A method and system for electromagnetic signal data enhancement based on variational mode recombination
CN122241018APending Publication Date: 2026-06-19ARTIFICIAL INTELLIGENCE INNOVATION RES INST OF ZHEJIANG UNIV OF TECH BINJIANG DISTRICT HANGZHOU
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ARTIFICIAL INTELLIGENCE INNOVATION RES INST OF ZHEJIANG UNIV OF TECH BINJIANG DISTRICT HANGZHOU
- Filing Date
- 2026-02-03
- Publication Date
- 2026-06-19
AI Technical Summary
Technical Problem
Existing deep learning-based automatic modulation recognition models have poor generalization ability on small datasets and low recognition accuracy in complex interference environments. Existing data augmentation methods cannot meet training requirements.
Method used
The variational mode reconstruction method is used to preprocess, decompose and reconstruct electromagnetic signals to generate expanded samples, which are then combined with the original samples to form an expanded dataset for training the model.
Benefits of technology
It improves the model's generalization ability and robustness, and increases the signal recognition accuracy on low-sample datasets.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122241018A_ABST
Abstract
This invention discloses an electromagnetic signal data augmentation method and system based on variational mode recombination, belonging to the field of electromagnetic signal processing technology. The specific steps are as follows: receiving the original electromagnetic signal and preprocessing the signal to be augmented; adaptively decomposing the preprocessed signal into a series of intrinsic mode functions (IMFs) with center frequencies using a variational mode decomposition algorithm to extract multi-scale intrinsic features from the signal; cross-recombining and linearly superimposing the IMFs of different signals to generate new and diverse electromagnetic signal samples by recombining the modal components of different samples, thereby achieving data augmentation and generating expanded samples; mixing the expanded samples with the original samples to form an expanded dataset for training a signal recognition model. This invention, by mixing the modal components of different signals, can generate a large number of diverse and realistic new samples for signal recognition model training while preserving essential features, effectively improving the model's generalization ability.
Need to check novelty before this filing date? Find Prior Art