Optimization Of Central Air Conditioning Cooling System Based On Intelligent Algorithm

The central air conditioning system of large public buildings is long-term operated under partial load,and there are no effective control methods to adjust the operating parameters with the change of load.So that the equipment operation efficiency is low.There are many plants in centralized chilled water cooling system,and the operation parameters are interrelated.The traditional control methods,such as empirical control,PID control and fuzzy control,are difficult to achieve the overall high efficiency operation of the chilled water cooling system.It is an effective optimization method to establish simulation model to find the settings of operation parameters with the highest energy efficiency,and then adjust the optimization parameters through the slave computer.Considering the central air conditioning system is a typical nonlinear,coupling system,it is difficult to establish the mechanism model.With the development of big data technology and machine learning,data mining has been widely used in HVAC field.In this paper,the deep learning algorithm is introduced to build the energy consumption model of centralized chilled water cooling system,and heuristic algorithm is used to solve the optimization problem.Since the parallel operation mode of multiple chillers is widely used in large central air-conditioning system,this paper takes the central air-conditioning cooling system of a metro station in China as the research object,and proposed optimization strategy based on its parallel operation characteristics.Base on the measured data and hierarchical modeling method,the energy consumption prediction model of each component was established by residual neural network,and the chilled water branch flow model was established to connect the load distribution between chillers with chilled water flow and chilled water supply temperature.According to the actual operation characteristics,the equipment models are integrated into the complete cooling system simulation model.Compared with the traditional neural network model,the residual neural network model has higher accuracy,the average relative error is1.4613%,and the maximum prediction error is less than 6%.Based on the established energy consumption prediction model,the optimization model of operation parameters of centralized chilled water cooling source system is established under equipment safe operation,energy conservation and indoor comfort constraints.And the environmental parameters include real-time cooling load,outdoor temperature and humidity are taken as the given inputs,the minimum total energy consumption of the cooling system is taken as the optimization objective,the chilled water supply temperature,cooling water inlet temperature,chilled water flow and cooling water flow of each chiller are taken as control parameters,the gray wolf optimization algorithm is introduced to solve the optimization problem.Compared with genetic algorithm(GA)and particle swarm optimization(PSO),the gray wolf optimization algorithm has higher accuracy and better convergence,which is suitable for the optimization of operating parameters.Compared the optimization results of only one control parameter,two or more control parameters,and equal chilled water supply temperature of each chillers,average energy saving effect on 30 group operating points reach the highest level when all control parameters are optimized.The highest average energy saving rate of the system,the chillers,the pumps and the cooling towers is 10.39%,8.14%,23.79% and 15.17% respectively,while the energy saving rate of chiller decreases to 5.38% when the chilled water supply temperature of chiller is equal.