分布式动态状态估计算法及其在电力系统中的应用

发布时间：2017-12-28 02:32

本文关键词：分布式动态状态估计算法及其在电力系统中的应用　出处：《山东大学》2017年博士论文　论文类型：学位论文

【摘要】：电力系统状态估计是能量管理系统的重要组成部分,其估计精度和可靠性直接影响着电力系统的调度、安全性分析和执行操作任务的准确性.动态状态估计不仅可以对系统状态进行更准确地估计,而且拥有静态状态估计不具备的预测功能,能够为电力系统的经济调度和预防控制等在线功能提供先验信息和更多的操作时间.所以动态状态估计具有非常重要的应用价值,并得到了广泛研究.随着电力系统规模的不断扩大,电网互联程度的日益加强以及精度更高、更新速度更快的量测装置(PMU)的迅速发展和广泛应用,传统的集中式状态估计方法很难满足准确性和实时性的需求,从而促进了电力系统分布式状态估计算法的发展.本文通过将电力系统划分为若干个不重叠的子系统,研究了电力系统的分布式动态状态估计问题,提出的算法适用于大规模电力系统,具有重要的理论意义和应用价值.主要贡献和创新点如下:(1)通过将电力系统分区并基于SCADA和PMU混合量测,提出了分布式动态状态估计算法,使得各子系统可以平行而且相对独立地计算,加快了整体的计算速度.(2)各子系统不需要全局拓扑可观测性信息,只需极少的量测数据和邻居传递的信息即可估计本地状态,降低了分布式动态状态估计算法的计算复杂度,并且该复杂度与网络的大小无关.所提出的算法没有中央调度中心,所以避免了集中式算法在数据传输过程中的瓶颈问题,而且便于实施和管理.(3)所提算法的估计精度比集中式算法稍差,但是优于分布式静态状态估计算法.当系统出现异常情况时,所提算法良好的估计性能和鲁棒性以及参数辨识的有效性得到证实.(4)通过减弱限制条件,证明了算法得到的估计和预测误差协方差矩阵是正定并且有上界的,保证了算法的可行性.按照章节顺序,具体的研究内容和研究成果包括以下几个方面:1.研究了离散时间电力系统线性化模型的分布式动态状态估计问题.基于最大后验估计技术,提出了一种分布式状态估计算法,其中量测数据由SCADA和PMU混合量测系统提供.首先将电力系统划分为若干个不重叠的区域,对应的子系统利用先验信息、本地和边界量测以及邻居子系统传递的信息对本地状态进行估计,而不是估计整个系统的状态.与集中式方法相比,该分布式算法有效地降低了每个子系统状态的维数和计算复杂度.其次,当划分后的子系统组成的网络不含有环时,证明了各子系统在每个时刻的本地状态估计经过有限次迭代收敛于集中式方法在修正目标函数下的估计值.最后,仿真结果验证了所提算法对于大规模电力系统状态估计的有效性和可行性.2.对于非线性电力系统,基于扩展卡尔曼滤波技术,给出了一种分布式动态状态估计算法.当子系统的本地量测无法得到时,证明了所提出的算法仍然是可行的,同时利用边界量测和邻居子系统传递的信息,各子系统能够得到理想的本地状态估计.通过对模型参数进行在线辨识,提高了状态预测的精度,并加强了算法的鲁棒性.当电力系统出现负荷突变、存在不良数据和拓扑结构改变等异常情况时,详细的仿真结果验证了所提算法的鲁棒性、较为准确的估计值和参数辨识对于状态估计的有效作用.与其它算法的性能比较结果显示了所提算法在应用中的优势.3.进一步研究了非线性动态系统的分布式状态估计问题.通过减弱算法所需的限制条件,分析了算法的有界性.基于数学归纳法,证明了各子系统的本地状态估计和预测的误差协方差矩阵正定.根据时变系统的能观测性秩判据,证明了误差协方差矩阵有上界,保证了算法的可行性.利用新的参数辨识方法,有效抑制了负荷突变对于估计精度产生的不良影响.
[Abstract]:Power system state estimation is an important part of energy management system, its accuracy and reliability directly affects the power system scheduling, security analysis and implementation tasks. The accuracy of dynamic state estimation can more accurately estimate the system state, but also have a static state estimation prediction function does not have the power to. The system of economic dispatch and the prevention and control of online function provides a priori information and more operation time. So the dynamic state estimation has very important application value, and has been extensively studied. With the expansion of power system, power grid interconnection degree increasing and higher precision and faster update speed measuring device (PMU) the rapid development and extensive application. The method is very difficult to satisfy the real-time and accuracy requirements of the traditional centralized state estimation, and from To promote the development of the algorithm of distributed state estimation of power system. The power system is divided into several non overlapping sub system, distributed dynamic state estimation problem of power system, the proposed algorithm is suitable for large-scale power system, has important theoretical significance and application value. The main contributions and innovations are as follows: (1) the power system partition and SCADA and PMU mixed measurement based, we propose a distributed dynamic state estimation algorithm is that each subsystem can be parallel and independent calculation, accelerate the overall computation speed. (2) the system does not require global topological observability information with minimal amount the measured data and neighbor message can be estimated local state, reduce the distributed dynamic state estimation algorithm computational complexity, and the complexity and independent of the size of the network is proposed. No central control