考虑配电网电压无功资源时空耦合特性的协调优化控制研究

发布时间：2018-01-15 04:04

本文关键词：考虑配电网电压无功资源时空耦合特性的协调优化控制研究　出处：《中国农业大学》2017年博士论文　论文类型：学位论文

【摘要】：随着电力负荷需求的快速增长,以及分布式电源(Distributed Generation,DG)的大规模接入,传统的配电网规划方法与运行控制策略难以满足用户日益增长的电压质量要求以及电力系统运行的安全、稳定和经济性需要。在此背景下,本文围绕配电网的电压无功时空解耦控制方法,对变电站无功优化配置、DG接入后的配电网无功优化运行及具体的求解算法进行研究,主要工作包括:(1)提出一种考虑无功需求曲线时序性特征的变电站并联补偿电容器优化分组方法。该方法首先以段内距与段间距的比值最小为目标建立曲线分段的数学模型,对无功需求曲线分段;然后,提出一种改进模糊C均值聚类算法对曲线的分段结果聚类,根据聚类结果即可确定电容器分组方案,分组数等于聚类数,分组容量等于相邻聚类中心的差值。最后,采用九区图无功控制方法,综合考虑电压质量、网络损耗以及电容器组动作次数等技术指标,对比不同电容器分组方案在系统运行中的优化效果。(2)针对大规模DG接入后的配电网电压无功控制问题,提出考虑配电网时空耦合特性的协调控制架构。基于常见DG的电压无功特性以及不同类型电压无功设备的控制特性分析,提出配电网电压无功两级协调控制模式实现配电网空间上的解耦,两个控制层级通过对关口电压和功率因数的不断修正实现电压的最优分布和无功功率的就地平衡,进而达到全网的最优控制;提出多时间尺度的控制方法实现配电网电压无功控制在时间上的解耦,通过日前计划、短时调度和实时控制这三个时间尺度的滚动修正来逐级降低负荷与DG出力预测误差对系统控制的影响,实现系统的最优控制。(3)提出一种基于最优分割一联合优化的配电网日前无功计划方法,以解决DG与有载调压变压器分接头(On-load Tap Changer, OLTC)以及电容器组的协调优化问题。该方法通过对OLTC和电容器组的静态最优出力序列进行最优分割,确定设备的动作时刻,实现了 OLTC档位及电容器组投切容量等离散型控制变量在时间上的解耦。考虑不同设备的电压无功调节特性,提出以静态优化结果为基础,先确定OLTC档位,再优化电容器组动作时刻,最后通过联合优化求解电容器组与DG最优电压无功出力的整体思路,通过OLTC、电容器组和DG三者的协调配合实现全网的最优控制。以某地实际85节点配电网为例,建立仿真模型,验证了本方法的正确性和有效性。(4)针对现有无功优化算法存在的不易收敛、求解速度较慢等问题,提出一种适用于配电网实时无功优化的熵权优选法,该方法包括负荷分布匹配和熵权法两个步骤。首先,基于无功优化方案与数据源的关系分析,提出一种基于典型场景划分和拉依达准则的负荷分布匹配方法,可以快速、准确地从大量历史数据中选取若干个备选方案;然后,考虑网损、电压偏移量和静态电压稳定性等多个指标,对各备选方案进行评价,采用熵权法将多属性决策转化为单属性决策问题,进而选出最优方案。并且,针对实际运行中历史数据不完善或历史方案优化效果较差等问题,提出一种混合算法,将熵权优选法与常规优化方法结合使用,以保证计算结果的最优性。最后,在IEEE 33节点系统、PGE 69节点系统以及某地实际173节点配电网和292节点配电网等多个标准算例和实际算例系统中进行仿真验证,并通过与遗传算法、序列二次规划算法等现有方法的对比,证明了所提方法的正确性和有效性。
[Abstract]:With the rapid growth of power load demand, and distributed power supply (Distributed Generation, DG) of the large-scale access, distribution network planning and operation method of the traditional control strategy is difficult to satisfy the requirements of users increasing voltage quality and power system security, stability and economy. Under this background, this paper focuses on the voltage distribution network work time decoupling control method, reactive power optimization of substation, distribution network with DG after reactive power optimization operation and specific algorithm are studied, the main work includes: (1) proposed a reactive demand curve time sequence features optimization grouping method for substation capacitor. Firstly, by segment from the ratio of the minimum section spacing mathematical model of target curve segment, the reactive power demand curve; then, we put forward an improved fuzzy C means clustering The results of the clustering algorithm piecewise curve, according to the clustering results can determine the capacitor grouping scheme, packet number is equal to the number of clusters, the packet capacity is equal to the difference between adjacent cluster centers. Finally, the nine area reactive power control method, considering the quality of voltage, network loss and capacitor action number and other technical indicators, the optimization effect comparison of different capacitor grouping scheme in the operation of the system. (2) according to the distribution voltage and reactive power after the large-scale DG access control problem, put forward the coordinated control architecture spatial and temporal characteristics of the distribution network. Considering the control characteristic of common DG voltage and reactive power characteristics and different types of voltage and reactive power equipment based on the proposed voltage distribution network reactive two coordinated control mode to realize decoupling of the spatial distribution network on the two level of pass through control of voltage and power factor correction to achieve the optimal constant voltage The cloth and the reactive power balance on the spot, and then achieve the optimal network control; decoupling control of reactive power at the time of the proposed control method of multi time scale implementation of voltage distribution network, the day before the program, rolling short-term scheduling and real-time control of the three time scale correction to progressively reduce the load and output forecast error DG the impact on the control system, realize the optimal control of the system. (3) proposed a distribution network on a joint optimal segmentation method based on reactive power planning before, in order to solve the DG and transformer taps (On-load Tap, Changer, OLTC) and the coordination and optimization of the capacitor. The output method the optimal segmentation by static optimal sequence of OLTC and capacitor group, determine the action time of the equipment, the OLTC shift and the capacitor switching capacity of the discrete control variables at the time of the consideration is not decoupled. The same equipment voltage and reactive power regulation characteristics, based on static optimization based on the results, to determine the OLTC file, and then optimize the capacitor action time, finally through the joint optimization of capacitor group and DG optimal voltage reactive power output of the whole idea, through the OLTC, coordinate the capacitors and the three with the realization of DG the whole network optimal control. Taking a practical 85 bus distribution network as an example, a simulation model is established to verify the correctness and validity of this method. (4) according to the existing reactive power optimization algorithms are not convergence, problem solving speed, this paper puts forward a suitable real-time distribution network reactive power optimization entropy optimization method, the method includes matching load distribution and entropy method in two steps. First, analysis of the relationship between the reactive power optimization scheme and data source based on a matching method of typical scene classification and Pauta criterion based on the load distribution, can Fast, accurate selection of a number of alternatives from a large number of historical data; then, considering network loss, voltage deviation and voltage stability and other indicators, to evaluate various options, using entropy method to multiple attribute decision making is transformed into a single attribute decision problem, and select the optimal solution. And, according to the the problems in the actual operation of historical data is not perfect or historical optimization effect is poor, this paper proposes a hybrid algorithm combining the entropy optimization method and conventional optimization methods, the optimality of the calculated result. Finally, in the IEEE 33 node system, PGE 69 bus system and a practical 173 bus distribution network and 292 node distribution network as a standard example and numerical simulation system, and through the comparison with the genetic algorithm, the sequence of two quadratic programming algorithm and other existing methods, to prove the validity of the proposed method and Efficiency.

【学位授予单位】：中国农业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TM714.3;TM732

【相似文献】