Optical Module Fault Prediction Using Classification Thresholds

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing optical module fault prediction methods are inadequate in accurately predicting faults before they occur, leading to delayed maintenance and service disruptions in large data centers.

Innovation Solution

A method that determines a classification threshold based on a classification sample set for operating parameters like bias current and receive power, using a fault prediction model to predict the likelihood and urgency of faults, allowing for timely maintenance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional fault monitoring methods are used to detect optical module faults, then faults can be detected after they occur, but the prediction accuracy is insufficient and maintenance cannot be performed proactively

Engineering Contradiction:
Improvefault prediction accuracyVSAvoidservice continuity
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent applies preliminary action by establishing classification thresholds and fault prediction models before faults occur. The system proactively analyzes operating parameters (bias current, receive power) against pre-defined thresholds and trained models to predict potential faults before they manifest, enabling preventive maintenance rather than reactive response. This is evident in the steps of determining classification thresholds based on historical data and using these thresholds to predict faults before service disruption occurs.

Inventive Principle:
Principle #10Preliminary action

2Loss of time

If classification thresholds are established based on classification sample sets to predict faults, then early fault identification is enabled, but the system complexity increases due to threshold determination and model training requirements

Engineering Contradiction:
Improvemaintenance response timeVSAvoidprediction system complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The system applies self-service by automatically determining classification thresholds and training fault prediction models using historical operating data without requiring manual expert intervention. The system autonomously collects bias current and receive power data, processes this data to establish thresholds, and continuously refines prediction models. This automation reduces the complexity burden on operators while maintaining high prediction accuracy, enabling the system to serve itself in the fault prediction process.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If multiple operating parameters are monitored continuously to improve prediction accuracy, then fault prediction capability is enhanced, but the energy consumption and measurement resources increase

Engineering Contradiction:
Improveparameter measurement accuracyVSAvoidmonitoring system energy consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent applies the extraction principle by selectively focusing monitoring resources on the two most critical operating parameters: bias current and receive power. Rather than continuously monitoring all possible optical module parameters, the system extracts and analyzes only these key parameters that have the strongest correlation with fault occurrence. This selective approach maintains high prediction accuracy while minimizing energy consumption and measurement resource requirements by concentrating on the most informative parameters.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP4037286B1Method and apparatus for predicting fault of optical module
Publication Date: 2024.10.30 HUAWEI TECH CO LTD
  • EP4037286B1 patent drawingFigure 1
  • EP4037286B1 patent drawingFigure 2
  • EP4037286B1 patent drawingFigure 3~4

AI summary

This application provides a method and an apparatus for predicting a fault of an optical module. In this application, a classification threshold of an operating parameter is determined based on a classification sample set corresponding to the operating parameter of optical modules; and whether a fault occurs in the future on an optical module corresponding to a sequence to be detected is predicted based on comparison results between the classification threshold and a plurality of measured values in the sequence to be detected, to help a maintenance engineer focus on an optical module on which a fault may occur, and reduce impact caused by the fault of the optical module on a service.