Hierarchical Fault Prediction for Installation Quality Auditing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inconsistent installation of equipment leads to varying degrees of correctness, resulting in potential curtailed lifespan, reduced performance, or susceptibility to 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 analyzing network hierarchies, detector outputs, performance measurements, environmental data, and historical fault data, and outputting a faulty installation likelihood predictor with corrective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual installation auditing is performed, then installation correctness can be verified, but the scale of auditing is limited due to fewer auditor operatives compared to installations
Solution Approach 1:
The patent replaces manual mechanical auditing with automated electronic detection systems including image capture devices, sensors, and machine learning models that automatically verify installation correctness, enabling scale expansion without proportional increase in auditor operatives
Solution Approach 2:
The patent introduces automated detection systems and machine learning models as intermediaries between installation operatives and human auditors, allowing most auditing to be performed automatically while human auditors focus on complex cases
2Reliability
If comprehensive installation checks are performed, then potential issues can be identified, but installation time and complexity increase
Solution Approach 1:
The patent performs detection and verification actions during the installation process itself rather than as a separate post-installation step, capturing images and sensor data in real-time to identify issues before they become problems
Solution Approach 2:
The patent implements real-time feedback loops where detection systems continuously monitor installation parameters and provide immediate feedback to operatives, allowing corrections to be made during installation rather than requiring rework later
3Measurement precision
If installation correctness is verified immediately, then errors can be caught early, but not all potential future failures can be predicted
Solution Approach 1:
The patent performs preliminary detection of installation conditions and potential risk factors during installation, using machine learning models to predict future failures based on detected parameters before they manifest as actual problems
Solution Approach 2:
The patent identifies potential future failure modes during installation and implements preventive measures or flags for monitoring, cushioning against future failures before they occur by addressing root causes during installation
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method comprising: (a) obtaining a hierarchy of two or more devices in a network, wherein the hierarchy describes connectivity of the two or more devices and/or describes relationship between the two or more devices; (b) obtaining a first detector's output of each of the two or more devices at a time of installation; (c) obtaining a second detector's output of an external element of the two of more devices at the time of installation; (d) obtaining performance measurement(s) of each of the two or more devices at the time of installation; (e) obtaining environmental information of each of the two or more devices at the time and/or a location of installation; (f) obtaining historical fault data associated with the network, device(s) and/or environment; (g) obtaining network performance metrics associated with the network; (h) inputting, to a machine learning model, at least some of the data, metrics and/or information, obtained in steps (a)-(g); (i) outputting, from the machine learning model, a faulty installation likelihood predictor configured to apply an overall confidence level to the installed equipment; (j) 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.