Analytical Remaining Lifetime Modeling for Electric Locomotive IGBT Modules Based on Failure Evolution Simulation
Traction converter as the most critical equipment in the electric locomotive system with the highest failure rate affects the safety of the system,where insulated gate bipolar transistor(IGBT)modules are the weakest components causing the failure of traction converters.Accurate evaluation of the IGBT module's remaining lifetime has important strategic value and economic significance for reducing the operation and maintenance cost and system failure rate.However,the current mileage repair operation and maintenance method does not consider the influence of load,condition,and locomotive operating lines difference on the service life,which leads to high operation and maintenance costs for the early replacement and system outage risk for the late replacement.Therefore,due to the high failure rate,the large reliability difference among operating lines,and the unreasonable mileage repair method,the common industry concern is:what is the remaining lifetime for the repaired and replaced IGBT modules.Aiming at the problem of accurate remaining lifetime estimation of IGBT module for the electric locomotive,a refined multi-physical field model considering the void spatial distribution in the IGBT module solder layer is established,and the failure evolution physical process simulation of multi-chip parallel IGBT modules is carried out.The analytical remaining lifetime prediction model is proposed,where the statistical law of solder layer voids distribution is selected as the aging state,and the accuracy of the improved model is verified by power cycling tests.The intended outcome of this paper enriches the power device's lifetime prediction and reliability evaluation theory,meanwhile,it provides the scientific basis for the system to change from mileage maintenance to condition-based maintenance.
electric locomotiveinsulated gate bipolar transistor(IGBT)modulemorphological evolution simulationremaining lifetime model
万方数据
维普
