PyAWD
收藏资源简介:
PyAWD是由比利时布鲁塞尔自由大学机器学习小组开发的合成地震数据集,旨在解决地震数据稀疏和分布不均的问题。该数据集通过模拟二维和三维异质介质中的声波传播,生成高分辨率的数据,支持机器学习模型的训练。数据集包含4608条模拟数据,涵盖了从地震波传播到特定特征(如震中或波时)的详细信息。PyAWD的创建过程允许用户精细控制参数,如波速、外部力、空间和时间离散化等,以生成复杂的地震波行为数据。该数据集主要应用于地震分析领域,旨在通过提供高质量的合成数据,增强机器学习在地震预测和震中检测中的应用。
PyAWD is a synthetic seismic dataset developed by the Machine Learning Group of Vrije Universiteit Brussel, which is designed to address the issues of sparse and unevenly distributed seismic data. This dataset generates high-resolution data by simulating acoustic wave propagation in two-dimensional (2D) and three-dimensional (3D) heterogeneous media, supporting the training of machine learning models. It contains 4608 simulated samples, with detailed information ranging from seismic wave propagation processes to specific features such as epicenter and wave travel time. The creation process of PyAWD allows users to finely control parameters including wave velocity, external forces, spatial and temporal discretization, and other relevant factors, to generate data exhibiting complex seismic wave behaviors. This dataset is primarily applied in the field of seismic analysis, aiming to enhance the application of machine learning in earthquake prediction and epicenter detection by providing high-quality synthetic seismic data.
PyAWD: Python Acoustic Wave Propagation Dataset
概述
PyAWD是一个用于生成包含自定义速度场中声波传播模拟的PyTorch数据集的包。
声波方程
声波传播方程为:$frac{d^2u}{dt^2} = c abla^2 u + f$,其中:
- $u(x, y, t)$ 是位移场,可以是标量或矢量场
- $c(x, y, t)$ 是波传播速度
- $ abla^2$ 是拉普拉斯算子
- $f(x, y, t)$ 是施加在系统上的外部力,其值随时间变化
PyAWD使用Devito Python库来解决从各种随机初始条件下的声波偏微分方程。
安装
通过PyPI安装: bash pip install pyawd
文档
API文档可在这里获取。
示例
提供了多个Notebook示例,位于examples目录中:
ScalarAcousticWavePropagation.ipynbVectorAcousticWavePropagation.ipynbVectorAcousticWaveDataset.ipynbMarmousi.ipynbInterrogators.ipynbGenerateVectorAcousticWaveDataset.ipynbSpatioTemporalVaryingWavePropagationSpeedField.ipynb
快速开始
基本导入: python from pyawd import *
生成包含10个模拟的数据集: python dataset = VectorAcousticWaveDataset2D(2, nx=128, dt=2, t=10, velocity_model="Marmousi") dataset.max_velocities[0] = 500
绘制第一个模拟: python dataset.plot_item(0)
绘制探测器响应: python dataset.plot_interrogators_response(0)
高级用法
使用VectorAcousticWaveDataset3D类生成3D模拟:
python
dataset_3d = VectorAcousticWaveDataset3D(1, nx=32, t=10, interrogators=[(0, 10, 15), (0, -10, 15)], velocity_model=300.)
生成视频: python dataset_3d.generate_video(0, "VAWD3D", 300)

- 1PyAWD: A Library for Generating Large Synthetic Datasets of Acoustic Wave Propagation with Devito比利时布鲁塞尔自由大学机器学习小组 · 2024年



