基于人工蜂群算法的最优潮流相关技术研究

发布时间：2018-03-22 22:24

本文选题：最优潮流　切入点：智能优化算法　出处：《北京交通大学》2016年博士论文　论文类型：学位论文

【摘要】：随着分布式电源的大规模并网以及电动汽车等柔性负荷的快速发展使得现代电力系统变得越来越复杂。最优潮流作为电力系统规划、经济调度和市场交易等方面的分析工具,可以有效地对复杂电力系统的安全性、经济性及稳定性进行综合优化计算,但其本质是一个多约束的、离散连续变量共存的、多维的非线性优化问题,选择合适的求解方法直接决定了最优潮流解的有效性及优越性。人工蜂群算法作为一种新颖的智能优化算法,在处理非线性、多约束、多变量、非连续、非凸等优化问题上具有一定优势,已在人工神经网络训练、图像识别、语音识别等领域得到广泛应用。然而,人工蜂群算法与其它智能优化算法发展类似,在初始研究阶段依然存在一些问题需要解决,例如提高该算法的收敛速度及计算精度。为此,本文以基于人工蜂群算法的最优潮流相关技术研究作为课题,通过对人工蜂群算法的深入研究,为电力系统的单目标／多目标最优潮流问题提供一种新的求解方法,从而为采用最优潮流作为计算工具的电力系统问题提供更加丰富的分析及决策信息。本文的主要研究内容如下：(1)分析了人工蜂群算法寻优时各阶段的数学模型,采用几个标准数值测试函数对人工蜂群算法的优化性能进行了仿真计算,结果表明：人工蜂群算法具有较高的收敛特性,能有效地处理数值优化问题。(2)针对人工蜂群算法在处理低维度优化问题时具有较高的寻优能力,但求解高维度优化问题时易陷入局部最优的缺点,提出了一种混沌差分人工蜂群算法。该改进算法采用差分进化算法的变异、交叉操作代替标准人工蜂群算法新蜜源的搜索操作,以提高算法的局部搜索能力；利用混沌映射中的Tent映射生成算法的初始种群、变异操作中的参考蜜源以及交叉操作中的参考维数,以增加种群的多样性。(3)分析了典型最优潮流的数学模型,分别从经济、环保、电能质量三方面建立最优潮流的目标函数,即总发电成本、有功网损、总污染物排放量及电压偏离量。针对多个目标函数,采用最大模糊满意度法将其模糊处理,形成模糊多目标最优潮流模型。建立了基于混沌差分人工蜂群算法的模糊多目标最优潮流求解模型。仿真结果表明：所建立的求解模型可以有效地、可靠地解决最优潮流问题,并且得到的最优潮流运行方案可以进一步提高系统的经济性、电压水平以及降低对环境的污染程度。(4)为了获得质量更高的Pareto最优前沿,研究并提出了一种改进的多目标人工蜂群算法。该改进算法通过变异和交叉操作获得新可行解,采用快速非支配排序获得各可行解支配信息以及更新外部存档,利用目标函数以及拥挤距离的综合信息来计算可行解被待工蜂选择的概率值,通过计算拥挤距离来实时控制外部存档的大小,利用外部存档中的Pareto最优前沿作为算法寻优时的参考蜜源。在此基础上,建立了基于改进的多目标人工蜂群算法的多目标最优潮流求解模型。该求解模型先采用所提出的算法获得Pareto最优前沿,再利用K均值聚类法对最优前沿进行聚类,最后采用模糊集理论方法进行决策分析。仿真计算表明：所提出的改进的多目标人工蜂群算法在求解Pareto最优前沿时具有有效性、可靠性及优越性；所建立的多目标最优潮流求解模型可以从Pareto最优前沿中选择出更满意、更优异的运行决策方案。(5)针对系统含有风电及负荷不确定因素的最优潮流问题,建立了考虑风电接入及负荷随机变化的多目标概率最优潮流模型。该模型以发电成本的期望值及标准差作为目标函数,将状态变量违反程度的期望值以惩罚项形式加入到目标函数中。针对所建立的模型,提出了两种求解方法。第一种方法：将多目标概率最优潮流采用模糊数学转换成单目标概率最优潮流,再利用基于拉丁超立方采样的改进人工蜂群算法进行求解。第二种方法：直接通过基于拉丁超立方采样的多目标人工蜂群算法获得多目标概率最优潮流的Pareto最优前沿。用改进IEEE30节点测试系统的仿真计算表明：所建立的模型可以有效地处理含有风电及负荷不确定性因素的概率最优潮流问题,同时也验证了所提出的两种方法在求解多目标概率最优潮流问题上的有效性及优越性。
[Abstract]:With the rapid development of large-scale distributed power grid and electric vehicle flexible load makes modern power system becomes more and more complicated. The optimal power flow in power system planning, analysis tools of economic dispatch and market transactions, can be effective for complex power system safety, economy and stability of the integrated optimization, but its essence is a multi constraint, discrete continuous variables, multidimensional nonlinear optimization problem, select the appropriate algorithm directly determines the effectiveness and superiority of the optimal power flow solution. Artificial bee colony algorithm is a novel intelligent optimization algorithm in dealing with nonlinear, multi constraint, multi variable, non continuous. Non convex optimization problem has certain advantages, has image recognition in artificial neural network, and is widely used in the field of speech recognition. However, artificial bee colony algorithm Method and other intelligent optimization algorithm, need to solve some of the problems still exist in the initial stage of research, such as improving the convergence speed and accuracy of the algorithm. Therefore, the optimal power flow based on artificial bee colony algorithm research as the subject, through in-depth study of the artificial bee colony algorithm, provides a new method to solve the problem for single target / multi-objective optimal power flow of power system, so as to provide analysis and decision making more information for the optimal power flow calculation of power system as tools. The main contents of this paper are as follows: (1) analysis of the mathematical model of each stage of artificial bee colony algorithm, using several standard numerical test function the artificial bee colony algorithm optimization performance were simulated. The results show that the convergence characteristics of artificial bee colony algorithm is high, can effectively. Physical and numerical optimization problems. (2) based on artificial bee colony algorithm is high in processing low dimension optimization optimization capabilities, but to solve the high dimension optimization problem is easy to fall in local optima, a chaotic differential artificial bee colony algorithm. The