Part abnormal sound detection method and device and electronic equipment
By calculating the cumulative energy generation time mask of each frame in the spectrum image, audio event images are extracted, and audio event categories are identified based on multi-dimensional features. This solves the misjudgment problem in the detection of abnormal noises in components in the existing technology, and achieves higher accuracy and anti-interference detection effect.
Patent Information
- Application Number
- CN202610568231.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-27
- Publication Date
- 2026-07-17
AI Technical Summary
Existing methods for detecting abnormal noises from components are difficult to effectively separate the corresponding audio events in complex audio environments and are prone to misjudgment.
By calculating the cumulative energy of each frame in the spectral image, a time mask is generated, audio event images are extracted, and the category of audio events is identified based on multidimensional features, including energy distribution, texture and shape features, and then classified using a neural network model.
It improves the accuracy and anti-interference ability of abnormal noise detection, reduces the probability of misjudging environmental noise and track impact as abnormal noise of parts, and improves the accuracy and reliability of detection.
Smart Images
Figure CN122417075A_ABST