Research On Fault Diagnosis Of Power Electronic Circuits Based On Bond Graph

Power conversion devices have been playing an increasingly significant role in industrial applications with the development of power electronic technology. Huge economic losses will be caused once the devices fail to get timely diagnosis and recovery when failure happens.It is of great significance, therefore, to research on the fault diagnosis of the power electronic circuits and its method to ensure the power conversion devices to work safely in a reliable way. In this paper, researches on fault diagnosis are operated with the fault diagnosis method of the quantitative bond graph, qualitative bond graph and support vector machine based on the theory of bond graph, which are introduced briefly bellow:(1)Fault diagnosis method based on quantitative bond graph. Firstly, according to the system bond graph model, the Analytical Redundancy Relations and Fault Signature Matrix are given. Then the binary coherence vector of the circuit is obtained through a comparison between the threshold values and the time domain response of the residual after the analysis of sensor output signal to get the trend of the residual error. The fault detection and isolation of the circuit are operated through the Fault Signature Matrix. Considering the issue of the fixed thresholds and the parametric can not fault isolation, approaches of interval estimation and Particle Swarm Optimization algorithm are proposed. In the last, the example the closed-loop control of Buck DC motor drive circuit is presented, the experimental results show that the approach is of correctness and validity.(2)Fault diagnosis method based on qualitative Bond Graph. What is firstly given are the fundamental principles of the causal analysis of the Bond Graph components, the Temporal Causal Graph generation, the construction of fault tree and the steps of fault location. Then a electrical and mechanical system of the mechanical parts for a car traction system is taken as an example, in which the system's Bond Graph model and also the Temporal Causal Graph are presented by the analysis of each system's BG component. On the basis upon theory the fault is built, in which the fault source is located with the method of Back propagation. In the last, the results of physical experiment have verified the effectiveness of the method in fault location.(3)Fault Diagnosis of combination-type converter. Firstly, a brief theory of support vector machine and approximate entropy is introducted. Then the AC/DC-DC/AC converter is taken as an example to the analysis of the current signal output of the inverter circuit about four kinds of typical fault situations. The simulation of converter failure is considered in the last, in which a three-phase output current signal is decomposed using the Empirical Mode Decomposition method. The entropy energy of Intrinsic Mode Function is calculated as the input of support vector machine. Meanwhile the support vector machine parameters are optimized using genetic algorithm. The results show that the method performs very well in solving the five fault problems of converter.