Study On Defect Monitoring Of IGBTs In The Power Electronic Converter

Reliability is one of the key factors that would limit the applications of powerelectronic converters, so developing evaluation method for the reliability of powerelectronic converters is an important measure to reduce the fault risk, realize the healthyoperation and save the cost of maintenance. In recent years, with the great developmentmade in power electronic technology, the application fields of power electronicconverters have been extended extremely, then power electronic converters start to playa vital role in promoting economic development and social progress. Therefore,reliability issues of power electronic converter are increasingly outstanding andbecoming a hot topic gradually, related research is just in the ascendant. In this paper,IGBTs as key components of power electronic converters are focused on, the defectconcept of IGBTs is proposed from condition monitoring viewpoints. Three aspects,namely, the reliability evaluation of IGBT is summarized to its defect off-line diagnosis,characterizing and on-line monitoring are systematically studied, which then beextended to the health management level, discussing the realization of state predictionfor IGBTs.Understanding and obtaining precursor information of failure is the basis ofreducing accident and saving maintenance cost. Defects are results of the fatigue, andessential factors in reducing the operational reliability of IGBTs and increasing the riskof an abrupt failure. Therefore, main defect types of IGBTs and their influence onparasitic parameters are analyzed. Then an off-line diagnostic based on frequencyresponse analysis is proposed in this paper. In the end, the correctness and applicationvalue is verified by test results. This method can be used in the regular overhauling ofpower electronic converters, is an effective measure for fault prevention.Tracing characteristic parameters of defects is a precondition for evaluatingreliability degradation of IGBTs during running. Unfortunately, most of physical wearout mechanisms of IGBTs give very few external indications of impending failure, andthose indications are not only difficult to be obtained but also affected by many factors.To circumvent this limit, the idea of using dynamic changes of gate voltage and currentas characteristic parameters of defects is proposed in this paper. In addition, thefeasibility is verified by experimental study. The failure of IGBTs depends on the fatigue damage accumulation with runningtime, and many uncertain factors would affect it. Therefore, on-line defect monitoringfor an individual IGBT is an important guarantee for avoiding an abrupt failure. For thispurpose, monitoring methods for defects under running condition are proposed in thispaper, which are based on the aging features of parasitic parameters. Moreover,dynamic changes of gate voltage and current are employed as observable variables tofind out the appearance of defects. These methods can be considered as pre-faultdiagnosis means, which are able to give operators a chance to replace defectivecomponents and avoid accidents enlargement.Future state evaluation is a very important content of health management forIGBTs, which is an evidence to make the scheduled maintenance and realize theCondition-Based-Maintenance. However, the traditional probabilistic methods based onlarge sample statistics are unable to be applied here, due to uncertain of runningcondition and other factors. For this reason, the gray theory is used in this paper, and agray state prediction model is built up. Besides, the failure rate function of IGBTs basedon the thermal fatigue damage accumulation is also preliminary discussed.