Training method of operation state recognition model, and operation state recognition method and device

By deploying acoustic sensors around the switch machine, collecting and analyzing acoustic signals, extracting effective acoustic signature features, and using machine learning models for automated monitoring of the switch machine, the problems of low efficiency and missed detection in manual inspections are solved, and efficient and accurate switch machine status identification is achieved.

CN121034342APending Publication Date: 2025-11-28BEIJING ZHONGKE DONGREN TECH CO LTD
View PDF 6 Cites 1 Cited by

Patent Information

Application Number
CN202511035267.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-25
Publication Date
2025-11-28

AI Technical Summary

Technical Problem

In the existing technology, the operation and status monitoring of switch machines mainly relies on manual periodic inspections, which has the problems of low efficiency, inability to capture anomalies in a timely manner, and easy to miss detection.

Method used

Acoustic signals are collected by acoustic sensors deployed around the switch machine, effective acoustic signature features are extracted, and the operation status recognition model is used for action and status monitoring, including frequency band features and training of machine learning models, to achieve automated monitoring of the switch machine.

Benefits of technology

It achieves efficient and accurate monitoring of switch machine operation and status, reduces labor costs, captures anomalies in a timely manner, avoids missed detections, and improves monitoring efficiency and accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121034342A_ABST
    Figure CN121034342A_ABST
Patent Text Reader

Abstract

The invention discloses a training method of a running state recognition model and a running state recognition method and device, and belongs to the technical field of audio processing. According to the invention, the acoustic sensor arranged around the target equipment is used for collecting the acoustic signal, and the action and state of the target equipment are monitored based on the collected acoustic signal. Because the monitoring mode does not need manual participation, compared with a monitoring mode of manual regular inspection, the manpower cost is saved, the efficiency is higher, and the abnormity of the target equipment can be captured in time. Besides, the machine learning model can be trained to recognize the action and state of the target equipment, effective voiceprint features used for operation state recognition can be extracted in the model training process, and the effective voiceprint features comprise frequency band features capable of representing the action features of the target equipment. Therefore, the effective voiceprint features serve as input of the model to participate in model training, and the above processing mode can greatly improve the recognition accuracy of the model.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Emotion recognition method and device, computer equipment and storage medium

    CN109003624A

  • Fault detection method and device, electronic equipment and readable storage medium

    CN116052653A

  • OLTC running state monitoring method and device and electronic equipment

    CN118522310A

  • Gearbox noise monitoring method

    CN119905107A

  • Voice Anti-counterfeiting method and apparatus, terminal device, and storage medium

    WO2021103913A1

Cited By

  • Multi-source fusion characterization method for low-frequency acoustic state of train passenger compartment

    CN122224223A