People in modern societies spend most of their time in buildings, and a considerable amount of energy is used to maintain indoor air quality (IAQ) and thermal comfort levels to ensure people’s health and productivity. Thus, it is important to balance IAQ and thermal comfort levels with energy consumption. However, traditional ventilation systems have difficulty in effectively capturing the detailed distributions of the indoor environment, making it challenging to meet the dynamic control of the indoor environment in real-time. Computational fluid dynamics (CFD) is a numerical modeling technique that can characterize detailed indoor airflow distributions. And artificial intelligence (AI) algorithms are highly capable of controlling indoor environmental parameters. Therefore, this thesis...[ Read more ]

People in modern societies spend most of their time in buildings, and a considerable amount of energy is used to maintain indoor air quality (IAQ) and thermal comfort levels to ensure people’s health and productivity. Thus, it is important to balance IAQ and thermal comfort levels with energy consumption. However, traditional ventilation systems have difficulty in effectively capturing the detailed distributions of the indoor environment, making it challenging to meet the dynamic control of the indoor environment in real-time. Computational fluid dynamics (CFD) is a numerical modeling technique that can characterize detailed indoor airflow distributions. And artificial intelligence (AI) algorithms are highly capable of controlling indoor environmental parameters. Therefore, this thesis aims to propose a series of AI algorithms that can be deployed in indoor environmental control for rapidly and accurately predicting and optimizing IAQ, thermal comfort levels, and energy efficiency in building rooms via the combination of CFD simulations and experiments.

Given that carbon dioxide (CO₂) is the primary fluid waste emitted by occupants inside buildings, the CO₂ concentration is considered a gas-type indicator of the IAQ. The CFD is used to establish the database that stores data on indoor airflow and CO₂ distributions of different building structures and indoor conditions. Using the database as a basis, a back-propagation neural network (BPNN) combined with a particle swarm optimizer (PSO) algorithm is proposed to rapidly predict and optimize the IAQ, with 6.44% mean reductions of CO₂ concentration.

Whereas, there are other types of indoor air pollutants besides CO₂, such as particulate matter (PM_2.5 concentrations). Therefore, developing a multi-objective optimization algorithm to rapidly and accurately predict and control indoor CO₂ and PM_2.5 concentrations to improve IAQ plays an important role. With the indoor pollutants database created by CFD simulations, the BPNN-based adaptive multi-objective particle swarm optimizer (AMOPSO) algorithm is initialized to predict and optimize the concentrations of indoor air pollutants. In test examples, the proposed optimization algorithm reduces CO₂ concentrations by up to 30.5%, while also reducing PM_2.5 concentrations by as much as 77.1%.

In addition, balancing thermal comfort levels with energy consumption is also important for indoor environmental control. The database that stores indoor airflow and temperature distributions is created by CFD simulations. Based on the database, the BPNN-based adaptive grey wolf optimizer (GWO) algorithm is applied to predict and optimize thermal comfort levels and energy efficiency. The results show that the proposed AI algorithm can rapidly predict thermal comfort levels and have a strong optimization ability. Meanwhile, 1.01% of energy savings are achieved.

The CFD technique is effective in obtaining a detailed indoor environmental database. Nonetheless, validation of the proposed CFD models also needs to be examined. Chamber experiments are conducted while taking IAQ, thermal comfort, and energy savings into consideration. The CFD simulation results are then combined with the experimental data to verify the accuracy of the results obtained and create a sufficient database. With such a database, the BPNN-based adaptive multi-objective particle swarm optimizer-grey wolf optimization (AMOPSO-GWO) algorithm is used to identify the optimal strategy for maximizing IAQ, thermal comfort levels, and energy savings. The results demonstrate that the mean reduction in air pollutants concentrations, increase in thermal comfort levels, and average energy savings are 31%, 45%, and 35%, respectively.

This thesis systematically investigates a series of AI algorithms that combine CFD simulations and experimental data to predict and optimize IAQ, thermal comfort, and energy consumption. Compared to state-of-the-art indoor environmental control approaches, on the one hand, this thesis benefits from CFD simulations which can acquire detailed indoor environmental information and experimental results to establish a sufficient indoor database. On the other hand, this thesis proposes a series of AI algorithms that can accurately and rapidly predict and optimize indoor environmental parameters, providing useful recommendations for smart indoor environmental control.