Abstract: This paper systematically reviews the current research status and development trends in modeling and control methods for building heating, ventilation, and air conditioning (HVAC) systems, offering a comprehensive perspective on both theoretical advancements and practical applications. First, it conducts an in-depth analysis of the current state of HVAC control algorithms, covering both traditional separable control methods and coordinated control approaches, while evaluating 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 and its latest developments in various forms, which have significantly enhanced system performance and energy efficiency. Regarding stochastic model predictive control (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 introduction of probabilistic models has improved system robustness under load fluctuations and weather variations, particularly in large-scale commercial buildings. For distributed MPC (DMPC), it thoroughly examines the technical approaches to reducing computational burden through decomposition of large-scale optimization problems, specifically introducing advanced solution techniques such as sequential quadratic programming (SQP), Benders decomposition, and subgradient dual decomposition, which have demonstrated remarkable efficiency in multi-zone HVAC systems. In the field of data driven MPC, the paper systematically discusses two major paradigms combining machine learning (ML) with MPC—using ML to substitute MPC for control decisions and employing ML to construct prediction models—while validating through multiple case studies 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 RL-MPC fusion frameworks, model-based RL (MBRL) methods, MBRL-MPC, and Safe Deep RL advanced fusion frameworks, demonstrating significant breakthroughs in combining MPC’s explainability with RL’s adaptive capabilities, which have shown promising results in dynamic building environments. Through comprehensive analysis, this paper reveals 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, the paper proposes two primary directions for future research: Universal control frameworks with knowledge transfer and adaptive capabilities, leveraging meta-learning, transfer learning, and federated learning to enable cross-building knowledge generalization and rapid adaptation; and 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 provides theoretical guidance and technical references for energy-efficient operation of building HVAC systems, contributing significantly to building energy conservation and the enhancement of 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, especially within the context of smart 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 optimizing modern HVAC systems.