center, so to avoid the bottleneck problem of centralized algorithm in data transmission process, and is easy to implement and management. (3) the estimation accuracy is slightly worse than the centralized algorithm, but better than the distributed static state estimation algorithm. When the system appears abnormal, the proposed algorithm good estimation performance and the robustness and parameter identification is proved to be effective. (4) by weakening the restrictions, proved that the estimation algorithm is obtained and the prediction error covariance matrix is positive and bounded, ensure the feasibility of the algorithm. According to the order of chapters, the specific research contents and results are as follows: distributed dynamic state 1. on the linear model of discrete time estimation of power system. Based on the maximum a posteriori estimation technique, proposes a distributed state estimation algorithm, the measurement data Provided by the SCADA and PMU mixed measurement system. Firstly, the power system is divided into several non overlapping regions, the corresponding subsystem using the prior information, and the local boundary value estimation of local state measuring subsystem and neighbor information, rather than estimating the state of the whole system. Compared with the centralized method, the distributed the algorithm effectively reduces the dimensionality and computational complexity of each subsystem state. Secondly, when the subsystems of the partitioned network does not contain a ring, that each subsystem of the local state at each time estimation after finite iterations converge to the centralized estimation method in the correction value of the objective function. Finally, simulation the results show that the proposed algorithm for large scale power system state estimation, the validity and feasibility of.2. for nonlinear power system, extended Calman filter technology based on a given A distributed dynamic state estimation algorithm. The local sub system can not be measured, it is shown that the proposed algorithm is feasible, at the same time measurement and information transmission subsystem neighbors using the boundary, the local state of each subsystem can obtain the ideal estimation. Through the online identification of the model parameters, improve the state of the prediction accuracy and enhance the robustness of the algorithm. When the power system load mutation and the existence of bad data and topology change and other abnormal situations, detailed simulation results verify that the proposed robust estimation algorithm, more accurate values of parameter identification and state estimation for effective performance with other algorithms. The comparison results show that the advantages of.3. algorithm in the application of the further study of the distributed state estimation problem of nonlinear dynamic system. The limits required by weakening algorithm And the analysis of the algorithm is bounded. Based on mathematical induction, proved positive definite covariance matrix of each subsystem of the local state estimation and prediction. The observability rank criterion based on time-varying system, proved that the error covariance matrix is bounded, ensure the feasibility of the algorithm. By using the method of parameter identification of the new. To restrain load mutation to the adverse impact of estimation accuracy.
【学位授予单位】：山东大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TM711;TM732

【参考文献】