【作者】 高毅；

【导师】 王成山；

【作者基本信息】 天津大学 ， 电力系统及其自动化， 2011， 博士

【摘要】 现代电力系统的发展使电力系统动态过程的多时间尺度特性愈发突出,这给电力系统仿真计算带来了挑战。本文对交-直流系统多速率稳定性仿真的相关问题进行了研究,主要研究内容包括:高压直流输电系统的稳定性建模和仿真,基于变量自适应分组的多速率仿真算法,十字链表稀疏矩阵技术。各部分内容的具体描述如下:归纳了稳定性仿真程序中高压直流输电系统的建模方式,主要可分为简单模型、响应模型和详细模型,探讨了直流系统稳定分析模型及其所包含的3种模型的适用性。分析了针对直流系统详细模型的交-直流双时步仿真算法的缺陷:该法假设在直流小步长仿真过程中换流母线的交流电压不变,这一假设可能延迟直流系统中一些具有开关特性控制器的动作时间,造成仿真结果不准确;对于多馈入直流系统,甚至可能引起直流开关特性控制的误动作,造成仿真结果失真。所提出的考虑直流系统开关特性控制的变步长仿真方法能正确计及直流开关特性控制的动作时间,从而获得准确的仿真结果,且很好地保持了传统双时步方法的计算效率。提出了一种基于变量自适应分组的多速率仿真方法,能够在仿真过程中自适应地给出具有较高加速比的变量分组方式,避免了传统的多速率仿真方法为进行变量分组而需要预先给定分组步长比的困难。构建了不诚实牛顿法(very dishonest Newton method, VDHN)的分组求解算法,包括预测、插值和校正三个环节,并针对VDHN法的特点给出了校正过程中电压分组计算的方法。算例证明本文多速率法能有效地提高传统电力系统和分布式发电供能微网系统的仿真计算效率。阐述了十字链表的构成以及检索和操作方式,说明十字链表能够高效地完成三角分解和前代-回代运算。指出了传统的十字链表内存分配方式对计算效率的影响:十字链表的存储结点通常散列于内存空间中,该内存分配方式下稀疏矩阵的运算效率低于存储结点连续分布在内存中的稀疏矩阵运算效率,这一现象是由计算机的高速缓冲存储器(cache)的工作原理造成的。针对这一问题提出了一种改进的十字链表方法。通过算例证明了本文的改进十字链表方法是一种高效的稀疏矩阵技术,既适用于传统的大规模电力系统仿真,也适用于分布式发电供能系统仿真。更多还原

【Abstract】 Due to the development of modern power system, the multi-time scale characteristic of power system dynamics has become much more prominent than before, which has brought a great challenge to the power system simulation technique. This thesis focuses on multi-rate simulation method for AC-DC power system stability analysis, including three topics: modeling and simulation method for stability analysis of HVDC, the multi-rate simulation method based on self-adaptive grouping of variables, cross chain table based sparse matrix technique. The work is detailed as follow:HVDC models commonly used in stability simulation program, including simple model, response model and detailed model, are summarized. The applicability of the HVDC models for stability analysis is analyzed. The drawback of the AC-DC system two-time step simulation method for HVDC detailed model is discussed: the assumption of the AC commutating voltages being unchanged during the fast HVDC simulation may delay the action of some HVDC controls with switching characteristics, resulting in errors in simulation results. And it may even cause misoperation of the HVDC switching controls, which is the case in the study of multi-infeed HVDC system. A variable-step simulation method considering HVDC switching controls is proposed, which can make the HVDC switching controls act in time during the simulation. The proposed method can obtain a better accuracy than the conventional one, and the calculation efficiency is not eroded.A multi-rate simulation method based on self-adaptive grouping of variables is proposed for power system stability analysis. A high speed-up ratio is obtained by the self-adaptive grouping algorithm of variables, and there is no need to decide the grouping step ratio in advance which is the case in the conventional multi-rate simulation method. The very dishonest Newton (VDHN) method is adjusted for solving the grouped variables, including three steps: prediction, interpolation and correction. According to the characteristic of the VDHN method, a method for solving the grouped bus-voltage variables is presented which is a step of the correction process. Case study shows that the proposed multi-rate method can be well applied in both traditional large scale power system simulation and distributed generation system simulation.The structure and operation manner of cross chain table are presented to illustrate that its convenience and efficiency in serching, adding and deleting matrix elements can help speeding up the calculation process of factorization and forward-backward substitution. The impacts of the memory allocation manner of cross chain table on the calculation efficiency are discussed: nodes of cross chain table are always stored seperately in memory because of its memory allocation manner, which can reduce the calculation efficiency, and this phenomenon is caused by the principle of cache. To solve this problem, an improved cross chain table method is proposed. Case study proves the efficiency of the improved cross chain table, and shows that it can be well applied in both traditional large scale power system simulation and distributed generation system simulation.更多还原