基于键合图的电力电子电路故障诊断研究

发布时间：2019-05-11 20:44

【摘要】：随着电力电子技术的发展,电力变换装置在工业应用中所占据的地位越来越重,一旦发生故障而不能及时诊断与恢复,会带来巨大的经济损失。为保证电力变换装置安全可靠运行,研究电力电子电路的故障诊断方法具有重要的意义。本文以键合图理论为基础,研究基于键合图模型下的定量、定性及支持向量机的故障诊断方法,主要内容有:(1)研究基于定量键合图的故障诊断方法。首先根据系统的键合图模型,建立解析冗余关系与故障特征矩阵。然后分析传感器输出信号得到残差趋势,将残差时域响应与阈值比较生成系统的二进制一致性向量,并利用故障特征矩阵实现系统的故障检测与隔离。针对固定阈值和参数故障不可隔离的问题分别采用区间估计法与粒子群算法。最后以闭环控制的Buck电路驱动直流电机为例,对该方法进行验证,实验结果表明该法是正确有效的。(2)研究基于定性键合图的故障诊断方法。首先讨论了键合图元件的因果关系、系统时间因果图的生成、故障树的建立及定位故障的过程。然后以汽车牵引系统为机械部分的机电系统为例,建立系统的键合图模型,并分析各键合图元件的因果关系生成时间因果图,在此基础上推导得到故障树,采用逆向逐层推理法,定位出引起故障发生的故障源。最后通过仿真和物理实验对其进行验证,实验结果分析表明该方法能有效识别故障位置。(3)研究组合型变流器的故障诊断。首先对支持向量机与近似熵的理论做了简要的介绍。然后以AC/DC-DC/AC变流器为例,分析逆变电路典型的4种故障情形下的电流信号输出。最后对变流器故障情况仿真,采用经验模态分解法对三相输出电流信号进行分解,计算分解得到IMF的能量熵输入支持向量机,同时对支持向量机的参数采用遗传算法优化,结果显示该法对于变流器的五类故障问题具有良好诊断效果。
[Abstract]:With the development of power electronics technology, the power conversion device occupies more and more important position in the industrial application. Once the fault occurs and can not be diagnosed and restored in time, it will bring huge economic losses. In order to ensure the safe and reliable operation of power conversion device, it is of great significance to study the fault diagnosis method of power electronic circuit. Based on the theory of bond graph, this paper studies the fault diagnosis methods of quantitative, qualitative and support vector machine based on bond graph model. The main contents are as follows: (1) the fault diagnosis method based on quantitative bond graph is studied. Firstly, according to the bond graph model of the system, the analytical redundancy relation and fault characteristic matrix are established. Then the residual trend of the sensor output signal is analyzed, the residual time domain response and the threshold are compared to generate the binary consistency vector of the system, and the fault detection and isolation of the system is realized by using the fault characteristic matrix. Interval estimation method and particle swarm optimization algorithm are used to solve the problem of non-isolation of fixed threshold and parameter faults, respectively. Finally, taking the closed-loop control Buck circuit-driven DC motor as an example, the experimental results show that the method is correct and effective. (2) the fault diagnosis method based on qualitative bond graph is studied. Firstly, the causality of bond graph elements, the generation of system time causality diagram, the establishment of fault tree and the process of fault location are discussed. Then, taking the mechanical and electrical system of automobile traction system as an example, the bond graph model of the system is established, and the causality generation time causality diagram of each bond graph element is analyzed, on the basis of which the fault tree is derived. The reverse layer-by-layer reasoning method is used to locate the fault source. Finally, the simulation and physical experiments are used to verify the fault location. The experimental results show that the method can effectively identify the fault location. (3) the fault diagnosis of the combined converter is studied. Firstly, the theory of support vector machine and approximate entropy is briefly introduced. Then, taking AC/DC-DC/AC converter as an example, the output of current signal in four typical fault cases of inverter circuit is analyzed. Finally, the fault situation of converter is simulated, and the three-phase output current signal is decomposed by empirical mode decomposition method, and the energy entropy input support vector machine of IMF is obtained by calculation and decomposition. at the same time, the parameters of support vector machine are optimized by genetic algorithm. The results show that this method has a good diagnosis effect for five kinds of fault problems of converters.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN710

【参考文献】