基于深度学习和注意力机制的心电信号分类方法研究

版权申诉

毕业论文

79 浏览量 2023-10-26 00:01:14 上传评论收藏 49.1MB PDF 举报

资源推荐

资源详情

资源评论

基于深度学习和注意力机制的心电信号分类方法研究

III

1 绪论 ................................................................................................................. 1

1.1 研究背景与意义 .................................................................................... 1

1.2 国内外研究现状 .................................................................................... 1

1.2.1 ECG 数据集不平衡处理研究现状 ............................................. 1

1.2.2 ECG 信号分类算法研究现状 ..................................................... 2

1.3 本文主要研究内容 ................................................................................ 4

1.4 本文组织结构 ........................................................................................ 4

2 ECG 信号分类技术与理论 ............................................................................ 6

2.1 引言 ........................................................................................................ 6

2.2 ECG 信号相关知识 ............................................................................... 6

2.3 ECG 信号不平衡处理算法 ................................................................... 7

2.3.1 SMOTE 算法 ............................................................................... 7

2.3.2 ADASYN 算法 ............................................................................ 8

2.4 深度学习网络 ........................................................................................ 9

2.4.1 1D-CNN 网络 .............................................................................. 9

2.4.2 Bi-LSTM 网络 ............................................................................ 11

2.5 注意力机制 .......................................................................................... 13

2.5.1 注意力机制原理 ........................................................................ 13

2.5.2 注意力机制形式 ........................................................................ 14

2.6 本章小结 ................................................................................................ 15

3 ECG 信号预处理 .......................................................................................... 16

3.1 引言 ...................................................................................................... 16

3.2 ECG 信号及国际标准数据库 ............................................................. 16

3.2.1 数据库介绍和心拍分类标准 .................................................... 16

3.2.2 信号噪声干扰分析 .................................................................... 17

3.3 ECG 信号预处理过程 ......................................................................... 17

3.3.1 ECG 信号去噪 ........................................................................... 18

3.3.2 QRS 波检测 ............................................................................... 19

3.3.3 心拍分割 .................................................................................... 20

3.3.4 数据不平衡处理 ........................................................................ 21

3.4 本章小结 .............................................................................................. 23

4 基于 PSTA-Net 模型的 ECG 信号分类 ...................................................... 24

4.1 引言 ...................................................................................................... 24

4.2 PSTA-Net 模型总体设计 .................................................................... 24

4.3 PSTA-Net 模型注意力模块设计 ........................................................ 26

4.3.1 基于门控机制和通道注意力机制的 GCA Block .................... 26

4.3.2 基于门控机制和时间步注意力机制的 GTSA Block .............. 27

4.4 实验结果与分析 .................................................................................. 28

4.4.1 实验平台及损失函数 ................................................................ 28

4.4.2 实验设置及评估指标 ................................................................ 28

4.4.3 结果与对比分析 .......................................................................... 31

4.5 本章小结 .............................................................................................. 36

5 基于 FA-Net 模型的 ECG 信号分类 ........................................................... 38

5.1 引言 ...................................................................................................... 38

5.2 FA-Net 模型总体设计 ......................................................................... 38

5.3 FA-Net 子模块设计 ............................................................................. 40

5.3.1 位置编码模块 ............................................................................ 40

5.3.2 多头注意力模块 ........................................................................ 40

5.3.3 密集插值模块 ............................................................................ 42

5.4 实验结果与分析 .................................................................................. 43

5.4.1 实验平台及损失函数 ................................................................ 43

5.4.2 实验设置及评估指标 ................................................................ 43

5.4.3 结果与对比分析 ........................................................................ 45

5.5 本章小结 .............................................................................................. 48

6 总结与展望 ................................................................................................... 50

6.1 总结 ...................................................................................................... 50

6.2 展望 ...................................................................................................... 50

参考文献 ....................................................................................................... 52

攻读硕士研究生期间科研成果与奖励 ............................................................. 57

致谢 ............................................................................................................. 57

