传感技术学报2025，Vol.38Issue(11)：1969-1977,9.DOI:10.3969/j.issn.1004-1699.2025.11.007
基于FCN与BiLSTM的心冲击信号J峰提取方法研究
Research on J-Peak Extraction Method of Ballistocardiogram Signal Based on FCN and BiLSTM
摘要
Abstract
Ballistocardiogram(BCG)technology is one of the promising approaches in non-invasive monitoring of vital signs.Accurately extracting J peaks from BCG signals is of great importance for calculating BCG based vital sign indicators.A deep learning model based on full convolutional neural network(FCN)and bidirectional long short-term memory(BiLSTM)network is proposed to extract the J peak from BCG signals,and a J peak correction algorithm is cooperatively designed.Compared with reported typical methods,the proposed method requires less training data and has no limitation on the BCG signal length and heart rate range,so it is applicable for a wider range of practical application scenarios.To verify the feasibility and effectiveness of the proposed method,BCG signals of 22 randomly-selected subjects are collected and processed.The test results show that the proposed J-peak extraction method achieves the best recog-nition accuracy and better robustness,which provides an optimized solution for the J-peak extraction of the BCG signal.关键词
心冲击信号/J峰提取/全卷积神经网络/双向长短期记忆网络Key words
ballistocardiogram/J-peak extraction/fully convolutional network/bidirectional long short-term memory network分类
天文与地球科学引用本文复制引用
杨国伟,尹明杰,樊冰,崔守毅,何羽恒,周雪芳,毕美华,胡淼..基于FCN与BiLSTM的心冲击信号J峰提取方法研究[J].传感技术学报,2025,38(11):1969-1977,9.基金项目
国家自然科学基金项目(52175460) （52175460）
