基于虚拟仪器的电力电子装置在线故障诊断

发布时间：2018-01-31 08:08

本文关键词： 试验传动平台电力电子装置虚拟仪器核极限学习机故障诊断　出处：《西安科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：在航天动力范畴中,电力电子装置通常作为试验传动平台电源供给环节,具有非常重要的作用。其一旦出现故障,轻则设备停机造成经济损失,重则导致人员伤亡。因此课题设计了一套针对试验传动平台的在线监测与故障诊断系统,以确保平台正常运行和故障发生后的及时排查。论文主要从以下几点展开研究:首先,针对原有试验平台信号测点位置分散、传统仪表监测、人工读取记录等弊端,通过现场考察,设计了基于PXI总线的高精度、多线程虚拟仪器的硬件采集系统,利用EPLAN电气软件绘制了采集系统电气接线图,完成了传感器转接箱和测试柜的研制,搭建出系统在线监测硬件平台。其次对于试验传动平台电力电子装置环节出错率较高的问题,利用Simulink软件搭建故障仿真模型。采集不同故障下单周期固定采样点数据,将此数据利用小波包分解重构的方法提取能量作为特征,并将该特征作为极限学习机故障分类输入。但因电力电子电路具有较强的非线性,传统的极限学习机算法得不到较好的分类结果。因此课题引入核函数的思想,利用核极限学习机的方法对其进一步分类测试,经过调整参数可以得到较好的分类效果,从理论的角度说明了此方法的合理性。然后,基于系统在线监测和故障诊断的功能需求,利用LabVIEW图形化编程环境完成软件功能设计。该设计主要包括试验台数据采集、分析显示、存储回放、故障诊断等部分。最后,课题依靠数据采集硬件系统和上位机程序设计,在航天试验现场实现了对传动平台各类信号的在线监测,并完成了现场1#整流装置、2#整流装置、直流电动机励磁模块、同步发电机励磁模块的故障监测与诊断。结果表明,本文所设计的基于虚拟仪器的电力电子装置在线故障诊断系统有较高的灵活性和可操作性,同时缩减了故障排查时间,在工程上具有一定的应用价值。
[Abstract]:In the aerospace power category, power electronic device is usually used as the power supply link of the test transmission platform, which plays a very important role. Once it breaks down, the light equipment shutdown will cause economic losses. Therefore, a set of on-line monitoring and fault diagnosis system for test transmission platform is designed. In order to ensure the normal operation of the platform and timely troubleshooting after the occurrence of fault. The paper mainly from the following points: first, the original test platform signal location scattered, traditional instrument monitoring. Based on the field investigation, the hardware acquisition system of high precision and multi-thread virtual instrument based on PXI bus is designed. The electrical wiring diagram of the acquisition system is drawn by using the EPLAN electric software, and the development of the sensor connecting box and the test cabinet is completed. Build the system on-line monitoring hardware platform. Secondly for the test transmission platform power electronic device link error rate is high. The fault simulation model is built by using Simulink software. The fixed sampling point data of different fault issuing period are collected, and the data is extracted as the feature by wavelet packet decomposition and reconstruction method. This feature is used as the fault classification input of LLM. However, because of the strong nonlinearity of power electronic circuits, the traditional LLM algorithm can not get better classification results. Therefore, the idea of kernel function is introduced in this paper. The method of nuclear extreme learning machine is used to further classify and test it. After adjusting the parameters, a better classification effect can be obtained. The rationality of this method is explained from the theoretical point of view. Then. Based on the functional requirements of on-line monitoring and fault diagnosis, the software function design is completed by using LabVIEW graphical programming environment. The design mainly includes data acquisition, analysis and display, storage and playback. Finally, based on the design of data acquisition hardware system and upper computer program, the on-line monitoring of all kinds of signals of transmission platform is realized in the space test site, and the field 1# rectifier device is completed. Fault monitoring and diagnosis of rectifier device DC motor excitation module and synchronous generator excitation module. The on-line fault diagnosis system of power electronic device based on virtual instrument designed in this paper has high flexibility and maneuverability, at the same time, it reduces the time of troubleshooting and has certain application value in engineering.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM507

【参考文献】