Intelligent Prediction On Heat Dissipation Capability Of The New The High-power LED Street Lamp

In recent years, because of the compactness and the miniaturization of LEDs, the heat flux of per unit increases rapidly which brings great threaten to the reliability of LEDs. It indicates that the invalid rate rises with the rule of index when the temperature exceeds a certain value. Therefore, it is one of the key technical problems to impmve the heat sinking capability for the industrialization of high power LEDs. Many production companies and institut have adopt different cooling way to solve this problem. But there is always excessive heat-sink and heat resistance. In this paper, the oscillating heat pipes (OHP) are applied on radiator whose structure is optimized, the heat dissipation of high power street lamp equiped with new raditor is tested by experiments and temperature signal which is predicted by artificial intelligence to evaluate cooling performance is gathered. The work of this thesis primary has four parts:1) The grey relational analysis method is used to deal with affecting factors of heat transfer performance and further define the primary and secondary impacting factors. OHPs with the best cooling performance are selected in view of primary factors impacting heat transfer performnce and the size of radior.2) The structure of radior with fins, OHP and air hole is optimized by field synergy principle in Fluent. The new radior is contrasted with other three typical raditor in effects of field synergy.3) The experimental system used to gather temperture signal is estabishd. Moreover, the temperture signal is denoised by Empirical mode decomposition.4) The prediction model of temperture signal is established by support vector machine and neural network. After that, the parameters of support vector machine are optimized by genetic algorithm. The support vector machine whose parameters are optimized have higher forcast accuracy.