Abstract:In recent years, rapid development of information technology has led to an increasing demand for computing and storage, especially in cloud computing field. Optical ports have become one of indispensable hardware in cloud computing services, and their stable operation is crucial. Performing remaining useful life (RUL) prediction on optical ports can effectively prevent accidents from occurring and reduce economic losses. Most existing RUL prediction methods focus on continuous degradation modeling, which often require 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 aforementioned function. Therefore, this paper 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” to realize online adaptive model parameters updating. RUL prediction is ultimately achieved by extrapolating the updated model to reach a preset failure threshold. 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 paper 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. The paper highlights the superiority of the method by comparing it with existing continuous degradation modeling methods.

Degradation Prediction of Semiconductor Lasers Using Conditional Variational Autoencoder

A Machine Learning-based Framework for Predictive Maintenance of Semiconductor Laser for Optical Communication

Prediction of Operational Lifetime of Perovskite Light Emitting Diodes by Machine Learning

Machine learning and FPGA implementation for predicting luminance decay and temperature distribution in micro-LED displays

Processes of the Reliability and Degradation Mechanism of High-Power Semiconductor Lasers

A Deep Learning Method for Damage Prognostics of Fiber-Reinforced Composite Laminates Using Acoustic Emission

Prognostics of Radiation Power Degradation Lifetime for Ultraviolet Light-Emitting Diodes Using Stochastic Data-Driven Models

Experimental Investigation of Machine Learning based Soft-Failure Management using the Optical Spectrum

Federated Learning Approach for Lifetime Prediction of Semiconductor Lasers

Physics-Informed Machine Learning for Degradation Modeling of an Electro-Hydrostatic Actuator System

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

Degradation Reliability Modeling for Nonlinearly Recoverable Products Under Random Shock

Degradation-based residual life prediction under different environments

Fusing physics-inferred information from stochastic model with machine learning approaches for degradation prediction

Vertical Cavity Surface Emitting Laser Performance Maturing through Machine Learning for High-Yield Optical Wireless Network

A Hybrid Physics-Informed Machine Learning Approach for Time-Dependent Reliability Assessment of Electromagnetic Relays

Gradual failure prediction of electronic equipment based on performance degradation data

Physics‐based and data‐driven approaches for lifetime estimation under variable conditions: Application to organic light‐emitting diodes

Physics-Informed Deep Learning-Based Approach for Probabilistic Modeling of Degradation

Fast Modeling of the Capacity Degradation of Lithium-Ion Batteries Via a Conditional Temporal Convolutional Encoder–Decoder

Predicting VCSEL Emission Properties Using Transformer Neural Networks