LED Driver’s Life Prediction And Its Impact On The Whole Lamp Life Based On The Accelerated Test

Accelerated test achieve the rapid aging of the product by accelerated stress. Reliability analysis is a method to establish the relationship between aging information and product reliability and longevity. There are many disadvantages in the conventional aging failure test, such as the long test period and the high economic cost. LED driver is the key LED component. In this paper, the life prediction of LED driver in the accelerated test and its impact on the whole lamp life are studied in order predict the life information of LED products which can improve the production process as a feedback.(1) For the problem of the LED driver’s impact on the whole lamp life, a test method is proposed which LED driver as a subsystem is isolated from LED in experimental group and controlled group. For the problem of LED driver’s life prediction, a black box testing is proposed under the accelerated stress because of its high confidentiality. At the same time, an auxiliary test is designed to provide a database for selecting the parameter of performance degradation. A test platform is build, and the databases of luminous flux and electric parameters are established in order to support the next work.(2) The driver mode and control mode of LED driver are analyzed to determine the driver structure characteristics which we study in this paper. The temperature simulation analysis of driver which has classical topology structure is produced. And the result appears the feasibility of life prediction and reliability analysis with the output electric parameter under temperature stress. Meanwhile these parameters’ changing trends are calculated with Daniel testing method based on Sperman correlation coefficient.(3) In order to give qualitative analysis of the driver’s impact on the whole lamp life based on luminous flux, some works are done as follows. First, this paper establishes a model about the samples’ individual lives and their lives are calculated with nonlinear fitting. Second, a lifetime model obeying Weibull distribution is established for the overall samples and then the individual lives and failure probabilities are transformed to calculate the statistical lifetimes of these two experimental samples with LSR. Third, this paper makes some analyses such as prediction interval analysis, residual analysis and confidence interval analysis of residual. These results show that integrated LED driver similar to linear structure has little influence on the whole lamp life based on luminous flux according to method proposed in this paper.(4) A statistical distribution model is established on the basis of previous work to predict the lifetime of integrated LED driver. First, this paper makes some inspections including P-P figure and KS and then it establishes a statistical model combined with the Arrhenius Accelerated model. The statistical model is based on Wiener-Einstein. Second, an improving model and a parameter estimation method combined with MCMC are proposed to derive the reliability under the normal temperature and reduce the demand of samples. Third, this paper gives some results of life prediction on the control of the whole lamp life.