直觉模糊粒子群算法在发电机组启动策略中的应用

发布时间：2018-04-17 10:39

  本文选题：直觉模糊集 + 粒子群算法　； 参考：《电子科技大学》2014年硕士论文

【摘要】：大停电事故后的恢复控制问题是现代电力系统安全防御的一个重要课题。其中,机组启动是整个恢复控制的基础。机组启动的关键是如何在恢复过程中根据系统的实际状况进行机组启动的优化。因此,如何在恢复控制过程中合理选择被启动机组是机组启动优化策略研究的核心问题。由于求解机组启动策略本身属于优化问题,所有文章的主要研究都是建立在如何解决优化问题的基础上,侧重点为粒子群算法研究。所以在这一前提下,文章提出应用粒子群算法(PSO)及直觉模糊集理论(IFS)来解决发电机组启动最优策略问题,并对发电机组的启动问题进行了相关的分析与研究。论文主要工作与成果如下:第一,分析了算法的运行机制及其相关改进,并对算法的收敛性进行了相关的推导,研究算法存在的缺陷和及其产生的原因。确定了对PSO算法改进的可能性及改进的方向。第二,对大停电事故后的恢复控制的核心机组启动问题进行了深入分析,在已有的机组启动的优化策略研究中,通常以运行经验为基础,在目标函数和求解方法上都存在需要完善的地方。文章在分析与发电机组启动策略密切相关有关因素后,给出发电机组启动目标函数和相关约束条件,建立了发电机组启动问题的0-1规划模型,提出了分时间周期进行启动的办法来解决含时间约束条件的机组启动问题。第三,针对PSO算法运行中种群多样性难以测定的问题,分析了算法运行过程中粒子的运动状态,结合IFS理论提出了直觉模糊种群熵(IFPE)作为种群多样性的测度,并证明了IFPE比其他常用多样性测度的优越性。第四,充分研究了离散二进制粒子群算法(BDPSO)的求解原理,通过实验手段在分析了可能影响的求解的因素及IFPE在求解过程中的相应变化。在此基础上,提出了两种基于IFPE的离散粒子群算法(IFDPSO)及其衍生型。将这些算法和原始DPSO进行了对比研究和实验,发现IFDPSO系列算法更适合解决0-1背包问题。第五,在研究发电机组启动特点及可能影响发电机组启动因素的基础上,编写基于IFDPSO的机组启动策略决策支持系统,并在此平台上验证了发电机组最有启动策略,并与常规启动策略、不同初始功率、不同优化时间段的启动策略对比分析,从而验证算法的有效性。
[Abstract]:The problem of recovery control after blackout is an important issue in modern power system security defense.Among them, the unit start-up is the basis of the whole recovery control.The key of unit start-up is how to optimize the unit start-up according to the actual condition of the system during the recovery process.Therefore, how to select the start-up unit reasonably in the recovery control process is the core problem in the research of the start-up optimization strategy of the unit.Because solving the start-up strategy of generating unit is an optimization problem, the main research of all papers is based on how to solve the optimization problem, with emphasis on particle swarm optimization algorithm.In this paper, PSO (particle swarm optimization) and IFS (intuitionistic fuzzy set theory) are proposed to solve the problem of optimal starting strategy of generator set, and the related analysis and research on the start-up problem of generator set are also given in this paper.The main work and achievements are as follows: first, the mechanism of the algorithm and its related improvements are analyzed, and the convergence of the algorithm is derived, the defects of the algorithm and its causes are studied.The possibility and direction of improving PSO algorithm are determined.Secondly, the start-up problem of the core unit after blackout is analyzed deeply. In the research of the optimization strategy of the existing unit start-up, it is usually based on the operation experience.Both objective function and solution method need to be improved.Based on the analysis of the factors closely related to the start-up strategy of the generator set, the objective function and the relevant constraints are given, and the 0-1 programming model of the start-up problem of the generator unit is established.In this paper, a method of time cycle start-up is proposed to solve the problem of unit start-up with time constraints.Thirdly, aiming at the problem that population diversity is difficult to measure in the operation of PSO algorithm, the motion state of particles is analyzed, and the intuitionistic fuzzy population entropy is proposed as the measure of population diversity combined with IFS theory.It is proved that IFPE is superior to other commonly used diversity measures.Fourthly, the principle of discrete binary Particle Swarm Optimization (Dbinary Particle Swarm Optimization) algorithm is fully studied. The possible factors affecting the solution and the corresponding variation of IFPE in the process of solution are analyzed by means of experiments.On this basis, two discrete particle swarm optimization algorithms based on IFPE and their derivation are proposed.By comparing these algorithms with the original DPSO, it is found that the IFDPSO series algorithms are more suitable to solve the 0-1 knapsack problem.Fifthly, on the basis of studying the characteristics of generator set startup and the factors that may influence the start of generator set, the decision support system of unit start-up strategy based on IFDPSO is developed, and the most effective startup strategy of generator set is verified on this platform.The algorithm is compared with conventional startup strategy, different initial power, and different optimization time period to verify the effectiveness of the algorithm.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM31
，

本文编号：1763252