Application of Dynamic Mode Decomposition for Temperature Analysis in Smart Building

The smart building having multiple subsystems are gaining momentum due to the growing trend of smart cities. These multiple subsystems interact with each other through a communication channel and coordinate through a building management system (BMS) for the effective operation of the smart building. The communication channels are prone to vulnerabilities (cyber attacks) which may lead to anomalies condition. However, with a proper prediction of future data beforehand various types of anomalies can be avoided. The task of predicting data requires extensive knowledge of the system model as well as the process. In view of this, the paper proposes a prediction technique known as the Dynamic Mode Decomposition (DMD) which can predict future temperature profile data with the help of available past data in an equation-free environment. The temperature data of the major component of BMS i.e. heating, ventilation, and air conditioning (HVAC) system is predicted using past temperature data with the help of DMD. After the prediction of the temperature profile, the concept of a process control chart is used for analyzing the HVAC system as normal or anomalies condition. The effectiveness of the proposed method for prediction of data using DMD where all system states may not be observable and the analysis of predicted data using the process control chart is verified using different test scenarios. Finally, from the result, it can be highlighted that DMD predicts the data effectively without the need for a system model, and the process control chart helps to identify the presence of anomalies.