Numerical investigation on model predictive control of portable electronic devices based on MATLAB/FLUENT co-simulation framework

Temperature control of electronic devices is becoming increasingly important. A better temperature control strategy could improve system performance without any change in the hardware-based thermal design. However, the development of temperature control algorithms is still challenging due to the multiple inner heat sources and thermal constraints, which causes a widespread use of the basic look-up table method on current devices. In this paper, the control performance of model predictive control (MPC) on a laptop is numerically evaluated based on the MATLAB/FLUENT co-simulation framework. In the MPC algorithm, a skin temperature reconstruction model is contained to online estimate the temperature of exterior surfaces based on the information of inner sensors, and a simplified temperature prediction model is utilized to predict the future temperature of both the exterior surfaces and inner sensors. The individual control of the charging current and fan speed show that compared to the baseline look-up table method, the MPC could improve the cumulative charging capacity in 20 min by 20% - 40% under different scenarios, and reduce the average and maximum fan speed by about 30% and 23% respectively. The outstanding performance and low computational cost altogether demonstrate that the MPC algorithm has tremendous potential for practical applications.