基于深度学习和注意力机制的心电信号分类方法研究

TABLE OF CONTENTS

1 Introduction ..................................................................................................... 1

1.1 Background and significance of the study ............................................. 1

1.2 Current status of domestic and international research ........................... 1

1.2.1 Status of Research on Imbalance Processing of ECG Datasets ... 2

1.2.2 Status of Research on ECG Signal Classification Algorithm ...... 2

1.3 Main research content of this paper ....................................................... 4

1.4 Organization of this paper ...................................................................... 5

2 ECG Signal Classification Techniques and Theory ........................................ 6

2.1 Introduction ............................................................................................ 6

2.2 Knowledge about ECG signals .............................................................. 6

2.3 ECG signal imbalance processing algorithms ........................................ 7

2.3.1 The SMOTE algorithm ................................................................. 7

2.3.2 The ADASYN algorithm .............................................................. 9

2.4 Deep Learning Networks ....................................................................... 9

2.4.1 1D-CNN network ......................................................................... 9

2.4.2 Bi-LSTM networks .................................................................... 13

2.5 Attentional Mechanisms ....................................................................... 15

2.5.1 Principle of the attention mechanism ......................................... 15

2.5.2 Forms of Attention Mechanisms ................................................ 16

2.6 Summary of this chapter ...................................................................... 15

3 ECG signal pre-processing ............................................................................ 18

3.1 Introduction .......................................................................................... 18

3.2 ECG signals and the international standard database ........................... 18

3.2.1 Introduction to the database and heartbeat classification criteria18

3.2.2 Signal Noise Interference Analysis ............................................ 19

3.3 ECG signal pre-processing process ...................................................... 20

3.3.1 ECG signal denoising ................................................................. 20

3.3.2 QRS wave detection ................................................................... 22

3.3.3 Heartbeat segmentation .............................................................. 23

3.3.4 Data imbalance processing ......................................................... 24

4 ECG signal classification based on the PSTA-Net model ............................. 24

4.1 Introduction .......................................................................................... 24

TABLE OF CONTENTS

4.2 Overall design of the PSTA-Net model ................................................ 24

4.3 Design of the PSTA-Net model attention module ................................ 26

4.3.1 GCA Block ................................................................................. 26

4.3.2 GTSA Block ............................................................................... 27

4.4 Experimental Results and Analysis ...................................................... 28

4.4.1 Experimental Platform and Loss Function ................................. 28

4.4.2 Experimental Setup and Evaluation Metrics .............................. 30

4.4.3 Results and Comparative Analysis ............................................. 31

4.5 Summary of this chapter ...................................................................... 39

5 Classification of ECG signals based on the FA-Net model .......................... 38

5.1 Introduction ......................................................................................... 38

5.2 Overall design of the FA-Net model .................................................... 38

5.3 FA-Net sub-module design ................................................................... 42

5.3.1 Location coding module ............................................................. 42

5.3.2 Multi-headed attention module .................................................. 42

5.3.3 Dense interpolation module ....................................................... 44

5.4 Experimental results and analysis ........................................................ 45

5.4.1 Experimental platform and loss function ................................... 45

5.4.2 Experimental Setup and Evaluation Metrics .............................. 46

5.4.3 Results and Comparative Analysis ............................................. 48

5.4 Summary of this chapter ...................................................................... 51

6 Summary and outlook ................................................................................... 52

6.1 Summary .............................................................................................. 52

6.2 Outlook ................................................................................................. 53

References ........................................................................................................... 54

Research achievements and awards during the Master's degree ......................... 59

Acknowledgements ............................................................................................. 60

基于深度学习和注意力机制的心电信号分类方法研究

VII

图目录

图 2.1 心电图 ............................................................................................................................. 6

图 2.2 单个周期心电波形 ........................................................................................................ 7

图 2.3 SMOTE 算法 .................................................................................................................. 8

图 2.4 一维卷积计算过程 ...................................................................................................... 10

图 2.5 激活函数计算过程 ...................................................................................................... 11

图 2.6 激活函数 ....................................................................................................................... 11

