Labeled Datasets for Air Handling Units Operating in Faulted and Fault-free States

This folder documents eight new air handling unit (AHU) fault datasets, covering various building configurations, control strategies, seasons, and injected faults. Additionally, a comparative study of publicly-available datasets is conducted to identify suitable dataset pairs and their corresponding faults, providing researchers with resources for evaluating their transfer learning (TL) algorithms. Notably, the results of this section can also be used for other inter-class studies, such as cross-building fault-symptoms comparisons, but TL remains the primary use case of this work.

本文件夹收录了8套全新的空气处理机组（Air Handling Unit, AHU）故障数据集，涵盖多种建筑配置、控制策略、运行季节以及人工注入故障场景。此外，本研究针对公开可用数据集开展对比研究，以筛选适配的数据集组合及其对应故障类型，为研究人员提供用于评估迁移学习（Transfer Learning, TL）算法的实验资源。值得注意的是，本章节的研究结果还可应用于其他跨类别研究，例如跨建筑故障症状对比分析，但迁移学习仍是本研究的核心应用场景。