Abstract: This paper provides a systematic review of the current research status and development trends in modeling and control methods for heating, ventilation, and air conditioning (HVAC) systems, offering a comprehensive perspective on both theoretical advancements and practical applications. First, it presents an in-depth analysis of the current state of HVAC control algorithms, covering both traditional separable control methods and coordinated control approaches, and evaluates their applicability across different building types and operational conditions through comparative case studies. The paper places particular emphasis on innovative applications of model predictive control (MPC) in HVAC systems, as well as recent developments of MPC in various forms, which have significantly improved system performance and energy efficiency. With respect to stochastic MPC (SMPC), it not only summarizes recent research achievements in building energy management—including hierarchical SMPC, Markov chain-based SMPC, and chance-constrained SMPC—but also highlights how the incorporation of probabilistic models has improved system robustness under load fluctuations and weather variations, particularly in large-scale commercial buildings. Distributed MPC (DMPC) provides a thorough examination of technical approaches for reducing computational burden by decomposing large-scale optimization problems. Specifically, it introduces advanced solution techniques such as sequential quadratic programming, Benders decomposition, and subgradient dual decomposition, which have demonstrated substantial efficiency in multi-zone HVAC systems. In the field of data-driven MPC (DDMPC), the paper systematically discusses two major paradigms that combine machine learning (ML) with MPC—using ML to substitute MPC for control decisions and employing ML to construct prediction models—and, through multiple case studies, validates the advantages of these methods in maintaining control performance while reducing MPC implementation barriers, especially in retrofit projects with limited historical data. Additionally, it provides an in-depth analysis of integration methods between MPC and reinforcement learning (RL), including basic MPC-RL fusion frameworks, model-based RL (MBRL) methods, MBRL-MPC, and safe deep RL advanced fusion frameworks, demonstrating significant progress in combining the explainability of MPC with the adaptive capabilities of RL, which has shown promising results in dynamic building environments. Through comprehensive analysis, this paper identifies the major technical bottlenecks currently faced by control methods for building HVAC systems: the challenges of transferability and real-time performance in MPC methods (including high-dimensional optimization, cross-building portability, and model dependency), as well as the multi-time-scale control challenges of MPC in HVAC systems (such as coordination among multiple subsystems, hybrid model architectures for multi-scale system coupling, and computational complexity). Based on these findings, this paper proposes two primary directions for future research. The first is universal control frameworks with knowledge transfer and adaptive capabilities, which leverage meta-learning, transfer learning, and federated learning to enable cross-building knowledge generalization and rapid adaptation. The second is hybrid intelligent control systems with multi-time-scale coordination, which integrate MPC and RL in a hierarchical structure to manage control tasks ranging from instantaneous device responses to seasonal energy planning. The study offers theoretical guidance and technical references for the energy-efficient operation of building HVAC systems, making a significant contribution to building energy conservation and enhancing indoor environmental quality. It also outlines future research pathways for the application of artificial intelligence, the Internet of Things, and edge computing technologies in HVAC systems, particularly in the context of innovative city development and sustainable building practices. Finally, the paper emphasizes the necessity of interdisciplinary collaboration among building science, control theory, and computer science to address the complex challenges involved in the optimization of modern HVAC systems.