DBbun/Extreme_Environment_Generator_v1.0

Extreme_Environment_Generator 是由 DBbun 创建的一个大型合成数据集，通过一个可配置的 Python 模拟器生成。它从以下极端物理环境中生成多传感器时间序列数据：地球深部地下、海底环境以及深冰世界和冰行星结构。每个生成的场景包括环境随时间的真实物理状态、被腐蚀/噪声的传感器测量值、层和材料剖面、环境配置、罕见故障事件以及场景级别的摘要统计信息。该数据集旨在用于机器学习研究，包括多变量预测、传感器融合、异常和罕见事件检测、在噪声、漂移和缺失数据下的鲁棒性测试、在极端合成环境中的强化学习以及时间序列的基础模型预训练。

Extreme_Environment_Generator is a large synthetic dataset produced by a configurable Python simulator created by DBbun. It generates multi-sensor time-series data from extreme physical environments inspired by deep-Earth subsurface, subsea environments beneath oceans, and deep ice-world and icy-planet structures. Each generated scenario includes the true physical state of the environment over time, corrupted/noisy sensor measurements, layer and material profiles, environmental configuration, rare failure events, and scenario-level summary statistics. The dataset is designed for machine learning research, including multivariate forecasting, sensor fusion, anomaly and rare-event detection, robustness testing under noise, drift, and missing data, reinforcement learning in extreme synthetic environments, and foundation model pretraining for time series.