基于遗传算法的配电网孤岛划分方法研究

发布时间：2018-08-16 14:09

【摘要】：出于对安全性,经济性和环境保护等诸多方面因素的综合考量,分布式电源的广泛接入是当今和未来很长一段时间内配电网发展的一个主要趋势。配电网中数量逐渐增多,容量逐渐增大的分布式电源给现阶段配电网络带来了深刻的变革,在电能质量,继电保护,故障恢复等方面的研究工作都有待于进一步深入和扩展。配网自愈功能建设是构建智能电网不可或缺的环节,而孤岛划分是自愈功能建设的主要技术之一,同时也是一个亟待解决的难题。本文针对这一问题,借助图论中加权连通图等相关概念对含分布式电源的配电网进行了拓扑抽象,以供电恢复量最大为目标建立了孤岛划分模型,并提出云自适应遗传算法和改进量子遗传算法进行上述模型的求解。云自适应遗传算法将云理论与遗传算法进行了有效结合,对交叉和变异算子的生成方法做出了改进。以一系列具有稳定倾向的点取代常规的隶属曲线,既能够使交叉和变异概率随适应度值的增大而迅速降低,又能够保持进化过程中的不确定性。在改进量子遗传算法中,采用了量子比特,旋转门,概率幅等量子力学中的概念进行算法设计。采用概率幅对染色体进行编码,通过旋转门操作使概率幅朝目前搜索到的最优值方向进行进化,并随着更优值的发现调整搜索方向。通过对算法的改进,避免了观测操作对种群上下代之间个体连续性的不利影响,使算法更加适用于孤岛划分这种存在拓扑关联的多重树背包问题。采用经过改进的含多个分布式电源的美国PGE69节点系统对提出的方法进行了验证,结果证明了两种算法的有效性。通过比较分析指出,云自适应遗传算法和改进量子遗传算法代表了进行遗传算法设计的两类不同思路:云自适应遗传算法的本质是个体之间的信息交换,通过影响交叉和变异算子来控制种群的活跃程度和多样性,交叉和变异本身具有一定的盲目性;相比之下,改进量子遗传算法则具有了明显的方向性,通过旋转门操作使每个个体朝向当前的最优值进行旋转,在旋转过程中不断发现更优值。初始种群的选取对改进量子遗传算法的搜索结果有较大影响,因此必须对种群的初始化和进化过程进行干预,好的人工经验将显著改善搜索结果。两种算法都能够充分发挥联络开关的作用,将更多更优秀的可行解纳入到搜索范围中,相比于目前不考虑联络开关的划分方案,能够增大故障恢复的灵活性并得到更好的恢复方案。
[Abstract]:Due to the comprehensive consideration of many factors such as security, economy and environmental protection, the wide access of distributed power generation is a major trend of distribution network development in a long period of time. The quantity of distribution network is increasing and the capacity of distributed generation is increasing gradually. The research work on power quality, relay protection, fault recovery and so on need to be further deepened and expanded. The construction of self-healing function of distribution network is an indispensable link in the construction of smart grid, and islanding is one of the main technologies of self-healing function construction, and it is also a difficult problem to be solved urgently. In order to solve this problem, this paper abstracts the topology of distribution network with distributed generation by means of the concept of weighted connected graph in graph theory, and establishes the islanding model aiming at the maximum recovery of power supply. Cloud adaptive genetic algorithm and improved quantum genetic algorithm are proposed to solve the above model. Cloud adaptive genetic algorithm combines cloud theory with genetic algorithm effectively and improves the generation method of crossover and mutation operators. Replacing the conventional membership curve with a series of points with stable tendency can not only reduce the probability of crossover and variation rapidly with the increase of fitness, but also keep the uncertainty in the evolution process. In the improved quantum genetic algorithm (QGA), the concepts of quantum bit, rotary gate and probability amplitude are used to design the algorithm. The probabilistic amplitudes are used to encode the chromosomes, and the probabilistic amplitudes are evolved towards the optimal values currently searched through the rotary gate operation, and the search direction is adjusted with the discovery of the better values. Through the improvement of the algorithm, the adverse effect of observation operation on the individual continuity between generations is avoided, which makes the algorithm more suitable for the multi-tree knapsack problem of island partition with topological correlation. The proposed method is validated by an improved American PGE69 node system with multiple distributed power sources. The results show that the two algorithms are effective. Through comparative analysis, it is pointed out that the cloud adaptive genetic algorithm and the improved quantum genetic algorithm represent two different ideas for genetic algorithm design: the essence of cloud adaptive genetic algorithm is the exchange of information between individuals. By influencing crossover and mutation operators to control population activity and diversity, crossover and mutation have certain blindness; in contrast, the improved quantum genetic algorithm has obvious directionality. Each individual is rotated towards the current optimal value through the rotary gate operation, and the better value is constantly found during the rotation. The selection of initial population has a great influence on the search results of improved quantum genetic algorithm, so it is necessary to intervene in the initialization and evolution of the population, and good artificial experience will improve the search results significantly. Both algorithms can give full play to the role of the contact switch, and bring more and more excellent feasible solutions into the search scope, compared with the current division scheme that does not consider the contact switch. It can increase the flexibility of fault recovery and get a better recovery scheme.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM73

【相似文献】