Collaborative optimization method for solving the diffusion and allocation issues in complex variable flow rate HVAC systems

The complexity of current optimization methods for HVAC systems is increasing, resulting in relatively lower computational efficiency, particularly in more complex systems. This difficulty makes real-time optimization and control challenging in practice. Therefore, there is an urgent need to simultaneously improve both system energy efficiency and computational efficiency to enhance system robustness. Present optimization methods predominantly emphasize enhancing system energy efficiency, often overlooking computational efficiency. Consequently, these methods become infeasible or unstable when implemented in practical systems. In our research, a multi-agent-based collaborative optimization method is proposed to solve the global optimization problem of complex HVAC systems. Under the multi-agent framework, the global optimization problem is decomposed into multiple sub-optimization problems considering the interaction characteristics among components, thus reducing the complexity of the global optimization problem in HVAC systems. The proposed AH-AFSA algorithm supports the solution of optimization problems containing hybrid decision variables (continuous and discrete variables) and can directly search for optimal discrete variables in the binary space. This feature is suitable for searching the optimal ON/OFF sequence and setpoints simultaneously during the global optimization process. The results demonstrate that the proposed method can save 18.9 % of electricity consumption with an average computing time of 12.2 s for each operating condition, saving about 54 % of the time cost compared to centralized methods. The methodology used in our research holds significant theoretical and practical value for enhancing the computational efficiency and productivity of optimization methods in complex HVAC systems.