【作者】 杨靖萍；

【导师】 徐政；

【作者基本信息】 浙江大学 ， 电力系统及其自动化， 2007， 博士

【摘要】 我国电网正处于高速发展时期，随着特高压输电工程的展开，不远的将来将形成全国性的大规模特高压互联电网。随之会出现一系列的重大技术问题，如稳定性、动态安全评估和协调控制等问题。直接对如此大规模的电网进行详细分析是很难进行的，因此本文利用电力系统仿真软件PSS／E，重点研究了在新环境下大规模互联电力系统的动态等值方法。主要内容包括：(1)基于电力系统仿真软件PSS／E研究了电力系统同调动态等值原理，其中同调机聚合采用了保留发电机方程系数矩阵结构不变的时域聚合方法。该方法不同于传统的在频域中拟合传递函数的聚合方法，它是在时域中直接求取等值机参数，不需要迭代优化。利用PSS／E中的IPLAN语言编制了程序，程序可以直接在PSS／E中使用。并将该方法应用于实际大电网—华东电网，对华东电网进行动态等值简化。通过对华东电网的原型系统和等值系统的动态仿真分析和比较，结果表明建立的动态等值系统能够很好地反映原型系统的动态特性。(2)提出了一种基于物理等效的适用于电力系统电磁暂态仿真的动态等值方法，该方法根据等值后主网架母线短路容量保持不变的原理对系统进行动态等值。利用该方法对实际电网—江苏电网进行动态等值简化，通过对原型系统和等值系统的动态仿真分析比较，表明等值效果良好，能够满足工程分析要求。利用等值系统分析了江苏电网中交流系统故障或直流系统系统故障时系统的稳定性，由仿真得到的动态响应特性曲线表明无论是交流系统故障或直流系统故障，江苏电网都能保持稳定。(3)提出了一种新的基于全球定位系统(GPS)的同步相量测量单元(PMU)的动态等值方法，该方法根据同步相量测量单元测得的同步动态数据来等值简化外部系统，等值系统的参数利用参数辨识技术得到。该方法的优点1)PMU能够进行相量测量，量测数据同步且精度高，保证了量测量的准确性和可靠性。2)不需要外部系统的拓扑结构、参数及运行状态等详细信息。通过原型系统和等值系统仿真分析和模态分析比较，表明等值系统能够较好地再现原型系统的主要动态性能。(4)提出了一种改进的含变异算子的两粒子群(PSO)优化辨识算法来估计等值系统参数。该算法构造了两个独立的粒子群，通过对各粒子群分别设置不同的速度上限，来使两个粒子群以不同的步长在搜索空间中并行寻优。为充分利用两个粒子群的最优信息，在各粒子群的进化过程中引入另一粒子群的全局最优值，从而实现两粒子群间的信息交互和共享。并在此基础上设计了一个特殊的变异算子来保证寻优过程中种群的多样性。该算法具有跳出局部极值的能力，提高粒子的搜索优化效率。更多还原

【Abstract】 Power grids are developing quickly in China. Large-scale national power grid will be formed in the future with the progress of Ultra High Voltage (UHV) transmission project. However, some forthcoming technical problems, such as stability, dynamic security assessment and coordination control et al., will emerge. It is difficult to analyze such a large-scale power system thoroughly. So, this dissertation focuses on the dynamic equivalents methods for large-scale power systems using the power system simulation software PSS/E. The main results are organized as follow:(1) The theory of coherency equivalents is studied under power system simulation software PSS/E, in which the coherent generators are aggregated based on structure preservation of the coefficient matrices in time domain. The aggregation method obtains directly the parameters of the equivalent generator without iterations unlike traditional aggregation method which needs iterations to optimize the parameters by means of fitting the transfer function in frequency domain. The program in IPLAN language in PSS/E can be implemented directly in the system. The aggregation method was used to reduce East China power grid. The original system and the equivalent system are simulated respectively. The comparison shows that the equivalent system can preserve the main dynamic characteristics of the original system well.(2) A practical method for power system electromagnetic analysis is presented based on physical equivalence, which matches a short-circuit current after the reduction of the power system to that of the original system. It was applied to reduce Jiangsu power grid. Simulation on the original system and the reduced system were performed and compared. Results show that the reduced system can maintain the essential dynamic behavior of the original with good accuracy. The stability of Jiangsu power grid was studied in the reduced system when disturbance in AC system or DC system occurred. Results indicate that Jiangsu power grid is stable whether AC system or DC system is disturbed.(3) A new dynamic equivalent method based on PMU is proposed, which reduces the external system on the basis of the synchronous dynamic measurements obtained by PMU and parameter identification. There are the following advantages: (a) PMU can provide the high-precision phasor measurement synchronously insuring the accuracy and reliability of the measurements. (b) It does not need the detailed information about the configuration, parameters and operating state of the external system. Results of simulation and mode analysis on the original system and the reduced system show that the reduced system can resemble the main dynamic characteristic of the original system well.(4) An improved two PSO algorithm with mutation operator was presented to optimize and identify the parameter of the equivalent system. The algorithm constructs two populations separately, and makes them seek the best position with different steps in the problem space by setting different maximum velocities. In order to share optimal information between two populations, the global optimum of one population is used by the other in the evolution process. Furthermore, a mutation operator is introduced to assures the diversity of the individual in the evolution process. The algorithm can escape from the local optimum and enhance the search ability of the particles.更多还原