返回
Lightweight Wavelet Convolutional U-Net for Seismic Phase Recognition
DOI:10.1109/ACCESS.2026.3662343.png)
摘要
En 中文
为调和相位拾取精度与边缘部署效率之间的冲突,本研究提出了一种轻量级小波卷积U-Net。其核心创新在于用可逆小波范式替代标准采样。我们引入了一种轻量级小波卷积注意力模块,将离散小波变换与轻量化卷积相结合,以在不损失信息的情况下捕捉多尺度结构线索。此外,设计了一种频率分支处理与融合模块,通过逆离散小波变换解耦频带并执行高保真重建,有效解决了传统上采样固有的频谱混叠问题。这种频率感知架构为实时地震监测建立了一种新方法,在嵌入式平台上实现了低延迟和高结构保真度之间的稳健平衡。
Keyword:
Accuracy
Real-time systems
Discrete wavelet transforms
Feature extraction
Convolution
Computer architecture
Computational modeling
Wavelet analysis
Time-frequency analysis
Decoding
Computational efficiency
convolutional neural networks
deep learning
feature extraction
microcontrollers
neural networks
real-time systems
seismic measurements
signal detection
wavelet transforms
期刊
IF:
3.6
论文数:
9.8W
被引数:
29.4W
机构
引用论文
Multivariate singular spectrum analysis of seismicity in the space–time-depth-magnitude domain: insight from eastern Nepal and the southern Tibetan Himalaya空间-时间-深度-震级域中地震活动性的多变量奇异谱分析:来自东尼泊尔和南藏喜马拉雅的见解
IPIML: A Deep-Scan Earthquake Detection and Location Workflow Integrating Pair-Input Deep Learning Model and Migration Location MethodIPIML: 集成对输入深度学习模型和迁移定位方法的深度扫描地震检测和定位工作流

