Frequency spectrum back noise prediction method based on multi-scale attention mechanism and domain self-adaption

By employing a multi-scale attention mechanism and a domain-adaptive spectral background noise prediction method, the problems of parameter sensitivity and multi-site universality in background noise extraction in radio spectrum monitoring are solved, achieving high-precision noise prediction and stable signal processing in complex environments.

CN122241044APending Publication Date: 2026-06-19ZHEJIANG YUANCHU DATA TECH CO LTD
0 Cites 1 Cited by

Patent Information

Application Number
CN202610153785.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-02-03
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing technologies for background noise extraction in radio spectrum monitoring suffer from problems such as high parameter sensitivity, poor versatility across multiple sites, and limited feature extraction capabilities, leading to noise baseline distortion and affecting signal processing performance.

Method used

A spectral background noise prediction method based on multi-scale attention mechanism and domain adaptation is adopted. By improving the convolutional neural network and combining multi-scale feature extraction, channel attention mechanism, domain adaptation technology and composite loss function, data preprocessing, model training and postprocessing are performed to achieve accurate prediction of background noise.

Benefits of technology

It achieves high-precision and robust prediction of background noise in complex spectrum environments, adapts to different devices and electromagnetic environments, reduces deployment and maintenance costs, and generates a noise baseline that conforms to physical laws, avoiding the step effect of traditional algorithms.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a spectrum back noise prediction method based on a multi-scale attention mechanism and domain adaptation, and aims to provide the spectrum back noise prediction method based on the multi-scale attention mechanism and domain adaptation, which can realize accurate prediction of background noise in a complex spectrum environment. Based on an improved convolutional neural network, a multi-scale attention mechanism and a domain adaptive technology are fused, and through four core steps of data preprocessing, model construction, model training, prediction and post-processing, and through model structure innovation and training strategy optimization, accurate prediction of background noise in a complex frequency spectrum environment is realized. The method has the advantages of being high in full-band adaptability, good in multi-site universality, outstanding in robustness, smooth and natural in prediction curve, suitable for scenes of radio frequency spectrum monitoring, signal detection, signal-to-noise ratio calculation, spectrum situation sensing and the like, and capable of effectively adapting to complex and changeable spectrum environments and multi-site monitoring requirements.
Need to check novelty before this filing date? Find Prior Art