| Solar simulator is an equipment used for simulating the different spectrum and radiation intensity of the sunlight,and is indispensable in photovoltaic industry.Compared with xenon lamps and halogen light sources in traditional solar simulator,LEDs have stronger controllability and wider adjustable range,so that multi-LED solar simulators have great potential in the research of new simulators.The light source is one of the core components in simulator,and the LED light source is a nonlinear system with strongly coupled multi-parameter and the photometric,electrical,and thermal features of LEDs are highly dependent on one another.With the increasing current,the LED junction temperature rises sharply,which will decrease the light efficiency and stability of optical properties so the heat dissipation problem becomes a major obstacle to ensure the stable output of the LED light source.The photoelectric thermal(PET)numerical model describes the coupling relationship between photometric,electrical,and thermal features and then provides a reference for thermal design.However,the junction temperature in the model is difficult to measure and estimate accurately,and there are many parameters which leads to complex calculation of the model.In this paper,a new test and modeling method is proposed to analyze and model the thermal characteristics of LEDs,and the heat dissipation structure is optimized to improve the heat dissipation performance of the system.Aiming at the heat dissipation problem of LED light source,in this paper,an RBF neural network model based on PET theory is built,and the photoelectric thermal properties obtained from the model are used as the boundary conditions for the design of the heat dissipation structure,and then the fin heat sink and rectangular channel are designed and optimized to improve heat dissipation performance.The main contents are follows:(1)The LED photoelectric thermal boundary conditions are indispensable to analyze the relationship between the heat dissipation structure and the heat dissipation performance.In this paper,the PET numerical model is used to analyze the relationship of LED photometric,electrical,thermal coupling.Given the heat sink,the LED photoelectric thermal characteristic test platform with temperature-controlled is built and the RBF neural network model is established between the temperature of heat sink,driving current and luminous flux output.The results show that the predicted values of RBF model show good consistency with the measured values to accurately describe the nonlinear relationship between photoelectric thermal parameters,and then get the photoelectric thermal properties under different conditions.(2)The photometric,electrical,thermal features predicted by the R_BF neural network model are used as the boundary conditions for thermal model of thermal design.Based on the multi-physics coupling analysis,the finite element analysis is used to calculate the fin heat sink under natural convection in this paper,and the RBF neural network model is built between the thickness of the substrate,the fin height,the thickness and resistance,and then particle swarm optimization algorithm is used to optimize the structure parameters.The results show that the effect of heat dissipation is the best when the substrate thickness is 5.25mm,the fin height is 53.27mm,and the fin thickness is 4.35mm.(3)The active cooling is necessary because the heat transfer efficiency of natural convection is not high.In this paper,the air cooling is adopted and the heat transfer behavior of the rectangular channel is analyzed.And then the relationship between the thickness of channel substrate,channel height,number and thermal resistance is used to optimize the channel structure.The experimental results show that the optimized channel structure has improved heat dissipation performance for 48.7%compared to the heat sink by natural convection.(4)Due to the heat accumulation effect in the heat and mass transfer process,the temperature in the surface contacting the rectangular channel and heat source is ununiformed.Therefore,a wedge-shaped channel with large inlet and small exit is chosen,and the temperature distribution is calculated by finite element analysis and computational fluid dynamics to get the temperature difference under different ratio.The experimental results show that when the area ratio is 2,the temperature uniformity is the best,which is up to 32.5%higher than that of rectangular channels.In this paper,the RBF neural network model based on photoelectric thermal theory provides boundary conditions for thermal design,then the fin heat sink and rectangular channel are designed and optimized.Compared with fin heat sink,the heat dissipation performance of rectangular channel increases for 48.7%,and the temperature uniformity of the wedge channel is higher for 32.5%than that of the rectangular channel,which improves the heat dissipation performance of the light source system and provides guarantee for the stability of optical performance. |
PDF Full Text Request