神经网络有着较好的预测表现,但是不太稳定。最后,为了进一步提高预测精度,提出了以人工蜂群算法优化的BP神经网络预测模型。以相同的样本数据训练之后,选取同样的典型测试样本数据进行预测精度的验证。结果表明,该方法能大大减小BP神经网络的预测误差。
[Abstract]:Wind energy as a kind of green and clean energy, with its advantages of low cost and easy development and utilization, began to change from supplementary energy to strategic alternative energy. About 20% of our country has abundant wind energy resources, both the development scale and the development level have great progress and promotion, wind power in China has a great potential for development. However, due to the randomness and intermittency of wind energy, the instability of power output and the challenge to the normal and stable operation of power system are brought about. Therefore, the wind power prediction can effectively manage the operation of the wind farm and ensure the safety and power quality of the power system. Based on this background, this paper takes the short-term power forecasting method of wind farm as the research content, through the means of neural network forecast, carries on the research and the discussion to the wind farm forecast method. The main work of the paper has the following aspects: first, The methods of wind farm power prediction are classified. After a comprehensive comparison of the existing methods, this paper decides to use BP neural network to predict wind power. On the basis of introducing the principle of BP neural network, the factors influencing wind farm output are analyzed in detail, and the wind speed, wind direction sinusoidal and cosine are determined as the most important factors affecting wind power output. Secondly, BP neural network is selected to predict wind power, and the historical operation data of a wind farm is used as the source of model training data. Then, typical test sample data are selected to verify the prediction accuracy. The results show that the BP neural network has a good prediction performance, but is not very stable. Finally, in order to further improve the prediction accuracy, a BP neural network prediction model optimized by artificial bee colony algorithm is proposed. After training with the same sample data, the prediction accuracy is verified by selecting the same typical test sample data. The results show that this method can greatly reduce the prediction error of BP neural network.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM614;TP183

【参考文献】