Drone Audio Graph Analysis for Automated Spinning Equipment Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the spinning process, the long and complex nature of the process along with numerous devices involved makes manual fault detection labor-intensive and inefficient, necessitating an automated solution for device fault detection.
Innovation Solution
An audio-based device fault detection method and apparatus that utilizes a drone to collect initial audio data from target devices, preprocesses the data, extracts features, constructs an information graph, and employs a graph neural network model to obtain fault detection results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual troubleshooting method is used, then fault detection can be performed, but it is labor-intensive and inefficient
Solution Approach 1:
The patent replaces manual mechanical troubleshooting with an automated audio-based detection system. Drones equipped with microphones collect audio data from devices, and a graph neural network model automatically analyzes the data to identify faults, eliminating the need for manual inspection and significantly improving detection efficiency while reducing labor intensity.
Solution Approach 2:
The system enables devices to self-diagnose faults through audio analysis. The graph neural network model processes audio data collected by drones and automatically generates fault detection results, allowing the system to perform self-monitoring and self-diagnosis without human intervention, thereby improving productivity and reducing operational complexity.
2Extent of automation
If automated fault detection is implemented, then labor resources are saved, but system complexity increases
Solution Approach 1:
The patent introduces audio data as an intermediary medium between the physical device state and the fault detection system. Drones collect audio signals that indirectly reflect device conditions, and the graph neural network model processes these audio features to determine faults. This intermediary approach simplifies the automation process by converting complex physical state monitoring into audio signal analysis.
Solution Approach 2:
The system replaces complex mechanical inspection procedures with audio-based sensing and computational analysis. Instead of manually examining device components, the system uses audio data collected by drones and processed by graph neural networks to automatically detect faults, reducing system operational complexity while maintaining high automation levels.
3Loss of time
If audio-based detection is used, then real-time fault identification is achieved, but data processing complexity increases
Solution Approach 1:
The patent segments the audio data processing into distinct stages: audio data collection by drones, audio feature extraction, construction of information graphs from audio features, and fault detection using graph neural networks. This segmentation allows real-time processing by breaking down complex audio analysis into manageable computational steps, reducing overall processing time while managing data processing complexity through structured methodology.
Solution Approach 2:
The system performs preliminary audio feature extraction and information graph construction before final fault detection. By pre-processing audio data into structured graph representations, the system prepares data in advance for rapid analysis by the graph neural network model, enabling real-time fault identification while managing processing complexity through staged computation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure provides an audio-based device fault detection method and apparatus and a related device, and relates to the field of data processing and in particular to technical fields of deep learning and voice technology. The method comprises: obtaining (S101) initial audio data collected by a drone for a target device; preprocessing (S102) the initial audio data to obtain audio data to be detected; performing (S103) feature extraction on the audio data to be detected to obtain an audio feature of the audio data to be detected; constructing (S104) an information graph based on the audio feature; and obtaining (S105) a fault detection result for the target device based on the information graph and a graph neural network model.