【作者】 余炳辉；

【导师】 张勇传；

【作者基本信息】 华中科技大学 ， 水利水电工程， 2007， 博士

【摘要】 I粒子群优化算法是一种新的智能优化算法,由于其具有编程简单,适应性强,较好的全局优化能力等特点得到了广泛的关注。同时也大量的应用到了电力系统的优化中。如何进一步提高粒子群优化算法的性能和设计出针对具体系统的有效算法是当前研究的重点。本文主要的任务就是通过试验的方法来探索粒子群优化算法的优化机理,在此基础上提出新的优化算法设计思路,以期设计出新的更有效的算法来。本文的具体安排如下:论文首先综合评述了水火电力系统短期优化调度的研究现状,重点回顾了现代启发式优化算法在该系统中的应用。然后对粒子群优化算法的产生,发展,取得的成果和存在的不足进行了综合评述。在此基础上确立了论文的研究方向,并给出了论文的内容安排。首先简单的介绍了基本粒子群优化算法的产生,原理,在此基础上,给出了粒子群优化算法的实现技术及具体步骤。同时还重点介绍了粒子群优化算法的4种典型形式。同时为了进一步改善算法的性能提出了一种改进的粒子群优化算法——随机摄动粒子群优化算法。对几个典型的函数进行优化,验证了算法的有效性。在此基础上是通过试验来探究粒子群优化算法的工作机理。首先根据在优化过程中,速度更新公式的不同表现定义了典型情形的概念,并给出了4种典型情形的具体形式。在此基础上,对各种典型情形下算法的动态特性和静态特性进行了试验研究,以探究粒子群优化算法的工作原理。同时对于不同的参数组合对优化性能的影响进行了试验研究,得到了一些有用的结论。最后做了几个探索性试验,为提出新的算法提供了思路。通过在系统试验过程中出现的现象对基本的粒子群优化算法进行了扩展,提出3种由粒子群优化算法衍生的新算法。1通过模拟每次迭代后粒子的分布规律,提出了模拟粒子群优化算法;2通过去掉速度更新公式中的惯性项,而加上最差极点,提出了最优最差粒子群优化算法;3通过模拟每次整个种群的分布规律,提出了整体分布优化算法。在提出上述算法的同时,还和粒子群优化算法进行了比较,通过比较表明,整体分布算法具有编程简单,鲁棒性好,优化性能强等特点。尝试用整体分布优化算法对一个水火电力系统的短期优化调度进行求解,同时将优化结果与遗传算法,进化规划求得的结果进行比较,验证了整体分布优化算法的有效性,说明整体分布优化算法比较适合于水火电力短期发电计划这类复杂系统的优化。最后,对本文的成果进行总结,并提出有待进一步研究的问题。更多还原

【Abstract】 Particle Swarm Optimization (PSO) is a new intelligent optimization algorithm, programming to be widespread concern due to its simple, adaptable, better global optimization features etc. Particle Swarm Optimization also has a large number of applications to power system optimization. How to further improve the PSO algorithm design and performance of specific system efficient algorithm is currently the focus of the study. The main propuse of this paper is to explore the PSO algorithm optimization mechanism through a large of experimentation. Based on this, new optimization algorithm is design, in order to design new, more effective algorithm. In this paper, the specific arrangements are as follows:The paper reviewed the hydro-thermal power system short-term optimal operation of the status quo, focus on reviewing the modern Heuristic Optimization Algorithm in the system application. Then the PSO algorithm for the birth, development, and the results achieved, and the shortcomings of the Comprehensive are reviewed. Based on this established a research paper, and gives arrangements of the paper.introduceing PSO algorithm for the birth, principle, on the basis of this, the PSO algorithm technology and the realization of concrete steps are given. Also focuses on the PSO algorithm four typical forms. In order to further improve the performance of the algorithm proposed a modified PSO algorithm -- random perturbation PSO Algorithm. Through some of the typical function optimization, the algorithm’s effectiveness is proved.The working mechanism for the PSO algorithm is probed through experimentation study. At first, according to the different performance of the speed of updating the formula in different optimize phase, the typical definition of the concept of circumstances and four typical cases of specific forms are given. On this basis, under typical circumstances of dynamic and static of a pilot study to explore PSO algorithm works. Moreover, different combinations of parameters to optimize the performance of a pilot study to be some useful conclusions. Finally a few exploratory test of the new algorithm is provided.The three new optimization algorithms derived from the basic PSO algorithm are proposed. 1. the simulation PSO algorithm is given through simulate each particle distribution after each iteratives ; 2.The optimal worst PSO algorithm is proposed through removing the formula for the speed of inertia, adding the worst particle; 3. by simulating the entire population distribution, the overall distribution optimization algorithm is given. The overall distribution optimization algorithm is simple, robust and strong performance optimization features among above algorithms through comparing.The overall distribution optimization algorithm is used to solve short-term scheduling for hydro-thermal power systems. The result is compared with the results of genetic algorithm and evolutionary programming for the same system, the validity of the overall distribution optimization algorithm is verificated. This example notes the overall distribution optimization algorithms are more suitable for short-term hydropower electricity generation schemes such complex systems optimization.Finally, the paper summarized the results and made pending further study.更多还原