HVAC system - Attack-detection

The dataset can be used to study the cybersecurity aspect of the HVAC system by evaluating the different attack detection and mitigation strategies. The dataset was collected from a simulation model of a 3-floor, 12-zone HVAC system for cooling using the Transient System Simulation Tool (TRNSYS), which is a graphical software environment for simulating a dynamical system. It consists of three logs: Dataset log 1 contains normal operational data collected for four months, Dataset log 2 represents normal operational data collected for 20 days. Dataset log 3 consists of the normal and attack data of 16 different attacks. It consists of 65 features: the hour of the year, the hour of the day, the temperature sensor measurements, the control signals, the control system's setpoints, the zones' thermal comfort indices, and the total estimated power usage of the HVAC system. Four files are provided as supplementary materials for training machine learning-based detection models using the Isolation Forest algorithm [1]. The details of the supplementary codes are as follows: File "HVAC - IF Training.ipynb" is for developing an attack detection model using Isolation Forest on the raw data, File "HVAC - PCA-IF Training.ipynb" is for developing an attack detection model using Isolation Forest on the data features extracted using Principal Component Analysis (PCA), File "HVAC - 1D CNN Training.ipynb" is for developing a feature extraction model using 1D Convolutional Neural Network (1D CNN), and File "HVAC - 1D CNN-IF Training.ipynb" is for developing an attack detection model using Isolation Forest on the data features extracted using the 1D CNN model.

For more information about the dataset refer to the following publications:

[1] Elnour, M., Meskin, N., Khan, K., & Jain, R. (2021). Application of data-driven attack detection framework for secure operation in smart buildings. Sustainable Cities and Society, 69, 102816. https://doi.org/10.1016/j.scs.2021.102816

[2] Elnour, M., Meskin, N., Khan, K., & Jain, R. (2021). HVAC System Attack Detection Dataset. Data in Brief, 107166. https://doi.org/10.1016/j.dib.2021.107166

* This dataset was supported by the Qatar National Research Fund (a member of the Qatar Foundation) under NPRP Grants number 10-0206-170360 and the Open Access funding was provided by the Qatar National Library

本数据集可用于研究暖通空调（HVAC）系统的网络安全维度，通过评估不同的攻击检测与缓解策略开展相关研究。本数据集采集自一套三层12分区冷却用暖通空调系统的仿真模型，该模型采用瞬态系统仿真工具（Transient System Simulation Tool, TRNSYS）搭建——TRNSYS是一款用于动态系统仿真的可视化软件环境。

数据集包含三类日志：日志1为四个月的正常运行数据，日志2为20天的正常运行数据，日志3则包含16种不同攻击的正常与攻击场景数据。该数据集共涵盖65项特征，具体包括年度小时数、当日小时数、温度传感器测量值、控制信号、控制系统设定点、各分区热舒适指数，以及暖通空调系统的总估算功耗。

本次提供四份辅助文件，用于基于隔离森林（Isolation Forest）算法训练机器学习攻击检测模型[1]。辅助代码详情如下：文件"HVAC - IF Training.ipynb"用于基于原始数据，通过隔离森林算法构建攻击检测模型；文件"HVAC - PCA-IF Training.ipynb"用于基于主成分分析（Principal Component Analysis, PCA）提取的数据特征，通过隔离森林算法构建攻击检测模型；文件"HVAC - 1D CNN Training.ipynb"用于基于一维卷积神经网络（1D Convolutional Neural Network, 1D CNN）构建特征提取模型；文件"HVAC - 1D CNN-IF Training.ipynb"用于基于1D CNN模型提取的数据特征，通过隔离森林算法构建攻击检测模型。

如需了解本数据集的更多详情，请参考以下出版物：
[1] Elnour, M., Meskin, N., Khan, K., & Jain, R. (2021). 面向智能建筑安全运行的数据驱动攻击检测框架应用. 可持续城市与社会, 69, 102816. https://doi.org/10.1016/j.scs.2021.102816
[2] Elnour, M., Meskin, N., Khan, K., & Jain, R. (2021). 暖通空调系统攻击检测数据集. 数据简报, 107166. https://doi.org/10.1016/j.dib.2021.107166

本数据集受卡塔尔国家研究基金（卡塔尔基金会成员）资助，资助项目编号为NPRP Grants 10-0206-170360，开放获取资金由卡塔尔国家图书馆提供。