Energy-aware Vehicle HVAC Control Strategies for Regulating In-cabin PMV and CO2 Levels

Addressing the complexities of energy-efficient climate control within vehicle cabins, this thesis investigates innovative strategies for managing the heating, ventilation, and air-conditioning (HVAC) systems to optimise occupants’ thermal comfort and indoor air quality (IAQ), while minimising energy consumption. The research pivots around the Predicted Mean Vote (PMV) index and CO2 concentration as primary indicators of thermal comfort and IAQ, respectively. The first contribution of this study is a comprehensive literature review focusing on common parameters in HVAC system design. It highlights PMV as a critical metric for assessing passenger thermal comfort and adopts CO2 levels to evaluate IAQ. Building upon this foundation, the thesis proposes a dual-focused control strategy to regulate both PMV and CO2 levels without significantly burdening the vehicle's energy resources. Developing a fuzzy logic control (FLC) system is the second key contribution. This system is engineered to balance the often-conflicting criteria of maintaining PMV, stabilising CO2 concentration levels, and conserving energy. Existing vehicle HVAC FLC controllers primarily aimed to regulate thermal comfort and IAQ based on temperature and humidity. In other closed compartments like buildings and working environments, FLCs typically focused on either thermal comfort or IAQ. Only a few controllers managed to regulate both thermal comfort and IAQ while also considering energy consumption, often facing challenges in balancing all parameters simultaneously. In this research, FLC is designed to incorporate both PMV index and CO2 concentration into its control parameter with minimizing the energy consumption for vehicle HVAC systems. The advantage of the FLC lies in its low computational complexity, enabling efficient operations without requiring significant computational resources. This efficiency is crucial for vehicle HVAC systems, which need to respond rapidly to changes in passenger load and environmental conditions. MATLAB simulation results demonstrate the FLC's superiority over conventional threshold-based control methods, especially in adapting to manage CO2 levels based on passenger count. However, its design and calibration processes are acknowledged as time-consuming and complex. To address these challenges, the third significant aspect of this thesis is creating a reinforcement learning (RL) based HVAC control strategy. Utilising deep deterministic policy gradient (DDPG) algorithms, this approach dynamically adapts to changing conditions. Comparative analysis under various simulated scenarios reveals the RL strategy's enhanced performance over the FLC method, particularly in its adaptability and efficiency. Collectively, these strategies present a forward-thinking approach to vehicle climate control, promising to regulate PMV and CO2 levels effectively without incurring high energy costs. This advancement can extend the mileage of internal combustion engine (ICE) and electric vehicles (EV), marking a significant step in sustainable automotive technology.