Component Reliability Parameters Prediction Based on an improved Discrete Grey Model

During the long-term service of electronic components, their key performance parameters gradually degrade. Accurate modeling and prediction of reliability parameters are essential for system life-cycle assessment and fault tolerance. However, due to factors like testing costs and degradation cycles, reliability-related samples are often limited, posing challenges for modeling. Traditional grey prediction models often assume fixed initial points, limiting their effectiveness in dynamic and small-sample scenarios. Based on the Discrete Grey Model (DGM), we used an ending-point optimized discrete grey model (EODGM) that overcomes this limitation by optimizing the initial iteration point based on the endpoint, fully utilizing the principle of new information priority. Simulation results demonstrate that the EODGM model improves prediction accuracy and robustness in reliability parameter forecasting. The model’s adaptability to changing data trends makes it suitable for practical engineering applications.