Study On The Fault Diagnosis Methods Of Switched Reluctance Motor System

Due to the advantages of rugged construction, excellent speed adjusting performance and superior fault-tolerant ability, Switched Reluctance Drive (SRD) System has been recognized as a research focus in contemporary electrical drive field. With constant improvement of modern industrial electrification and automation, the reliability requirement of electrical drive systems also becomes higher. Studying on the fault diagnosis and fault-tolerant control methods of SRD, which has certain foundation of fault-tolerant ability, would further enhance its reliability and protrude its fault-tolerant predominance, therefore, this subject has prodigious research and practical value. According to the fault types of SRD system, this dissertation focuses on fault diagnosis methods of inter-turn winding short-circuit fault, power converter faults and position sensors'faults. The main works are generalized as follows.The non-sinusoidal phase current and local saturation of iron core are two notable features of SRM. Based on the computation results of electromagnetic characteristics for a prototype SRM by finite-element method, a flux-linkage model using exponential function, a inductance model using Fourier series function and a Back-Propagation (BP) network model optimized by Genetic Algorithm (GA) are used to set up the nonlinear model of SRM. And a whole SRD system model is built up all by simulink modules based on the optimized BP network, which lays a good foundation for the fault analysis of SRD system.The insulation system of phase winding is the weakest part of the SRM body, and motor symmetry would be destroyed by the inter-turn winding short-circuit fault. This dissertation reconstructs every current phasors, and presents a calculation method of symmetrical current components for SRM. On this basis, the low-pass filtered amplitude ratio of negative sequence and positive sequence is established as a fault feature for winding faults in order to avoid the effects of changing operation conditions, which could effectively improve the detecting sensitivity and reliability.Power converter is considered to be one of the most probable failure parts in SRD system. Faults of converter would badly destruct the equilibrium condition of drive system, and cause serious problems under continue operation with faults. According to the asymmetrical half-bridge converter, all single-phase faults are analyzed via the nonlinear simulation model and experiments, and the change patterns of current spectral components under voltage PWM control strategies with converter faults are studied by Fourier Transform. Then the concept of relative spectral coefficient is given as a fault diagnosis eigenvalue. At last, experiments with simulated faults are carried out on the RT-LAB hardware-in-the-loop simulation system and a prototype SRM, and the sensitivity and veracity of the proposed diagnosis method is verified by experiment results.The operating characteristics of SRM determines that all phases could not work at the same time; thus combined with the drive signals, the working state of each bridge arm could be monitored by detecting bus current. With the objective of convenient realization, rapid and reliable detection, this dissertation presents three on-line fault diagnosis methods for power converter, including detecting the DC bus current and upper freewheeling bus current scheme, detecting the chopping bus current scheme, detecting the excitation bus current and upper freewheeling bus current scheme. These diagnosis methods could save hardware costs, and obtain the needed fault characteristic information rapidly through digitizing the detected current. All these methods could be easily implemented, and get a higher anti-interference ability and reliability. From the control and diagnosis experimental system based on DSP, the obtained results demonstrate that the detected currents have significantly fault features for rapid diagnosis, and it's easy for fault location, which would supply an effective guidance for further fault-tolerant control.Valid position signal is an important guarantee for reliable operation of SRD system. Through deep analysis of the effects on system under failures of single or more position sensors, a position fault detection method with a diagnosis buffer zone is presented in this dissertation, to guarantee the accuracy of diagnosis results effectively. Regarding to the invalidation of single or double position sensors fault types, a position signal fault-tolerant control scheme is given by reconstructing the lost commutation processes at proper positions. The validity and rapidity of the proposed detection and fault-tolerant control scheme is verified by simulated fault experiments.