The increased system complexity in electronic products brings challenges in a system level reliability assessment and lifetime estimation. Traditionally, the graph model-based reliability block diagrams (RBD) and fault tree analysis (FTA) have been used to assess the reliability of products and systems. However, these methods are based on deterministic relationships between components that introduce prediction inaccuracy. To fill the gap, a Bayesian Network (BN) method is introduced that considers the intricacies of the high-power light-emitting diode (LED) lamp system and the functional interaction among components for reliability assessment and lifetime prediction. An accelerated degradation test was conducted to analyze the evolution of the degradation and failure of components that influence the system level lifetime and performance of LED lamps. The Gamma process and Weibull distribution are used for component level lifetime prediction. The junction tree algorithm was deployed in the BN structure to estimate the joint probability distributions of the lifetime states. The degradation and prediction results showed that LED modules contribute a major part for lumen degradation of LED lamps followed by drivers and the least effect is from diffuser and reflector. The BN based lifetime estimation results also exhibited an accurate prediction as validated with the Gamma process and such improved reliability assessment outcomes are beneficial to LED manufacturers and customers. Thus, the proposed approach is effective to evaluate and address the long-term reliability assessment concerns of high-reliability LED lamps and fulfill the guarantee of high prediction accuracy in less time and cost-effective manner.

电子产品中日益增加的系统复杂性给系统级可靠性评估和寿命估计带来了挑战。传统上，基于图形模型的可靠性框图（RBD）和故障树分析（FTA）已用于评估产品和系统的可靠性。但是，这些方法基于引入预测误差的组件之间的确定性关系。为了填补这一空白，引入了一种贝叶斯网络（BN）方法，该方法考虑了大功率发光二极管（LED）灯系统的复杂性以及组件之间的功能相互作用，以进行可靠性评估和寿命预测。进行了加速退化测试，以分析影响系统级寿命和LED灯性能的组件退化和故障的演变。Gamma过程和Weibull分布用于组件级寿命预测。连接树算法被部署在BN结构中以估计生命状态的联合概率分布。退化和预测结果表明，LED模块是导致LED灯流明退化的主要部分，其次是驱动器，而影响最小的是扩散器和反射器。基于BN的寿命估计结果还显示出准确的预测，这一点已通过Gamma工艺验证，并且这种改进的可靠性评估结果对LED制造商和客户有利。因此，