含大规模风电的多源区域电网优化调度研究

发布时间：2018-11-07 18:03

【摘要】：化石能源发电所引起的环境污染问题已经成为制约国家能源可持续发展战略的一大障碍,利用无污染、可再生的新能源代替化石能源发电,是未来电力发展趋势之一。风电作为新能源发电中的一种,具有清洁、储存量大和易于开发等优点,被广泛开发和利用。由于风电的随机不确定性,大规模风电的接入,给电力系统稳定运行带来了一定的挑战。因此,研究含大规模风电接入的电力系统动态特性和风功率预测以及多源区域电网的优化调度,对提高风电的开发利用具有重要意义。本文对含大规模风电接入的多源区域电网优化调度问题,展开了如下研究:(1)构建了含风力发电机组、水力发电机组和汽轮发电机组的多源混合电力系统模型,在风速波动条件下,对该系统模型进行了仿真分析。仿真结果表明,所建的多源混合电力系统的稳定性受风电机组输出功率波动性的影响,并能够准确描述该电力系统主要参数的动态特性,为进一步研究含大规模风电接入的多源区域电网优化调度研究提供支撑。(2)提出了基于粒子群神经网络(Particle Swarm Optimization and Back-propagation Neural Network,PSO-BP)的风电功率预测方法,该方法利用粒子群算法的全局搜索能力来获得BP(Back-propagation,BP)神经网络的初始权值和阈值,很好地解决了常规BP算法收敛速度慢、易陷入局部极小等问题,并对PSO-BP算法和BP神经网络算法的预测结果进行了对比分析。根据实例预测结果表明,PSO-BP算法较BP神经网络算法预测的绝对平均误差(Mean Absolute Error,MAE)和均方根误差(Root Mean Square Error,RMSE)分别减少了7.02%,和9.37%,证明粒子群神经网络(PSO-BP)算法在风电场输出功率预测方面具较理想的效果。(3)基于所建立的含风电的多源混合电力系统模型和风电功率预测的基础上,研究了基于多智能体粒子群算法(Multi-agent and Particle Swarm Optimization,MA-PSO)的经济调度方法,该算法结合了粒子群(Particle Swarm Optimization,PSO)算法全局特性和多智能体系统(Multi-agent System,MAS)的智能特性,有效解决了高维数、非线性、多参数耦合的经济调度问题;通过对MA-PSO算法与基本PSO算法优化结果进行对比分析,MA-PSO算法求出的一天的最优值的发电成本为3.7964×10~4$,而PSO算法所求出的最优值的发电成本为4.1787×10~4$。MA-PSO算法所求发电成本较PSO算法节省了3.823×10~3$,即节省率高达9.14%。证明MA-PSO算法搜索性能好,收敛精度高。同时,MA-PSO算法应用于解决经济调度问题,能够获得较好的经济效益和环境效益。
[Abstract]:The problem of environmental pollution caused by fossil energy power generation has become a major obstacle to the national energy sustainable development strategy. The use of non-polluting renewable new energy to replace fossil energy power generation is one of the future power development trends. Wind power, as one of the new energy generation, has the advantages of clean, large storage and easy to develop, so it has been widely developed and used. Because of the random uncertainty of wind power and the connection of large-scale wind power, it brings some challenges to the stable operation of power system. Therefore, it is of great significance to study the dynamic characteristics and wind power prediction of power system with large-scale wind power access, as well as the optimal dispatching of multi-source regional power network, in order to improve the development and utilization of wind power. In this paper, the optimal dispatching problem of multi-source regional power network with large-scale wind power access is studied as follows: (1) the model of multi-source hybrid power system with wind turbine generator, hydrogenerator and turbine generator is constructed. Under the condition of wind speed fluctuation, the system model is simulated and analyzed. The simulation results show that the stability of the multi-source hybrid power system is affected by the fluctuation of the output power of the wind turbine, and the dynamic characteristics of the main parameters of the power system can be accurately described. It provides support for further research on optimal dispatching of multi-source regional power network with large-scale wind power access. (2) A wind power prediction method based on particle swarm optimization neural network (Particle Swarm Optimization and Back-propagation Neural Network,PSO-BP) is proposed. This method utilizes the global searching ability of particle swarm optimization algorithm to obtain the initial weights and thresholds of BP (Back-propagation,BP) neural network, which solves the problems of slow convergence speed and easy to fall into local minima of conventional BP algorithm. The prediction results of PSO-BP algorithm and BP neural network algorithm are compared and analyzed. The prediction results show that the absolute mean error (Mean Absolute Error,MAE) and root mean square error (Root Mean Square Error,RMSE) of the PSO-BP algorithm are 7.02 and 9.37 less than those of the BP neural network algorithm, respectively. It is proved that the particle swarm optimization neural network (PSO-BP) algorithm is effective in predicting the output power of wind farm. (3) based on the model of multi-source hybrid power system with wind power and the prediction of wind power, The economic scheduling method based on multi-agent particle swarm optimization (Multi-agent and Particle Swarm Optimization,MA-PSO) is studied. The algorithm combines the global characteristics of particle swarm optimization (Particle Swarm Optimization,PSO) algorithm and multi-agent system (Multi-agent System,). The intelligent characteristic of MAS effectively solves the economic scheduling problem with high dimension, nonlinear and multi-parameter coupling. By comparing and analyzing the optimization results of MA-PSO algorithm and basic PSO algorithm, it is found that the optimal value of MA-PSO algorithm is 3.7964 脳 10 ~ (4) 脳 10 ~ (4) / day. The optimal value of the PSO algorithm is 4.1787 脳 10~4$.MA-PSO, and the cost is 3.823 脳 10 ~ (-3) less than that of the PSO algorithm, that is, the saving rate is as high as 9.14%. It is proved that MA-PSO algorithm has good searching performance and high convergence accuracy. At the same time, the MA-PSO algorithm is applied to solve the economic scheduling problem, which can obtain better economic and environmental benefits.
【学位授予单位】：华北水利水电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TM73

【参考文献】