基于Transformer的心电信号降噪方法

CN115919325BActive Publication Date: 2026-07-17WUHAN UNIV
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
WUHAN UNIV
Filing Date
2022-11-16
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing ECG signal noise reduction technologies are unable to effectively handle different types and intensities of noise, resulting in poor noise reduction performance.

Method used

We construct a Transformer-based deep neural network, including an encoder and a decoder. We learn the temporal features of ECG signals through a self-attention module and a locally enhanced feedforward layer, and generate abstract features using a multi-head self-attention layer and a feedforward layer. We then optimize the model by combining mean squared error and smoothing loss.

Benefits of technology

It improves the noise reduction effect for different types of noise, enhances the efficiency and accuracy of ECG signal processing, and optimizes the noise reduction performance of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115919325B_ABST
    Figure CN115919325B_ABST
Patent Text Reader

Abstract

本发明公开了一种基于Transformer的心电信号降噪方法，包括：获取无噪心电图信号数据集ECG_D1和纯噪声数据集Noise_D；利用Noise_D对ECG_D1中的心电图信号数据进行加噪处理得到带噪声的心电信号数据，并将其分为训练集和验证集；构建基于Transformer的深度神经网络；将步骤S2得到的带有噪声的心电信号训练集送入步骤S3构建的深度神经网络中训练，直至训练至达到终止条件，并采用验证集验证验证得到优化后的深度神经网络并将其作为心电信号降噪模型M；利用心电信号降噪模型M对真实的心电信号进行降噪处理，得到去噪后的心电数据用于下游任务。本发明能够学习心电信号的时序长期依赖性以及局部相关性，且能迅速地对不同种类的心电信号数据进行降噪处理，提高了效率与准确率。
Need to check novelty before this filing date? Find Prior Art