粒子群和声搜索混合算法及其在水库长期优化调度中的应用研究

发布时间：2018-04-08 21:55

本文选题：水库　切入点：长期优化调度　出处：《昆明理工大学》2016年硕士论文

【摘要】：随着我国水利工程建设的推进,水电站水库优化调度成为了水利系统长期经济运行需解决的核心问题。运用优化调度技术,有效、合理地对水库水电站进行调度研究,提高水资源可持续利用,使水库的综合效益最大化具有重要意义。受诸多因素的影响,水库的调度模型发展成具为有多目标、多极值、高维度、非线性等特性的复杂优化问题,用传统的优化方法难以解决。随着计算机技术的迅速发展,众多智能优化算法层出不穷,它们打破了传统优化问题采用精确模型进行计算的模式,非常适合用来处理传统方法难以解决的复杂问题。和声搜索算法作为一种新兴高效的启发式全局搜索算法,它通过模拟音乐创作原理来求解优化问题,具有结构简单、容易实现、跳出局部最优等优点,但是算法的收敛性和收敛速度都有待提高。本文通过研究和声算法的寻优机理,与粒子群算法的特点相结合,提出了粒子群和声搜索混合算法,并应用在小湾电站的长期优化调度问题中。本文阐述了研究水库调度的背景和意义以及国内外研究进展,并介绍了水库优化调度常用的模型方法,分析不同模型方法的优缺点。接着对和声搜索算法做了深入的研究,针对该算法收敛性差和收敛速度慢的不足,结合粒子群算法的优点对算法进行改进,提出了粒子群和声搜索混合算法(PSOHS)。然后根据小湾水库30年径流资料,以发电量最大为目标函数建立长期优化调度模型,分别采用常规算法和提出的PSOHS优化算法对模型进行求解。常规算法以月为时段,按等流量调节进行计算,得出多年平均发电量为186.5736亿kW·h。在运用PSOHS优化算法时,为简化计算,先以年为时段进行多年调节计算,再对典型年进行年调节计算,用来修正多年调节计算的结果,得出多年平均发电量为201.3864亿kW·h,相比常规计算结果,效益提高了7.36%。为更好的分析比较优化的效果,在30年的时间序列中选取1967-1980年时段,针对枯水期和丰水期的发电量比较时段优化效益,枯水期(1967-1977年)优化效益为7.08%,丰水期(1978-1980年)优化效益为-2.72%,充分发挥出水库的蓄丰补枯的调节作用；整个时段内发电量提高了5.25%,体现了算法有很好的优化效果。实例应用结果表明,粒子群和声搜索混合算法在水库长期优化调度中的应用是有效、可行的。
[Abstract]:With the development of water conservancy construction in China, the optimal operation of hydropower stations and reservoirs has become the core problem in the long-term economic operation of water conservancy system.It is of great significance to study the operation of reservoir hydropower stations effectively and reasonably by using optimal dispatching technology to improve the sustainable utilization of water resources and maximize the comprehensive benefits of reservoirs.Influenced by many factors, the reservoir operation model has developed into a complex optimization problem with multi-objective, multi-extremum, high-dimensional, nonlinear and other characteristics, which is difficult to solve by traditional optimization methods.With the rapid development of computer technology, many intelligent optimization algorithms emerge in endlessly. They break the traditional optimization problem using accurate model to calculate, and are very suitable for dealing with complex problems which are difficult to solve by traditional methods.As a new and efficient heuristic global search algorithm, harmony search algorithm solves the optimization problem by simulating the principle of music creation. It has the advantages of simple structure, easy realization, jumping out of local optimum and so on.However, the convergence and convergence rate of the algorithm need to be improved.By studying the optimization mechanism of the harmony algorithm and combining the characteristics of the particle swarm optimization, a hybrid algorithm of particle swarm optimization and acoustic search is proposed in this paper, and it is applied to the long-term optimal scheduling problem of Xiaowan Hydropower Station.In this paper, the background and significance of reservoir operation research and the research progress at home and abroad are described. The common model methods for reservoir optimal operation are introduced, and the advantages and disadvantages of different model methods are analyzed.Then, the harmonic search algorithm is deeply studied. Considering the shortcomings of poor convergence and slow convergence speed, the PSO algorithm is improved by combining the advantages of PSO, and a hybrid PSO HSG algorithm is proposed.Based on the 30-year runoff data of Xiaowan Reservoir, a long-term optimal dispatching model is established with the maximum power generation as the objective function. The conventional algorithm and the proposed PSOHS optimization algorithm are used to solve the model respectively.The conventional algorithm takes the month as the time period and calculates according to the constant flow regulation. The average annual power generation is 18.65736 billion kW / h.In the application of PSOHS optimization algorithm, in order to simplify the calculation, the multi-year adjustment calculation is carried out in the period of year first, and then the annual adjustment calculation is carried out for the typical year, which is used to modify the results of the calculation of multi-year regulation.The average power generation for many years is 20.13864 billion kW / h. Compared with the conventional calculation results, the efficiency is increased by 7.36%.In order to better analyze and compare the effect of optimization, the 1967-1980 period is selected in the 30-year time series.The optimization benefit is -2.72 in the dry season and 1978-1980 in the high water period, and the electricity generation has been increased by 5.25 in the whole period, which shows that the algorithm has a good optimization effect. The optimum benefit is -2.72 in the dry season and 1978-1980 in the high water period, and the regulating function of the reservoir is brought into full play, and the electricity output increases by 5.25% in the whole period of time, which shows that the algorithm has a very good optimization effect.The application of particle swarm optimization and acoustic search hybrid algorithm in reservoir long-term optimal operation is proved to be effective and feasible.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TV697.11

【相似文献】