A channel adaptive hybrid weighting low slow small unmanned aerial vehicle acoustic detection method

By employing a channel-adaptive hybrid weighting method, combined with convolutional neural modules and Transformer encoders, the problems of long-term feature modeling and cross-domain generalization in the acoustic detection of low-speed, small UAVs are solved, achieving high-precision, low-complexity real-time detection.

CN122201340APending Publication Date: 2026-06-12HANGZHOU EBOYLAMP ELECTRONICS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HANGZHOU EBOYLAMP ELECTRONICS CO LTD
Filing Date
2026-02-03
Publication Date
2026-06-12

AI Technical Summary

⚠Technical Problem

Existing acoustic detection technologies for low-speed, small unmanned aerial vehicles (UAVs) suffer from insufficient long-term feature modeling capabilities, weak cross-domain generalization capabilities, and a conflict between computational resources and real-time performance in complex environments, resulting in insufficient detection accuracy and generalization capabilities.

⚗Method used

We employ a channel-adaptive hybrid weighting method, combining convolutional neural modules and Transformer encoders, to extract time-frequency features and perform temporal modeling. Furthermore, we enhance the robustness and generalization ability of the model through adaptive weighting and data augmentation strategies.

🎯Benefits of technology

It improves the accuracy of acoustic detection for low-speed, small UAVs and the adaptability of the model to different devices and complex environments, reduces computational complexity, and is suitable for real-time detection on edge computing devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122201340A_ABST
    Figure CN122201340A_ABST
Patent Text Reader

Abstract

The application discloses a kind of passage adaptive hybrid weighting low slow small unmanned plane acoustic detection methods, comprising: obtaining environmental real-time audio stream, using sliding window to frame processing environmental real-time audio stream, obtain multiple preset length audio segments;The passage adaptive hybrid weighting low slow small unmanned plane acoustic detection method of the application is through convolution neural module and Transform encoder module, both capture the unique high-frequency rotor voiceprint details of unmanned plane, and utilize the time continuity of signal, compared with single network structure has higher feature expression ability and detection accuracy;And the convolution layer of last convolution module is compressed in frequency dimension to feature map, significantly reduce the input sequence length and computational complexity of Transform encoder module, so that the algorithm model has the characteristics of small parameter quantity, fast inference speed, very suitable for deployment in the edge computing device or embedded system with limited computing power for real-time detection.
Need to check novelty before this filing date? Find Prior Art