This paper addresses the catastrophic failure of power converter systems caused by the aging of the insulated-gate bipolar transistor (IGBT) module. Monitoring the aging state of IGBT in real time enables the defective module to be found and be replaced timely. Solder fatigue increasing the internal thermal resistance of the module has been identified as one of the main root causes of IGBT module aging. In this paper, a novel event-driven solder fatigue monitoring mechanism is proposed. The proposed method consists of two models which are a dynamic solder aging monitoring model and an event-driven online calculation model of the increase in thermal resistance. The dynamic solder aging monitoring model detects solder fatigue in real time by the variations of case temperatures with thermal sensors on the bottom surface of the baseplate. The event-driven online calculation model estimates the increase in thermal resistance at the occurrence of the solder aging event by the junction temperature from thermosensitive electrical parameters, which are not affected by the solder fatigue. With the cooperation of two models, the aging level of the IGBT module caused by solder fatigue is evaluated. Simulation and experimental results are provided to verify the effectiveness of the proposed method.

Implementation of real-time health assessment and thermal management of insulated gate bipolar transistors (IGBTs) require thermal equivalent circuit models that can be used to predict the junction temperature of the modules. Solder aging in IGBT’s has a substantial impact on the accuracy of junction temperatures estimated by the models. This paper proposes an adaptive thermal equivalent circuit model that can estimate the junction temperature of IGBTs with precision under solder aging conditions. Firstly, the solder aging process is monitored in real time by the temperature gradient of the baseplate of the IGBT module, which is easily implement by placing two thermal sensors at the interface between the baseplate and the cold plate. Then, when the solder aging is detected, the actual junction temperature obtained by an on-state collector-emitter voltage of the IGBT is utilized to update model parameters based on the thermal behavior of the device. By combining the two stages, the effect of solder aging on the accuracy of the junction temperature estimate is removed in time. Simulation and experimental results are provided to verify the effectiveness of the proposed method.

This paper proposes a novel method for optimizing the Cauer-type thermal network model considering both the temperature influence on the extraction of parameters and the errors caused by the physical structure. In terms of prediction of the transient junction temperature and the steady-state junction temperature, the conventional Cauer-type parameters are modified, and the general method for estimating junction temperature is studied by using the adaptive thermal network model. The results show that junction temperature estimated by our adaptive Cauer-type thermal network model is more accurate than that of the conventional model.

Condition monitoring (CM) of insulated-gate bipolar transistor (IGBT) modules is significant for improving power converter reliability and reducing the loss caused by IGBT failure. In this paper, we present an advanced method to evaluate the aging condition of an IGBT based on the changes of thermal characteristics in the module, which can be described by the junction and case temperatures. The variation of the case temperature distribution is used to monitor the solder fatigue. The estimated level of solder fatigue is used to update the parameters in the electrical-thermal model. Then, the junction temperature can be accurately estimated by the updated electrical-thermal model. The difference between junction temperatures estimated from the updated model and temperature-sensitive electrical parameter (TSEP) is utilized to monitor the aging of bond wires. The proposed CM method permits both of health level evaluation and accurate junction temperature estimation of IGBT in real time to prevent catastrophic failure caused by defective modules. The experimental results on a commercialized IGBT product have validated the feasibility of the proposed method.

Junction temperature is a key parameter that influences both the performance and the reliability of the insulated gate bipolar transistor (IGBT) module, while solder fatigue has a significant effect on the accuracy of junction temperature estimates using the electro-thermal model. In this paper, an improved electro-thermal model, which is independent of solder fatigue, is proposed to accurately estimate the junction temperature of IGBT module. Firstly, solder fatigue conditions are monitored in real time with the information of the case temperatures. Secondly, when solder fatigue is found, the update process of the electro-thermal model parameters is performed to match the model parameters with the fatigue device. With the above two-step process, the influence of solder fatigue on the accuracy of temperature estimates can be removed in good time. Experimental results are provided to validate the effectiveness of the proposed method.

The dynamic matrix control (DMC) strategy is extended to a nonlinear system using Hammerstein model that consists of a static nonlinear polynomial function followed in series by a linear step response dynamic element. A new nonlinear Hammerstein DMC algorithm (named NLH-DMC) is presented in detail. The set point tracking and disturbance rejection simulation results of a strong nonlinear PH neutralization process control show that NLH-DMC gives better control performance than linear DMC (LDMC) and the nonlinear PID (NL-PID) controller commonly used in process industry. Further simulation experiment demonstrates that NLH-DMC not only gives good control response, but also possesses good stability and robustness even existing large modeling errors