Journal of Shanghai Jiaotong University ›› 2016, Vol. 50 ›› Issue (04): 636-640.

Previous Articles     Next Articles

Analysis of Transient Thermal Resistance of LED Based on Bayesian Probability and Statistics

CHEN Weia,b,QUE Xiufua,FENQ Weia,b,ZHANG Jianhuaa,YANG Lianqiaoa   

  1. (a. Key Laboratory of the Ministry of Education of Advanced Display and System Applications; b. School of Materials Science and Engineering, Shanghai University, Shanghai 200072, China)
  • Received:2015-03-15 Online:2016-04-28 Published:2016-04-28

Abstract: Abstract: Based on the theoretical basis of transient thermal test, backstepping algorithm from the result of transient thermal measurement to structure function generally include data fitting, smoothing, derivation, deconvolution and network structure conversion. The Bayesian probability and statistics method was presented for the network identification by deconvolution (NID) in analyzing transient thermal resistance of light emitting diode (LED), which effectively avoided the ill posed problems in the frequency domain method of the computer and the choice of the iteration number in the Bayesian iterative method. The relationship between the optimization parameter ε and halfwave width, and its relationship with ripple of time constant spectrum were discussed. The proposed method successfully extracted the transient thermal resistance in the transformation from Foster network model to Cauer network model with an optimized ε. Meanwhile, the reliability of this method in deconvolution of noisy signal and the accuracy in obtaining the thermal resistance of LED package devices were discussed.

Key words: Key words: Bayesian, network identification by deconvolution (NID), light emitting diode(LED), transient thermal resistance, time constant spectrum

CLC Number: