基于多Agent的断路器故障诊断系统设计

【作者】 甄建聚；

【导师】 孙来军；

【作者基本信息】 黑龙江大学 ， 检测技术与自动化装置， 2012， 硕士

【摘要】 作为高压电器中一种重要的设备，高压断路器在电力系统中肩负着控制和保护双重任务，其自身的健康状况与运行的可靠性直接关系到电力系统的安全与稳定。因此，为了降低高压断路器的故障率，有效提高高压断路器的可靠性，本文设计了断路器状态监测与故障诊断系统，并着重从以下四个方面对该系统进行了研究。首先，针对传统故障诊断系统存在的缺点，本文将多Agent技术应用于断路器的故障诊断当中。并对多Agent故障诊断系统的任务分解、组织结构以及通信与协作机制进行了研究。其次，研究了特征提取Agent提取振动信号特征参数的方法。将经验模态分解和信息熵理论相结合提出了一种新的特征提取方法。测试与仿真结果表明，利用该方法提取的特征向量可以正确地反映高压断路器状态的变化，为断路器的故障诊断提供了新方法。再次，研究了诊断Agent进行断路器故障诊断所使用的诊断策略。在实际应用中，由于断路器不可能频繁动作，所以无法获得大量的故障样本进行训练。支持向量机是一种基于结构风险最小化的学习机器，为小样本学习分类问题提供了一条有效的解决途径。同时本文又对多诊断Agent诊断结果的模糊融合原理进行了介绍与应用分析。最后，根据目前国内电力设备管理与运行模式以及断路器状态监测与故障诊断的特殊性，设计了一种基于多Agent的高压断路器状态监测与故障诊断系统，并开发了基于该设计的硬件设备和上位机管理软件。试验结果表明，该系统可以将大型数据库、研究机构专家、设备维护人员和现场监测仪有效地结合起来，实现了对多方诊断资源的有效利用，提高了断路器诊断系统诊断的准确率和效率。更多还原

【Abstract】 As an important device in high-voltage electrical apparatus, high-voltage circuitbreakers play the dual role of control and protection in the power system. The healthstatus and operating reliability of circuit breakers are directly related to the security andstability of power system. Therefore, the condition monitoring and fault diagnosticsystem was designed in order to reduce the fault rate of high-voltage circuit breakersand effectively improve the reliability of high-voltage circuit breakers. The system wasstudied from the following four aspects.Firstly, to overcome the shortcomings of traditional systems of fault diagnosis, themulti-agent technology was applied to fault diagnosis of circuit breakers in this paper.The paper have made research on the task decomposition, organization structure, themechanism of communication and coordination which are involved in the systems offault diagnosis based on multi-agent.Secondly, the paper have studied the extract method of characteristic parametersabout vibration signals which were used by characteristic-extracting agents. A newmethod of feature extraction was proposed by associating the empirical modedecomposition with the information entropy theory. The results of simulation and testshow that the extracted characteristic parameters can sensitively reflect changes ofhigh-voltage circuit breakers status with this method. It provides a new method for faultdiagnosis of circuit breakers.Thirdly, fault diagnostic strategy of diagnosis agents about high-voltage circuitbreakers was studied. In practical applications, on account of the circuit breakers cannot act frequently, a large numbers of fault samples can not be obtained for training. Thesupport vector machine is a learning machine based on the structural risk minimization,which provides an effective solution for the classification problems of small sample learning. At the same time, introduction and application analysis about fuzzy fusion ofagents’ diagnosis results were performed.Finally, according to the management and operation mode of currently domesticelectrical equipment and the particularity of the condition monitoring and faultdiagnosis of circuit breakers, a condition monitoring and fault diagnosis system ofcircuit breakers based on multi-agent was designed. At the same time, the hardware andmanagement software of computer based on the design were developed. The test resultshows that the system can effectively unite the large-scale database, the experts ofinstitutes, the equipment maintainer and field monitor to realize the effective utilizationof multi-diagnosis resources. Consequently, the accuracy and efficiency of circuitbreakers fault diagnosis were improved.更多还原