大型风电场短期风电功率预测技术研究

发布时间：2018-07-26 16:13

【摘要】：随着我国风电装机容量的快速增长,风电在电网中占的比重越来越大,大规模风电并网对电力系统安全运行带来严重冲击,有效的风电场功率预测可以为电网稳定运行和调度提供参考依据。本文针对大型风电场传统功率预测精度不高和不稳定性问题,提出了一种大型风电场分组的智能优化功率预测模型,具体研究内容如下:首先,对大型风电场的参数特征进行分析和规律统计。研究了大型风电场风速和风向等参数的特性,分析了风电场风速、风向、温度与发电功率之间的关系,统计了整个风场的发电功率的规律特征,对大型风电场的参数特性有准确的定位。其次,针对大型风电场采集数据的不完整和坏点问题,通过风机实际功率特性曲线对数据进行剔除并采用相关系数矩阵法对数据进行填补。针对噪声等因素导致采集的风速数据产生毛刺和尖峰现象,采用了新型粒子滤波器对风场风速数据进行滤波处理,消除风速的毛刺并使得数据得到平滑性处理,并将处理后的数据作为预测模型的输入数据。接着,针对大型风电场中单机功率预测模型的参数选取问题,通过改进人工鱼群算法对模型参数进行优化选取。对鱼群算法中固定的视野和步长产生的局限性问题,本文通过增加自适应调节因子来自动调节鱼群在觅食和追尾行为中的视野和步长,解决了鱼群寻优速度慢及易陷入局部最小问题,通过不同测试函数实验验证改进的算法有较好的寻优效果,最后建立了改进鱼群优化支持向量机的风电场单机的功率预测模型并对两个典型的风场的风机进行功率预测研究。最后,针对大型风电场多台风机预测不稳定以及传统预测方法精度不高等缺陷问题,本文采用了一种基于风速分布特征采样互相关的风电场分组的功率预测策略,将此策略结合改进鱼群优化支持向量机建立了大型风电场分组的智能优化功率预测模型,通过陆地和近海两个典型风场实例仿真验证预测模型的应用效果,并结合实习项目设计了一套风电场功率预测系统软件,对提出的方法进行了工程性验证。
[Abstract]:With the rapid growth of the installed capacity of wind power in China, the proportion of wind power in the power network is increasing, and large-scale wind power grid connection brings serious impact on the safe operation of power system. Effective wind farm power prediction can provide reference for power grid stable operation and dispatching. Aiming at the problem of low precision and instability of traditional power prediction in large-scale wind farms, this paper presents an intelligent optimized power prediction model for grouping large wind farms. The specific research contents are as follows: firstly, The parameter characteristics of large scale wind farm are analyzed and regular statistics are made. The characteristics of wind speed and wind direction of large scale wind farm are studied. The relationship between wind speed, wind direction, temperature and generation power is analyzed. The parameter characteristics of large scale wind farm are accurately located. Secondly, aiming at the problem of incomplete and bad points of data collected by large-scale wind farms, the actual power characteristic curve of fan is used to eliminate the data and the correlation coefficient matrix method is used to fill the data. Aiming at the phenomenon of burrs and spikes caused by noise and other factors, a new particle filter is used to filter the wind speed data in the wind field to eliminate the burr of the wind speed and to smooth the data. The processed data is taken as the input data of the prediction model. Then, aiming at the parameter selection of single-machine power prediction model in large-scale wind farm, the parameters of the model are optimized by improved artificial fish swarm algorithm. For the limitation of fixed visual field and step size in fish swarm algorithm, this paper automatically adjusts the visual field and step size of fish herd in foraging and rear-end behavior by adding adaptive adjustment factor. The problem of slow searching speed and easy to fall into local minimum is solved, and the improved algorithm is proved to be effective by different test function experiments. Finally, the power prediction model of wind farm with improved fish swarm optimization support vector machine is established, and the power prediction of two typical wind field fans is studied. Finally, in view of the instability of multi-typhoon prediction in large-scale wind farms and the low precision of traditional forecasting methods, a power prediction strategy of wind farm grouping based on wind speed distribution characteristic sampling and cross-correlation is adopted in this paper. This strategy is combined with improved fish swarm optimization support vector machine to establish the intelligent optimal power prediction model for large wind farm grouping. Two typical wind field examples on land and offshore are simulated to verify the application effect of the prediction model. A set of wind farm power prediction system software is designed, and the proposed method is verified by engineering.
【学位授予单位】：上海电机学院
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM614

【参考文献】