monster-monash/LenDB

LenDB数据集包含来自全球多个不同地震检测网络的地震图记录，由1,244,942个多元时间序列组成，每个时间序列包含3个通道，每个通道长度为540，采样频率为20 Hz。数据集分为两个类别：地震和噪声。该数据集已被划分为交叉验证折，确保来自同一地震检测网络的地震图不会在训练和验证集中重复出现。

The LenDB dataset comprises seismograms recorded from various seismic detection networks around the world, consisting of 1,244,942 multivariate time series, each with 3 channels and a length of 540, sampled at 20 Hz. The dataset is categorized into two classes: earthquake and noise. This dataset has been split into cross-validation folds to ensure that seismograms from the same detection network do not appear in both the training and validation sets.