电工技术学报2024，Vol.39Issue(16)：5104-5120,17.DOI:10.19595/j.cnki.1000-6753.tces.231755
用于IGBT模块温度观测的3-D降阶混合型热模型
A Hybrid 3-D Reduced-Order Thermal Model for Temperature Observation of IGBT Modules
摘要
Abstract
Online monitoring of IGBT module temperature is a key technology for enhancing the reliability of high-power converters.Among the various online temperature monitoring methods for IGBT modules,the thermal network method has garnered attention due to its ability to estimate the internal temperature distribution within the IGBT module.The temperature observer based on the thermal network model can correct the model by utilizing measurement information to reduce the impact of aging,temperature-dependent material parameters,and power loss calculation errors on temperature estimation accuracy.However,existing thermal models struggle to balance system observability and modeling accuracy,and their high model order complicates their application in temperature observers.Therefore,an observable reduced-order hybrid 3-D thermal model suitable for temperature observer for forced air-cooling multi-chip IGBT modules is proposed in this paper. Firstly,a hybrid thermal network combining the strengths of the 3-D Cauer model and the Foster model is introduced for air-cooling multi-chip IGBT modules.This model ensures both thermal modeling accuracy and observability for air-cooled IGBT modules.Based on this model's structure,the corresponding state-space expression is also derived.Secondly,a nonlinear optimization algorithm is employed to refine and calibrate the thermal network model using experimental temperature measurements,reducing the deviation between thermal model parameters and actual parameters caused by uncertain material parameters in the computational fluid dynamics(CFD)model.Finally,the balanced truncation method is utilized to further reduce the order of the proposed thermal network model's state space while maintaining model accuracy. In summary,the main contributions of this paper are: (1)The proposed lumped parameter 3-D thermal network model addresses the challenge that existing reduced-order thermal models cannot accurately model air-cooling IGBT modules with system observability.The error between the proposed thermal network model and CFD simulation is less than 5％,whereas the existing 3-D Cauer model exhibits a maximum error exceeding 34％. (2)The nonlinear optimization algorithm is applied to further correct model parameters.Experimental results demonstrate that this model parameter correction method can decrease the discrepancy between the model and the actual IGBT module by up to 78.69％. (3)The order of the thermal model's state space is further reduced from 270 to 38 by the balanced truncation method,while the model accuracy is still guaranteed within 100 Hz,which further improves the computational efficiency compared to lumped parameter thermal network models. In the future,based on the observable 3-D reduced-order thermal model proposed in this paper and incorporating sensor measurement information,the temperature observer will be further designed to reduce the impact of aging and loss calculation errors on temperature estimation,which can enhance the reliability of thermal monitoring for high-power IGBT modules.关键词
IGBT模块/温度监测/热网络模型/温度观测器Key words
IGBT module/temperature monitoring/thermal network model/thermal observer分类
信息技术与安全科学引用本文复制引用
田野,卜凯阳,李楚杉,李武华,何湘宁..用于IGBT模块温度观测的3-D降阶混合型热模型[J].电工技术学报,2024,39(16):5104-5120,17.基金项目
国家重点研发计划(2022YFE0138400)和浙江省重点研发计划(2023C01061)资助项目. （2022YFE0138400）
