Research On In-Situ Monitoring Of Wire Bonding Faults In IGBT Modules Based On Multi-dimensional Data Model

Today,IGBT power modules are widely used in photovoltaic power generation,wind power generation and other fields.With the increasingly harsh application environment,the reliability of power modules has become an important factor affecting system security.In the long-term operation of the power module,bond wires failure and solder layer failure are two basic failure modes.This paper takes the bond wires failure as the research content and conducts theoretical analysis and experimental exploration based on monitoring method using the saturation voltage drop.In theory,when the bond wires fail,the saturation voltage drop UCE of power modules increases.Therefore,UCE is usually used as a feature to monitor the bond wires failure.This paper starts with the physical structure of the power module,essentially analyzes the cause of power module failure and proves that the thermal expansion coefficient of the physical material is a key factor affecting the reliability of the power module.Secondly,from the perspective of health monitoring data,this paper analyzes the traditional wire bonding health monitoring method based on two-dimensional data model.The method establishes a two-dimensional curve of the saturation voltage drop and the junction temperature,and determines the health status of the bond wires of power modules according to whether the two-dimensional curve is moved upwards or not,thereby realizing the health monitoring.This monitoring method based on the two-dimensional data model needs to provide a current source with a large rated value in practical application,and thus it cannot realize on-line monitoring of the power module.In order to overcome this shortcoming,this article adds the collector current variable and proposes a three-dimensional data model to meet the conditions required for on-line monitoring.In order to verify the feasibility of on-line monitoring,this paper builds a 15KW three-level NPC photovoltaic inverter experimental platform for experimental demonstration.And the experimental results show that based on the three-dimensional data model,the monitoring method establish the three-dimensional surface of saturation voltage drop,junction temperature and collector current,and the health status of the bond wires is determined according to whether the three-dimensional surface is upwardly moved,thereby realizing on-line health monitoring of power modules and proving that this monitoring method is highly feasible.In order to reduce the influence of the process of online monitoring on the inverter operation as much as possible,this paper uses MATLAB simulation to find the best monitoring time and the best monitoring duration to achieve the maximum benefit of online monitoring.In addition,this paper studies the influence of different working states of the power module on the parameters of diagnostic features,and build the multi-dimensional data model of power modules from the perspective of parasitic parameters and load changes,which provides a new perspective for the research of power module reliability.