Study On Fault Diagnosis And Fault-tolerant Control For The Cascaded STATCOM

Cascaded static synchronous compensators(STATCOM) are widely used inreactive compensation and harmonic suppression on medium voltage distributionnetwork because of their advantages such as modularity, high power capacity, goodharmonic characteristic, and easily realized redundancy control. The reliability of theSTATCOM is directly related to the safe operation of the power grid. Therefore, theSTATCOM system with fault-tolerant capabilities is recently proposed and hasbecome a research focus for scholars both domestic and abroad. The STATCOMsystem with fault-tolerant capabilities can diagnose and analyze faults in a timelymanner as well as reconstruct both the software and hardware structures of the systemupon fault diagnosis to ensure that the device continues to operate safely without anyreduction or partly reduction in performance indicators. Research shows that thepower semiconductors in power converters are weak links of the STATCOM systembecause of their high frequency and energy transformation environment. Onlinereal-time fault diagnosis of power semiconductors and appropriate fault-tolerantcontrol strategies are effective means to improve the reliability of the STATCOMsystem. Therefore, research on fault diagnosis and fault-tolerant control strategies ofconverters in the STATCOM system has important significance both in theory andpractical value.The precondition of realization of the fault-tolerant STATCOM system isaccurate fault diagnosis. Immediacy and reliability of fault diagnosis is directlyimpacted to the effect of the fault-tolerant strategies. The converter in the STATCOMsystem is a typical hybrid system which include both continuous state variables anddiscrete switching variables. It is always in dynamic response of the load change.Therefore, the traditional analytical modeling method or signal processing methodwhich takes the output current or voltage as the original signal can not effectivelyreveal the fault features of the system. The traditional method can not be used foranalyzing and diagnosing faults inside converters. A comprehensive fault diagnosismethod for converters is proposed by using wavelet multi-resolution analysis(MRA),fault feature dimension reduction techniques and multi-kernel least squares supportvector machines(MLS-SVM). Capacitor voltages of power modules in the converter isselected for the original signal. The method overcomes the traditional shortcomings indiagnosis time, load changes, fault samples requirements, etc. First, the original signals are decomposed into eight layers via the db40wavelet, and the detailcoefficient energy of each layer is selected as the fault features. The fault classifier isthen constructed by using MLS-SVM. The parameters of the MLS-SVM is optimizedbased on chaotic particle swarm optimization algorithm(CPSO). Finally, thedimensions of the features are reduced by using kernel principal componentanalysis(KPCA) and hierarchical clustering algorithm(HCA). The simulation andexperimental results demonstrate that it overcome the complex calculation of thetraditional SVM, improved the robustness of fault diagnosis, maintained theadvantages on small sample learning, and reduced diagnostic time while maintainingdiagnostic accuracy.Due to the space vector pulse width modulation(SVPWM) method is difficult toachieve in the cascaded STATCOM, a generalized SVPWM(GSVPWM) method isproposed with the aim of reducing switch loss and easily implement. The SVPWMcan be equivalent to a carrier-based method by adding one zero sequence component.Different PWM methods are obtained easily by changing the zero vector distributioncoefficient. A simple digital realization method for the GSVPWM is also proposed toavoid calculations such as complex trigonometric function, coordinate transformations,space vector selections, and duration time calculations. The advantages of theGSVPWM method are highest DC voltage utilization ratio, low switch loss, quicklyresponse, and easily realized fault-tolerant control, etc.By reconfiguring software and hardware structures fault-tolerant controlstrategies of the STATCOM system can ensure post-fault systems have such reliableand stable operation capability. The fault-tolerant topologies and control strategies ofthe cascade STATCOM are discussed. Redundancy scheme with one stageredundancy and isolation scheme with only bypassing the fault module are applied inthe system. The power modules run less than its rated value when normal operation.When the fault happens, the drive pulses are secured and the fault module is isolatedfrom main circuit with bypass device. The influences of fault on each system controllink are analyzed. Fault-tolerant control strategies such as phase-to-phase capacitorvoltage balance control, one phase capacitor voltage balance control and PWMcontrol are proposed to guarantee the reliable and stable operation of the STATCOM.An eight cascaded STATCOM experiment platform is designed for faultdiagnosis and fault-tolerant control. The distributed control structure based on fieldprogrammable gate array(FPGA) is chose for its advantages such as modularity, easily control and maintenance. The experimental results demonstrate that the platform havegood dynamic compensation effects, accurate diagnostic ability, and fine fault-tolerantcontrol performance. It has a good prospect in industry application.