Life Prediction Of LED Based On Genetic-ant Colony RBF-BP Neural Network

Normally operating Light Emitting Diode(LED)products have a service life of hundreds of thousands of hours,and traditional test methods have long test times and high costs.In the process of LED degradation,various extreme conditions may be encountered,such as high temperature,low temperature,vibration,and thermal shock.This extreme environment will aggravate LED performance degradation.If the LED lifetime can be effectively predicted,the normal operation of production and life can be guaranteed,and the cost of engineering practice can be saved.This paper proposes an LED life prediction model optimized by genetic and ant colony algorithm,which can overcome the shortcomings of reduced individual diversity and premature convergence when optimized by genetic and ant colony algorithm alone,and improve the accuracy of LED life prediction.Firstly,this article introduces the structural principles and light-emitting characteristics of LEDs,draws out the physical mechanisms that affect LED failures,and obtains four types of LED failure factors,selects the luminous flux in the expected life,the current in the electrical parameters,and the temperature and color in the thermal parameters.The color coordinate drift change amount in the learning parameters is used as the input amount of the neural network,and the reasons for the selection are explained separately.It also introduces the related concepts of RBF(Radial Basis Function)and BP(Back Propagation)neural network,which provides a theoretical basis for subsequent neural network modeling.Secondly,in view of the slow convergence speed and inaccurate prediction of the RBF-BP life prediction model,the genetic and ant colony algorithm in the bionic intelligent swarm algorithm are introduced to optimize the life prediction model;the RBF-BP network is used to construct Life prediction model,and then use Genetic Algorithm(GA)and Ant Colony Optimization(ACO)to optimize the life prediction model and perform simulation verification.By comparing with the two models of RBF-BP and GA-RBF-BP,comparing the average relative error,mean square error and fitting accuracy,it proves that the constructed ACO-RBF-BP neural network model has higher prediction accuracy.Finally,in view of the local optimization and slow prediction speed of the optimized life prediction model,the genetic-ant colony optimization algorithm(GAACO)is used to optimize the life prediction model,and GAACO-RBF-BP life prediction model.By comparing the two life prediction models optimized by genetic and ant colony algorithm,it is verified that the LED life prediction model of GAACO-RBF-BP has the advantages of fast prediction speed,good output stability and high prediction accuracy.