improved algorithm uses deviation algorithm, crossover operator instead of the standard artificial bee colony algorithm new nectar search operation, in order to improve the local search ability of the algorithm; using Tent mapping algorithm to generate the initial population in the chaotic mapping, mutation and cross reference nectar in the reference dimension, in order to increase the diversity of the population. (3) analyzed the typical mathematical model of optimal power flow, respectively from the economic, environmental protection, power quality three aspects to establish the objective function of optimal power flow, the total cost of power generation, power loss, the total pollutant discharge quantity and voltage deviation. For multiple target function The number of the fuzzy satisfaction maximizing method fuzzy processing, fuzzy multi-objective optimal power flow model is established. Based on chaotic differential fuzzy multi-objective optimal power flow solution model of artificial bee colony algorithm. The simulation results show that the model can solve effectively and reliably solve optimal power flow problem, optimal power flow and operation plan can further improve the economic system, the voltage level and reduce the degree of pollution of the environment. (4) in order to obtain the Pareto optimal front for higher quality, research and proposes an improved multi-objective artificial bee colony algorithm. The improved algorithm by mutation and crossover operation to obtain a new feasible solution, the fast non dominated the sort of feasible solution of controlling information and update the external archive, comprehensive information of the objective function and the crowding distance to calculate the feasible solution by Daigong bee selection The probability value by calculating the crowding distance to the real-time control of the external archive size, using the Pareto optimal front external archive as algorithm reference nectar. On this basis, the establishment of multi-objective optimal power flow model to solve multi-objective artificial bee colony algorithm based on improved. The model is solved by using the proposed algorithm to obtain the first Pareto optimal front, then cluster the optimal frontier using K means clustering method, the fuzzy set theory in decision analysis method. The simulation results indicate that the multi-objective improved artificial bee colony algorithm proposed is efficient in solving Pareto optimal front, superiority and reliability; multi-objective optimal power flow solution model you can choose from the Pareto optimal front with more excellent operation decision. (5) for the system with wind power and load the most uncertain factors Optimal power flow problem, a multi objective optimal power flow probability of wind power and the load of model. In this model, the generation cost of expected value and standard deviation as the objective function, the degree of violation of state variable expectations to punish a form to join in the objective function. According to the model, put forward two methods. The first method: the multi-objective probabilistic optimal power flow using fuzzy mathematics into single objective probabilistic optimal power flow, then using the improved artificial bee colony algorithm to solve the Latin hypercube sampling based on second methods: directly through the Pareto optimal front to obtain optimal power flow of multi objective probabilistic multi-objective artificial bee colony algorithm Latin hypercube sampling based on the simulation. The results indicate that the improved IEEE30 node test system: model can effectively deal with the uncertainty of wind power and load The probabilistic optimal power flow problem with qualitative factors is also verified. It also verifies the effectiveness and superiority of the two methods proposed for solving multi-objective probabilistic optimal power flow problem.

【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TM744;TP18

【相似文献】