图 2.7 池化操作 ...................................................................................................................... 12

图 2.8 LSTM 模型内部结构 ................................................................................................... 13

图 2.9 Bi-LSTM 模型内部结构 .............................................................................................. 14

图 2.10 注意力机制结构 ........................................................................................................ 15

图 3.1 含噪声的原始心电信号 .............................................................................................. 19

图 3.2 ECG 信号预处理流程 .................................................................................................. 20

图 3.3 小波方法去噪效果对比 .............................................................................................. 21

图 3.4 自适应阈值 QRS 波检测 ............................................................................................ 22

图 3.5 五种类型心电信号热度图 .......................................................................................... 23

图 3.6 ECG 信号数据集划分比例 .......................................................................................... 24

图 4.1 PSTA-Net 模型结构 ..................................................................................................... 26

图 4.2 GCA Block 具体结构 ................................................................................................... 28

图 4.3 TSA Block 具体结构 .................................................................................................... 29

图 4.4 不同尺度的心拍分类结果对比 .................................................................................. 35

图 4.5 五折交叉验证的分类结果对比 .................................................................................. 35

图 4.6 PSTA-Net 模型收敛曲线 ............................................................................................. 36

图 4.7 心拍分类结果的混淆矩阵 .......................................................................................... 37

图 5.1 FA-Net 模型的结构 ...................................................................................................... 40

图 5.2 多头注意力结构 .......................................................................................................... 43

图 5.3 密集插值 ...................................................................................................................... 45

图 5.4 FA-Net 模型收敛曲线 .................................................................................................. 48

图 5.5 相关参数寻优测试 ....................................................................................................... 49

剩余61页未读，继续阅读

评论收藏

内容反馈

版权申诉

2301_77550592

粉丝: 17
资源: 7163

基于深度学习和注意力机制的心电信号分类方法研究_毕业论文.pdf

最新资源

基于深度学习和注意力机制的心电信号分类方法研究_毕业论文.pdf

一种基于深度学习的心电信号分类方法.pdf

基于深度学习的心电信号ST段识别方法研究_毕业论文.pdf

基于深度学习的心电信号心律失常分类方法研究_毕业论文.pdf

基于混合深度学习算法的ECG心电信号分类研究_毕业论文.pdf

基于深度自编码器的心电信号降噪方法研究_毕业论文.pdf

基于SVM分类Ⅱ导联心电信号的方法研究_毕业论文.pdf

基于生成对抗网络的心电信号降噪方法研究_毕业论文.pdf

基于超限学习机的不平衡主动学习心电信号识别方法研究_毕业论文.pdf

基于时间卷积编解码网络的胎儿心电信号提取方法研究_毕业论文.pdf

基于深度迁移学习的心电信号分类方法.pdf

基于深度学习的肌电信号预测的研究.pdf

基于深度学习和模糊C均值的心电信号分类方法.pdf

基于方差和深度学习的脑电信号分类算法.pdf

基于深度学习与心电信号的抑郁症识别研究_毕业论文.pdf

基于特征融合与深度神经网络的心电图分类方法研究_毕业论文.pdf

基于遗传BP神经网络的胎儿心电信号提取算法研究_毕业论文.pdf

基于ICA分离的胎儿心电信号分离技术研究_毕业论文.pdf

基于多任务深度神经网络的心电分类方法研究_毕业论文.pdf

基于机器学习的心电信号分类研究_毕业论文.pdf

基于度量学习的心电身份识别方法研究_毕业论文.pdf

基于稀疏表示和字典学习的心电身份识别方法研究_毕业论文.pdf

基于多层复合电极的非接触式睡眠心电信号检测系统研究_毕业论文.pdf

基于多尺度特征融合变换神经网络的心电信号重构算法研究_毕业论文.pdf

基于特征学习和多特征融合的心电身份识别方法研究_毕业论文.pdf

相关实用应用程序（Windows可用）

免费可用的ChatGPT网页版.zip

ChatGPT使用总结：150个ChatGPT提示词模板（完整版）

chromedriver-win64.zip

最新资源