Novel Approaches For Multiple IGBT's Open Switch Fault Diagnosis In CMLI Based Systems

In this decade,Cascaded Multilevel Inverter(CMLI)industry claps the pledge with substantial progress and significant accomplishments to improve the performance,evaluation,development and establishment of inverters in renewable energy sources,electrical machine drives and mechanical drive applications and so on.Fault diagnosis in multilevel inverters has taken worthwhile attention into account in this era.Researchers,Scholar,Engineers and Scientist are paying august attentions towards fault diagnosis in multilevel inverter they are also ascertaining new methods and techniques to diagnose accurate faults in minimum time.Because fast and precise detection is very important to make system more reliable and efficient.Our contribution is to scrutinize faults and provide their real time perspective solutions to make system reliable because in power electronics reliability is the most sensitive,salient and paramount concern.In power electronics,two ways to diagnose faults either online or offline.In online fault diagnosis,we implemented fuzzy logic and in offline faults,we applied probabilistic principle component analysis-k-nearest neighbors(PPCA-k-NN)algorithm and probabilistic principle component analysis-support vector machine(PPCA-SVM)schemes for fault detection,identification and classification in minimal time with highest accuracy achievements.In offline system,we applied new fault diagnosis method in CMLI system based on feature extraction and feature classification.we expounded PPCA for the reduction of dimension in output data for feature extraction due to its significant attributes such as optimize data based on probability model,facilitation statistical,testing,application of Bayesian methods,capability to combine multiple PCA,and PCA Projection.We also elucidated PPCA with help of k-NN and SVM for fault classifications and feature classifications due to their significant attributes such as k-NN has good stability,high accuracy and easy implementation.Whereas SVM is used to find the best hyper plane to maximize the distance between two classes input data from largest distance to nearest training data point with kernel techniques.The output voltage signals under different fault conditions of CMLI are taken as the fault characteristics signals to avoid the effect of load variation on fault diagnosis.PPCA is used to optimize the data without changing the original properties of the input data,both k-NN is used to identify the accurate fault location and diagnosis the fault.This technique is validated with real time experiment using Field-Programmable Gate Array(FPGA)to reduce fault diagnosis time with highest accuracy in minimum time.In PPCA-SVM method is used for controlled switches in single phase CMLI.The output voltage signals under different fault conditions of the CMLI are taken as fault features by using Phase Shift PWM(PS-PWM)technique.PPCA is used to optimize the data and reduce dimension of fault features.Finally,SVM classifier is used to diagnose the different fault modes.An experimental setup of CMLI has been designed to validate the proposed fault diagnosis method.In simulation three conditions are explained.Firstly,parameter initialization of PPCA and SVM,then output voltage signals from inverter conversion to sample singles.Sample signals are given to PPCA to optimize the data and optimize data is given to SVM.In real time experimental results same simulation parameters are used to achieve the highest accuracy of the fault location and to reduce fault diagnosis time.The simulation and experimental results described that the proposed fault diagnosis method is the most suitable method to handle the system failure.In online system,a fault diagnosis technique has been proposed for three-phase CMLI.We presented fuzzy logic fault diagnosis technique based on the average current Park's vector technique for three-phase CMLI fed permanent magnet synchronous motor(PMSM)drive.The three-phase normalized currents divided into positive(E_q+)and negative groups(E_q-).Furthermore,the average current Park's vector technique is used to measure the fault symptom variables with help of the phase current information.The proposed fault diagnosis technique can detect and locate the single or multiple open-circuit faults,and also the intermittent faults in IGBT's at any condition(Ex.,change in mechanical speed,change in load torque and change in system parameters),which can improve the reliability of the motor drive system.In simulations,we discussed three cases Firstly,change in speed at no fault condition,secondly two condition are explained,single IGBT open circuit faults and single IGBT open circuit(intermittent)faults.Finally,multiple IGBT's open circuit faults.To validate the defined fuzzy logic rules,real time experiments performed with same simulation parameters which mentioned in the simulation.The proposed method effectively validated in simulation and real time experimental with highest accuracy of fault detection achieved in minimum time.