A method and system for multi-target state detection based on a single sensor
By acquiring mixed signals with a single sensor and performing mode decomposition, adaptive filtering, and multimodal decoupling, the hardware deployment difficulties and signal aliasing problems of multi-target detection in confined spaces are solved, and accurate identification of multi-target states is achieved.
CN121744101BActive Publication Date: 2026-05-26JIANGXI NORMAL UNIV
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- JIANGXI NORMAL UNIV
- Filing Date
- 2026-02-14
- Publication Date
- 2026-05-26
AI Technical Summary
Technical Problem
Existing technologies cannot effectively deploy multiple sensors in confined spaces, making multi-target detection difficult, and single-sensor methods cannot accurately distinguish the states of multiple signal sources.
Method used
A single passive sensor is used to collect mixed signals. The signals are separated by mode decomposition, adaptive filtering and multi-mode mixed signal decoupling model. Combined with Mel-frequency cepstral feature extraction and multi-output classification, multi-target state recognition is achieved.
Benefits of technology
Achieving precise separation and feature extraction of multi-target signals in confined spaces improves the reliability and accuracy of detection, simplifies the system structure, and reduces power consumption.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN121744101B_ABST
Abstract
This application relates to the field of sensor signal processing technology, specifically disclosing a method and system for multi-target state detection based on a single sensor. The method includes acquiring mixed target signals, decomposing the mixed target signals into multiple modal components, determining the correlation coefficients between multiple intrinsic mode function components and the mixed original signal, and determining the number of target modes based on the correlation coefficients; fusing the remaining intrinsic mode function components to obtain a denoised mixed signal; converting the denoised mixed signal into time-frequency vector data, inputting the time-frequency vector data into a multi-target signal decoupling model based on variational autoencoder for decoupling, and outputting a demixed source signal; inputting the demixed source signal into an adaptive filtering model to obtain a global average value, and determining the Mel-frequency cepstral feature vector of each target source based on the global average value; inputting the Mel-frequency cepstral feature vector into a multi-output classification model to output the state recognition result. The method realizes state detection of multiple targets in a confined space based on a single sensor.
Need to check novelty before this filing date? Find Prior Art