Research Of Fault Detection Method For SRD Based On The Current Signalextracted Technology

Due to the simple structure and independent control among each phase, switchedreluctance motor （SRM） has higher reliability. Compared with other motors, SRM alsohas fault tolerance. However, the inherent fault tolerance is limited, so longterm operationin fault is bound to damage the entire system. Therefore, it is great significant to the studyof fault detection technology for switched reluctance motor drive system （SRD）, in orderto take appropriate fault tolerance strategy. This paper makes the fault detection technologyfor SRD as the study theme, and respectively presents new detection methods for statorwinding inter-turn short circuit fault, the power converter failure and the air gapeccentricity fault based on current signal extraction technology.Firstly, this paper establishes model for SRM based on GA-BP algorithm.Respectively make winding phase current and rotor position angle as inputs and make fluxand torque as outputs to establish two neural network models, and use the GA-BP hybridalgorithm to train the network. Compared with the traditional BP neural network model，themodel with this method has more accuracy and reliability. And based on Matlab/Simulinkenvironment, it also presents an independent functional modularity modeling method forSRM with8/6pole. The body of SRM, power converter, speed control, current control,rotor position angle selection, torque calculation modular and parameter calculationmodular are respectively built, then combinate each part together organicly to build thewhole simulation model for SRD. Speed controller model is constituted by the PIregulator, the current control module uses the control method combinated the angularposition control （APC） and the current chopping control （CCC）, that effectiveimplement the real-time control.Secondly, based on finite element analysis software Maxwell and circuit analysissoftware Simplorer, this paper establishes the simulation model for SRD and takes anin-depth look at the principle of short-circuit fault. And two cases are simulated, one wherea few turns were shorted, and one where a complete pole was shorted, respectively from theview of magnetic circuit and field circuit. Results of simulation and experiment show thatwinding short-circuit fault breaks the symmetry between the motor poles and influencesoutput performance negatively.On this basis, this paper extracts the fundamental component to reconstruct four phase symmetrical current based on spectrum analysis withphase current for SRM. And the mothod of symmetrical components is used to calculate thepositive and negative sequence component of reconstructed current, then amplitude ratio ofpositive sequence and negative sequence is seen as fault feature. Results of simulation andexperiment under different conditions show that the method can effectively achieveinter-turn winding short-circuit fault detection.Moreover, based on different behavior patterns of the power converter supply currentand each phase current under normal and fault operation, fault diagnosis eigenvalues aregiven, and combined with on-off states of power elements to make a comprehensiveanalysis. This paper presents a new fault detection method for power converter of SRM,that can timely and accurately detect the fault occurs, diagnose the fault phase, determinefault type and identify faulty component. Results of simulation and experiment show thatthe accuracy and reliability of detection method under low or high speed operation.Meanwhile, this paper obtains the function relation between air gap length and thenon-excited phase current based on the circuit equations of SRM, then uses finite elementanalysis to calculate non-excited phases current under motor in the normal state anddifferent eccentric level, and presents a new detection method of air gap eccentricity faultthat makes the difference of non-excited phase current as eigenvalues. This method canrealize comprehensive detection about occurrence of the fault and its location and type.Finally, this paper obtains the function relation between air gap length and differentialcurrent of two stator pole windings in one phase based on the circuit equations of SRM,then uses finite element analysis to calculate the differential current under motor in thenormal state and different eccentric level, and translates the differential current underdifferent excited modes into the matrix expression. Through analyzing the symmetry ofelements in the matrix expression, this paper presents a new fault detection method of airgap eccentricity that makes differential current of two stator pole windings as eigenvalues.This method can realize comprehensive detection about occurrence of the fault and itslocation, direction and type. Results of simulation and experiment show that this methodhas higher accuracy, reliability and robustness. And this paper also summarizes thecomparison of two eccentric fault detection scheme.