【作者】 井实；

【导师】 黄琦；

【作者基本信息】 电子科技大学 ， 检测技术与自动化装置， 2013， 博士

【摘要】 随着智能电网推广建设，智能化开关、电子式互感器、合并单元等新型设备被引入变电站。这些设备的使用有助于提高变电站内部信息组织、分配的灵活性。但是，从另一方面来看，智能变电站整个一次、二次系统的实现方式和应用方式也发生了较大的变化。传统以电缆为媒介传输电气信号的方式，被光纤数字化方式所取代。信息也不再是单一的点对点通信方式，可通过构建新型的通信网络加以传播。同时，智能变电站中单个设备与其他设备之间的关联性也更加紧密，部分新型保护装置需要依据多设备的输出信息进行判断。这些复杂的信息组织关系和综合保护、控制功能如何检验，是智能变电站投运前的检测工作必须解决的问题。以往，智能变电站二次系统的测试方法主要针对单体设备的功能检验问题，在现场测试中，难以涉及设备关联性检查和网络行为检查。因此，测试工作缺乏完整性，不利于电网安全和稳定运行。本文从系统级测试的角度出发，研究二次系统测试方法及其关键技术，为提升智能变电站安全运行水平和现场调试工作的效率提供解决方案。主要内容有以下几点：1）针对智能变电站二次系统的组成结构以及网络特性，在分析了二次系统测试应涵盖的内容以及技术需求的基础上，提出了一种新型智能变电站全场景测试方法。论文详细阐述了该方法的测试原理、测试流程以及测试结果的判别方式，探讨了方法实现的可行性，以及需要解决的各项关键技术。2）针对全场景测试方法的分布式特点和变电站现场所需的同步性能指标，结合变电站现场环境因素，研究一种适用于智能变电站全场景测试的无线IEEE1588时钟同步解决方案。论文在分析影响其精度各因素的基础上，通过改变对时方式以及基于平均速率进行修正的方法，研究了一种基于WLAN的无线1588对时方法，结合具体实现，对比研究了改进方法的性能。同时，通过构建无线电磁辐射传递模型，定量分析了无线同步过程中电磁辐射对二次设备的影响。3）针对全场景测试方法快速电磁暂态计算的需求，在分析仿真过程和计算速度影响因素的基础上，提出两种适用于现场全场景测试系统加速仿真的方法：基于CUDA的并行化计算方法和基于FPGA的硬件计算方法。论文基于因子表路径树的矩阵分解方法，将求解过程中的导纳矩阵转化为对角加边形式，将计算任务分配到CUDA提供的独立计算硬件资源，实现了单步长电磁暂态仿真中线性方程组求解的分解并行化计算；论文设计了构建电磁暂态元件以及故障的逻辑电路表达形式的方法和FPGA硬件计算方法，实现了基于隐式积分方法的电磁暂态实时计算；结合实际变电站测试实例，对两种计算方法的加速性能和计算精度进行了评估。4）为了进一步节省计算资源，论文基于频率相关网络等值的思想，探讨了在较大规模电网的电磁暂态仿真中，采用矢量拟合法进行等值电路计算的改进方法及其实现步骤。针对该方法存在的不足，论文引入混合矩阵法，研究计及三相耦合因素的电磁暂态仿真中的等值电路计算方法。并针对实际电路的等值计算分析，比较了矢量拟合法和混合矩阵法在稳态计算和暂态计算中的性能，评估了混合矩阵法在全场景测试中的适用性。更多还原

【Abstract】 As the development of smart grid, the concept of smart substation is proposed, inwhich some novel types of devices such as smart switch, electronic transformer,merging unit (MU) etc, are introduced. These devices can greatly increase the flexibilityin the organization and distribution of the information. However, the implementationand application of the whole primary and secondary systems are changed greatly. Thetraditional signal transmission by electrical cable is replaced by digital fiber.Information is not transmitted through point to point communication, but through novelcommunication network. Simultaneously, the interrelationship among different devicesin smart substation is closer. Some of the novel types of protection devices need tojudge based on the outputs from multiple devices. The utility engineers must solve theproblems such as how to verify the complex information distribution relationship andcheck the functionalities of comprehensive protection and control, before putting thesmart substation into operation.In the past, the test of secondary system in smart substation mainly involves withthe function verification of single devices. This approach is not able to check thenetwork behavior and correlationship of devices in field test. Therefore, the test lacks ofintegrity, threatening the security and stability power system operation. In thisdessertation, we study the test approach and its key technologies of secondary system ina smart substation from a systematic view. It is expected that the research would providesolution for enhancing the security level of the operation of smart substation and testefficiency. The main research contents and innovative results include:1. By analyzing the structure and network characteristics of the secondary systemin smart substation, a novel whole-view test approach is proposed based on the analysisof the test contents and requirements. The principle, test procedure, and processing oftest results are described in detail. The feasibility of the approach and the keytechnologies to implement are discussed.2. According to time synchronization performance requirement of substation fieldtest, a scheme of wireless time synchronization by IEEE1588and WLAN technology,suitable for distributed whole-view test, is proposed. The impact of field is alsoconsidered. Based on analyzing factors affecting accuracy of synchronization, an improved method based on wireless IEEE1588is proposed by changing the timingmethod and time correction based on average transmission rate of the network. Thesystem is implemented and the performance is comparatively studied. A model ofwireless radiation is built, and the effect of radiation of2.4GHz wireless communicationon secondary device is quantitatively evaluated.3. As the whole-view test needs fast electromagnetic transient simulation, twomethods for accelerating computation are proposed in the dissertation, based on theanalysis of the factors affecting the simulation process and computation speed. InCUDA based parallel computing method, the solution of conductance matrix istransformed into bordered diagonal form by using the matrix decomposition based onpath tree of factor table. The computer tasks are distrubted to independent computingresources provided by CUDA, accelerating the solution of linear equations in theprocess of electromagnetic transient simulation process. In FPGA based hardwarecomputing method, the method of contruction of logic circuit of electromagnetictransient components and faults is designed, which implemented the real-timecomputing of electromagnetic transient simulation based on implicit integral. Withactual examples, the performance of acceleration and computation accuracy is assessed.4. In order to further save computing resources, the improvement and realization ofnetwork equivalence based on VF (Vector Fitting) in large-scale electromagnetictransient simulation are explored in the dissertation, based the Frequency DependentNetwork Equivalence (FDNE) approach. To overcome the shortcomings of VF, MixingMatrix method is introduced. And a new method for network equivalence forelectromagnetic transient simulation considering three-phase coupling is discussed. Bysimulation with actual circuits, the performance of steady-state and transientcomputation with VF method and Mixing Matrix method is comparatively studied. Theapplicability of Mixing Matrix method in the whole-view test is evaluated.更多还原