大型风电场群运行特性与优化控制研究

发布时间：2018-03-24 11:33

本文选题：风电场群　切入点：马尔可夫链　出处：《华北电力大学(北京)》2017年博士论文

【摘要】：新世纪以来,全球经济快速增长,同时也带来了能源需求的快速增长。伴随着化石能源的日渐衰竭,大力开发利用新能源已经成为当今能源革命的主题。而风力发电是全球发展最为迅速的新能源发电形式,已经实现了连续十年装机容量20%左右的快速增长。然而,由于风电自身的特点,波动性和随机性大,不可控的问题严重,并网和消纳正逐步成为制约风电开发的最主要问题。由于我国国情所限,无论是风资源条件还是系统调峰能力都与欧美等国差距很大。“弃风”压力格外明显。在这样的背景下,大型风电场群需要提高自身的控制水平,使大型风电场群向“电网友好型”电源过渡。围绕这一目标,本文在风电数据预处理、大型风电场群平滑效应、大型风电场群聚类模型、短期风功率预测和大型风电场群优化控制等方面进行了研究,主要研究工作如下:在数据挖掘的过程中,数据预处理具有重要意义。本文基于马尔可夫链理论,建立了双向缺失点补充模型,对风电数据进行缺失点补充。仿真结果表明,基于高阶马尔可夫链的双向缺失点补充模型,具有很高的精度,可以满足数据预处理的要求。通过对大型风电场群运行数据的分析,定量分析了平滑效应在大型风电场群中空间尺度上的表现,获得了如下两条性质:在风电场群内部,风电出力波动随空间范围的变大而下降;距离较远的区域之间的组合能更好的平抑波动,具有更强的平滑效应。基于虚拟发电厂的理论和模糊C聚类算法,建立了大型风电场群聚类模型,将整个风电场群化作多台虚拟风力发电机,在虚拟风力发电机内部,遵循“同调等值”的原理,进行整体调度和控制。鉴于风功率信号在同一频率上具有更为接近的性质和表现,选择最优小波包变换作为信号分析手段。本文的短期风功率预测模型采用了最优小波包变换与最小二乘支持向量机结合的方式。通过实际风功率数据验证,结果表明加入最优小波包变换的预测方法,其预测精度有了明显的提升。在风机有功功率模型分析的基础上,建立了虚拟风机全程可调策略。在风况允许的条件下,小范围调节时通过发电机转矩,大范围调节时通过桨距角和转速联合调节,实现对风电机组功率全程可调,进而实现虚拟风机整体的全程可调。最后基于短期风功率预测的结果和虚拟风机整体的全程可调,提出了一种基于粒子群优化算法的大型风电场群优化控制策略,使得大型风电场群向“电网友好型”电源过渡。
[Abstract]:Since the new century, the rapid growth of the global economy has also brought about a rapid growth in energy demand. The development and utilization of new energy has become the theme of the energy revolution today. Wind power is the most rapidly developing new energy generation form in the world. It has achieved a rapid growth of about 20 percent of installed capacity for ten consecutive years. However, Due to the characteristics of wind power itself, the problems of high volatility and randomness, and uncontrollable problems, grid connection and absorption are gradually becoming the most important problems restricting the development of wind power. Both the wind resource conditions and the peak shaving capacity of the system are very different from those in Europe and the United States and other countries. The pressure of "abandoning wind" is particularly obvious. In this context, large wind farms need to improve their control level. In this paper, the wind power data preprocessing, the large wind farm group smoothing effect, the large scale wind farm cluster model, the wind farm cluster model, the wind power data pretreatment, the large wind farm group smoothing effect, the large scale wind farm cluster model, Short-term wind power prediction and large-scale wind farm group optimization control are studied. The main research work is as follows: in the process of data mining, data preprocessing is of great significance. A bi-directional missing point supplement model is established to complement the wind power data. The simulation results show that the model based on high order Markov chain has a high accuracy. By analyzing the operation data of large wind farm group, we quantitatively analyze the performance of smoothing effect on spatial scale in large wind farm cluster, and obtain the following two properties: inside the wind farm group, The wind power output fluctuation decreases with the increase of the spatial range, and the combination between the distant regions can better suppress the fluctuation and has a stronger smoothing effect. Based on the theory of virtual power plant and fuzzy C clustering algorithm, The cluster model of large scale wind farm is established. The whole wind farm is grouped into several virtual wind turbines. In the virtual wind turbine, the principle of "homology equivalence" is followed. Overall scheduling and control. Given the closer nature and performance of wind power signals at the same frequency, The optimal wavelet packet transform is chosen as the signal analysis method. The short-term wind power prediction model in this paper adopts the combination of the optimal wavelet packet transform and the least squares support vector machine. The results show that the prediction accuracy of the optimal wavelet packet transform is improved obviously. Based on the analysis of the active power model of the fan, the full-range adjustable strategy of the virtual fan is established. When the generator torque is adjusted in a small range and the pitch angle and rotational speed are combined to adjust in a large range, the power of wind turbine can be adjusted in the whole process. Finally, based on the result of short-term wind power prediction and the whole range adjustable of virtual wind turbine, an optimal control strategy based on particle swarm optimization (PSO) algorithm for large wind farm group is proposed. Make large wind farm group to "grid friendly" power supply transition.
【学位授予单位】：华北电力大学(北京)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TM614

【参考文献】