A method, device and medium for monitoring usage of a data center network device
By using an interface usage detection model in the inspection robot, combined with a neural network structure with multi-scale features, the problem of low accuracy and recall in detecting abnormal use of data center network equipment interfaces has been solved, enabling accurate monitoring and timely detection of interface usage status.
Patent Information
- Application Number
- CN202211285869.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-20
- Publication Date
- 2025-12-19
- Estimated Expiration
- 2042-10-20
AI Technical Summary
The existing data center network equipment abnormal usage behavior detection accuracy and recall rate are not high. In particular, the visual changes of abnormal interface usage are small and there are no clear trace characteristics, making it impossible to detect abnormal interface usage problems in a timely and accurate manner.
An inspection robot is used to acquire images of the interface in use and input them into a pre-trained interface use detection model along with a pre-set template image. An initial model is constructed by fusing multi-scale features into a neural network structure, generating a detection annotation file and determining the interface use status, including abnormal use status and normal use status.
It improves the detection accuracy of interface misuse, can promptly identify interface misuse issues of data center network equipment, and quantitatively displays the visual changes and ambiguous trace characteristics caused by misuse.