Local Fault Prediction for Equipment Installation Quality

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Inconsistent installation of equipment leads to varying degrees of correctness, resulting in potential early failure, reduced performance, or progressive degradation, which is not readily identified during initial operation, particularly in utilities industries like telecommunications.

Innovation Solution

A computer-implemented method using machine learning models to predict the quality of equipment installation by applying detectors to equipment and external elements, obtaining performance and environmental data, and outputting a faulty installation likelihood predictor to recommend corrective actions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual installation auditing is performed by operatives, then installation correctness can be verified, but the scale of auditing is limited by the number of auditor operatives available

Engineering Contradiction:
Improveinstallation correctness verificationVSAvoidauditing scale
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces manual auditing by human operatives with an automated machine learning-based detection system. The system uses detectors and machine learning models to automatically analyze installation images and identify defects, eliminating the need for human auditors to physically inspect each installation. This substitution enables scaling to thousands of installations without being constrained by the number of available auditor operatives, while maintaining or improving verification precision through consistent application of detection algorithms.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If equipment is installed quickly to meet demand, then installation productivity increases, but installation quality and correctness may deteriorate

Engineering Contradiction:
Improveinstallation speedVSAvoidinstallation correctness
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent implements automated detection and prediction systems that operate during or immediately after installation to identify potential defects before they manifest as failures. The machine learning models analyze installation data in real-time or near-real-time, providing feedback that allows corrective actions to be taken promptly. This preliminary detection approach enables fast installation processes to maintain high quality standards by catching errors early without requiring slow manual inspection of each installation.

Inventive Principle:
Principle #10Preliminary action

3Reliability

If comprehensive installation auditing is performed manually, then installation quality improves, but time and resources required for auditing increase

Engineering Contradiction:
Improveinstallation qualityVSAvoidauditing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces time-consuming manual auditing processes with automated machine learning-based detection systems that can process multiple installations simultaneously and instantaneously. The system uses pre-trained models to rapidly analyze installation images and data, providing quality assessment results in real-time or near-real-time without the delays inherent in manual inspection. This automation maintains comprehensive quality checking while eliminating the time loss associated with sequential manual auditing of each installation.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4671889A1Local fault predictor
Publication Date: 2025.12.31 BRITISH TELECOM PLC
  • EP4671889A1 patent drawingFigure 1
  • EP4671889A1 patent drawingFigure 2
  • EP4671889A1 patent drawingFigure 3~4

AI summary

A method comprising the steps: (a) applying a first detector to equipment; (b) applying a second detector to at least one external element; (c) obtaining a performance measurement associated with the equipment at time of installation; (d) obtaining data associated with the environment in which the equipment is being installed; (e) obtaining historical fault data associated with the equipment and/or environment; (f) inputting, to a machine learning model, at least some of the data and/or information, obtained in steps (a)-(e); (g) outputting, from the machine learning model, a faulty installation likelihood predictor configured to apply an overall confidence level to the installed equipment; (h) if the overall confidence level is below a predetermined threshold, recommending a corrective action until the overall confidence level is above a second predetermined threshold