【作者】 马丽丽；

【导师】 罗艳红；

【作者基本信息】 东北大学 ， 电力系统及其自动化， 2015， 硕士

【摘要】 随着传统能源的日趋枯竭和能源危机、发电带来的环境污染等问题日益严重,如果能在充分利用现有的电力网络的同时,以相对少的网络损耗和投资成本提供更大容量的高质量的电能。对网络进行重构能发挥电网现有设备的潜能,降低运营成本、提高网络运行的经济性和供电的可靠性。当分布式电源接入配网对配电网的潮流、电压、网损都将产生巨大的影响。含分布式电源的配网重构问题是典型的非线性优化问题,具有多约束、多变量、离散性等特点,在利用目前的优化方法求解时存在很多问题,例如DG的处理方法影响计算的时间和精度,传统优化算法易陷入局部最优或很难找到最优解,优化算法的求解时间长、精度低等。因此我们有必要研究新的方法或新的改进策略。以差分进化算法为代表的进化智能算法是求解配网重构这类复杂非线性优化问题的有力工具,但差分进化算法是一种相对新的基于进化的优化技术,在理论分析和应用研究等方面还处于初级阶段,有很多问题值得研究,例如如何提升算法跳出局部最优解的能力,如何提升算法求解复杂多峰值问题的精度和速度。本文首先建立各种DG的等效负荷模型和处理方法;针对具有二阶收敛特性的牛顿法对初值敏感的问题,采用基于种群进化状态动态调整参数的N-EAPSO潮流计算方法。并以IEEE33节点系统和加入DG的系统验证本文算法的性能。然后,本文针对DE算法优化停止在局部极值处和收敛速度慢等问题,使用PIADE算法。通过基于个体优势调整变异因子策略、保留精英思想与混沌遍历思想对差分进化算法进行改进,在优化过程中动态调整种群数量和变异因子,使得算法能快速有效的收敛,跳出局部最优的能力得到增强,并通过算例验证了算法的有效性。最后,针对差分进化算法的“贪婪”选择机制导致算法过早收敛,引入植物生长算法与改进的差分进化算法相融合,通过使用由形态素浓度所决定的随机性和方向性平衡较为理想的搜索机制,平衡算法局部搜索和全局搜索性,增强算法的全局收敛能力,并通过算例验证了算法的有效性。更多还原

【Abstract】 As the traditional energy depleted,power generation and energy crisis brought serious environmental pollution problems,making full use of existing power network of measures,with relatively less network loss and investment cost to provide larger capacity of the high quality of electric energy,should be taken.Reconfiguration of power distribution network can play to the potential of existing power grid equipment,reduce operating costs,improve the reliability and efficiency of the network operation and the power supply.Connected to distribution network of distributed power impact on the trend of the distribution network,the node voltage and network losses.The reconfiguration of power distribution network containing distributed generation is a typical nonlinear optimization problem with multiple constraints,multivariate,discrete and other characteristics.There are many problems in using the current optimization methods.for example,DG treatments affect the time and precision of computation;the traditional optimization algorithms are easy to fall into local optimum or difficult to find global optimal solution,optimization algorithms have a long time of computation,and optimize the accuracy of the results is low.Therefore,we need to study new methods or new improvable strategies.Represented by differential evolution algorithm is the evolution of intelligent algorithm,which is a powerful tool to solve the reconfiguration of distribution network for this type of complex nonlinear optimization problem.But differential evolution algorithm is a relatively new technique based on evolutionary optimization,which theoretical analysis and applied research are still in its infancy and have many issues worthy of studing:such as how to enhance the capacity of algorithm jump local optima,how to improve the accuracy and speed of high-dimensional complex multimodal problems.Firstly,the equivalent load model and the treatment methods of various kinds of DG is established.As to Newton’s method with quadratic convergence properties of sensitivity initial value,the paper proposed N-EAPSO flow calculation method with dynamically adjust the distance between the particles based on parameters.Aiming at the problem that the weight and the learning factor is more dependent on the particle swarm optimization,the dynamic adjustment strategy of the weight and the learning factor is adopted.The effectiveness of the algorithm was verified.by IEEE33 bus system and join the DG of IEEE33 node system.Secondly,for solving problems that differential evolution algorithm into local optimum and slow convergence,the PIADE algorithm is used in this thesis.Adaptive mutation factor strategy based on individual advantage,holding elite thoughts and thoughts on chaos are used to improve performance of differential evolution algorithm.Dynamic adjustment in the process of optimization of population and the mutation factor,makes the algorithm can quickly effective convergence and jump out of local optimal solution,and the effectiveness of the algorithm was verified by an example.Finally,as to the problem that differential evolution algorithm of "greedy" selection mechanism easily lead to premature convergence of the algorithm,by using the determined by the form element concentration in randomness and directional balanced ideal search mechanism,this thesis combine plants growth simulation algorithm to the improved differential evolution algorithm to further expand the search scope and enhance the global convergence performance of the algorithm.更多还原