Adaptive Generalized Time Interval Cumulative Methodology for Optical Port Degradation Modeling in Cloud Computing Service

In recent years, the rapid development of information technology has led to an increasing demand for computing and storage, especially in the cloud computing field. Optical ports have become indispensable hardware in cloud computing services, and their stable operation is crucial. Performing remaining useful life (RUL) predictionon optical ports can effectively prevent accidents from occurring and reduce economic losses. Most existing RUL prediction methods focus on continuous degradation modeling, which often requires continuous degradation trending to describe degradation uncertainty so as to achieve RUL distributions and uncertainty characterization. However, in some industrial applications, such as optical ports, their degradation is discrete and their trending is hard to describe by the aforementioned function. Therefore, this article aims to develop an adaptive generalized time interval cumulative methodology for discrete degradation modeling and RUL prediction. Here, the proposed new methodology is integrated with newly defined “virtual jump points (VJPs) to realize online adaptive model parameters updating. RUL prediction is ultimately achieved by extrapolating the updated model to reach a preset failure threshold. The results show that the proposed new methodology can well model optical port degradation and provide accurate RUL prediction so as to be beneficial to realizing condition-based optimal port replacement. Note to Practitioners—Optical ports are one of the important hardware in the field of cloud computing, and their physical characteristics determine their discrete characteristics of degradation. Practitioners need to know how to handle data with discrete degradation features. This article proposes an adaptive generalized time interval accumulation methodology, which can achieve adaptive selection of discrete degradation data distribution and RUL prediction, offering valuable guidance for equipment operation and maintenance. This article highlights the superiority of the method by comparing it with existing continuous degradation modeling methods.