A Study On Active Thermal Control Method Of Power Devices With Dynamic Current Limiting

As a new thermal management method for power semiconductor devices(shorten as power devices),active thermal control method has received widespread attention in recent years.Different from the traditional method,it can actively adjust the device losses with on-line estimation and control of junction temperature.Thus,it can safely improve the overcurrent capability and smooth the junction temperature fluctuation of power devices without changing the existing cooling structure and cooling mode.According to the influence factors on the losses,it can be implemented by dynamic current limiting,switching frequency adjustment or modulation strategy switching,etc.Among these,active thermal control method with dynamic current limiting is the core method to improve device overcurrent capability,which has very important theoretical significance and practical value because it can effectively excavate potential current capacity of power devices and alleviate the contradiction between overload of power equipment and thermal safety of power devices.However,the development of the method is not yet mature.Some common problems in practical application still exist,such as unreliable junction temperature estimation model,the lack of optimal setting of junction temperature control parameters and ignored influence of ambient stochastic convection,which affect reliability,accuracy and stability of improving overcurrent capacity of power devices.From the above point of view,in this paper,a new active thermal control method with dynamic current limiting by taking power MOSFET as the object,device thermal network model as the core and combining theoretical analysis with experimental verification was studied.The aim of this study was to overcome the shortcomings of the existing active thermal control method with dynamic current limiting,to improve the engineering practicability of the method,and to provide reference for further development of other kinds of active thermal control methods.The main research work can be summarized as following three points.(1)Research of active thermal control method with current dynamic limiting based on nonlinear junction to ambient thermal network model.Generally,existing active thermal control methods with dynamic current limiting use the junction to case thermal network model to estimate the junction temperature of the power device.However,when the case temperature sensor falls accidentally,the reliability of the model estimation results are hardly guaranteed,which reduces the thermal safety of the power device and affects the reliability of the improvement in overcurrent capability.For this purpose,a nonlinear junction to ambient thermal network model for junction temperature estimation was proposed at first.The model can describe the nonlinear effect of natural convection and heat radiation on the case temperature,effectively reducing the dependence on case temperature sensor and improving the reliability of junction temperature estimation.Then,in order to improve the accuracy of junction temperature estimation,a parameter identification method was proposed,the objective function was constructed and its characteristics were analyzed,and the prime-dual interior point method was employed to obtain its global optimal solution.Finally,an active thermal control method with dynamic current limiting based on nonlinear junction to ambient thermal network model was proposed.The experimental results showed that the proposed method could effectively realize the junction temperature estimation and control under the condition that the case temperature measurement was unreliable,which consequently guaranteed the reliability of device overcurrent.(2)Research of active thermal control method with current dynamic limiting based on junction temperature PI parameters optimization.Generally,existing active thermal control methods with dynamic current limiting adopt the classical PI strategy to control device junction temperature.However,researches on optimal setting of junction temperature PI parameters are few and the dynamic performance of junction temperature control is difficult to be guaranteed,which reduces the thermal safety of the power device and affects the accuracy of improvement in overcurrent capability.For this purpose,taking the nonlinear junction to environment thermal network model as the core,a junction temperature PI parameter optimization method for active thermal control was proposed,the objective function was constructed,and the genetic algorithm was used to obtain the optimal PI parameters results.Then,according to the results,the linear interpolation method of optimal PI parameters in multidimensional space was given by taking the operational and environmental conditions as dimensions.Finally,an active thermal control method with dynamic current limiting based on junction temperature PI parameter optimization was proposed.The experimental results showed that the proposed method could improve the dynamic performance of junction temperature control.Consequently,the accuracy of device overcurrent was ensured.(3)Research of active thermal control method with current dynamic limiting in stochastic convection environment.Usually,existing active thermal control methods with dynamic current limiting assume that power devices are operating in a stable cooling environment.Nevertheless,in practice,power devices may also operate outdoors.In this case,the heatsink of power devices is in a stochastic overcurrent environment,and the convective thermal resistance of the heatsink will always have random pulsation.Affected by this,the power device cannot work in a stable overcurrent state,which reduces the stability of the overcurrent capability improvement and the practicability of the active thermal control method.For this purpose,taking the nonlinear junction to environment thermal network model as the core and combining the principles of markov chain and wavelet packet transform,a stochastic prediction model for convective thermal resistance of heatsink was established to obtain the prediction results of convective thermal resistance with good agreenment in stochastic properties.On the basis of this,the method of evaluating overcurrent capability of the device in a steady state was put forward by using the idea of stochastic programming,and the maximum steady-state overcurrent value was estimated effectively in a stochastic convective environment.Finally,an active thermal control method with dynamic current limiting in stochastic convection environment was proposed.The experimental results showed that the proposed method could reasonably switch the current limiting value of the device to the maximum steady-state overcurrent value based on the output of the junction temperature controller in stochastic convective environment,so as to ensure the stability of the overcurrent capability of the device without overheating.