Research On Mutlti-Objective Tolerance Design Method Based On Degradation Characterictics For LED Driver

With the advantanges of long-life, energy efficiency, low carbon and environmental protection, LED lighting has a broad market. It has been widely used in backlight, architechtual and other fields. With the development of high-power white LED, there is a greater value if the traditional incandescent and fluorescent can be replaced by LED lighting. Due to the instability of output current, poor quality and consistency, short life and low reliability of LED driver, LED can not fully play its advantages; this has seriously affected the promotion of LED lighting. Therefore, the reliability design on LED driver has important practical significance and theoretical value to improve the robustness to design and envirtomental factors.Roubust design, especially tolerance design is an important method that improves the consistency of product quality and balances the relation between cost and quality. It has been widly applied on electronics reliability design. However, the traditional method of tolerance design only focuses on the output response characteristic; this method is increasingly unable to meet the development of electronic product reliability. Considering the deficiency, a multi-objective tolerance design method has been proposed, which not only designs the output response, more importangly, all components are considerd as design objects to reduce the degradation stresses on key components, thereby improving product reliability and service life.The principle of LED driver is analyzed to establish the PSpice simulation model. The FMMEA analysis is carried out to determine the potential failure causes and failure mechanisms. According to the application conditions, the failure mechanisms are prioritized. Based on the high priority failure mechanisms and degradation models, the key components of LED driver are determined through virtual reliability assessment.The output current and degradation stresses on key components are used to established objective functions and constraints. The sensitive factors that affect the objective function are determined by sensitivity analysis. And only the sensitive parameters are designed by tolerance analysis and allocation. This paper established the quality-cost model after the establishment of manufacuturing cost model and using cost model respectively. On the minimum cost conditon, the tolerance ranges of sensitivity paramenters are calculated by Genetic Algortithm.Based on experiment design to establish approximate model, a multi-objective tolerance method for LED driver is proposed. According to space filling experment design and response surface modeling method, we establish mutli-objective surface models of LED driver and verify their accuracy and predictive ability. Finally, the mutli-objective optimization function established is solved to determine the optiamal tolerance allocation scheme, and the results are demonstrated through comparison.