Research On Multi-model Fusion Optimal Energy-saving Control Of Central Air-conditioning Cold Station System

The central air-conditioning system is an important infrastructure in urban public buildings as well as the main energy-consuming equipment.For achieving optimized energy-saving control of the central air-conditioning system,it is necessary to concentrate on the research of the central air-conditioning cold station system.The point of energy-saving optimization of the central air-conditioning cold station system includes establishing an accurate static energy consumption model.Regarding the problems in the modeling of the central air-conditioning cold station system,because the refrigeration unit accounts for the largest energy consumption and the strongest nonlinearity in the central air-conditioning cold station system,this paper proposes a multi-model weighted fusion model based on the division of working conditions.According to the nonlinear characteristics of the working curve of the refrigeration unit,the input/output data is clustered into several sub-categories,and each sub-category adopts the Marquardt method to establish a sub-model.Thinking of the slow time-varying characteristics of the central air conditioning system and the characteristics of the operating conditions of the refrigeration unit,there is information redundancy between the sub-models.Therefore,the weighted method is used to fuse the sub-models to finally form a global model to eliminate its redundant information.The weighting coefficient takes the minimum mean square error as the index and is optimized by the particle swarm algorithm.The simulation results verify that the accuracy of the multi-model fusion model of the refrigeration unit established in the article is higher than that of the single global model.The multi-model fusion model of the refrigeration unit is applied to the energy-saving optimization of central air-conditioning cold station system,which will be more in line with actual working conditions.Accurate load forecasting is the basis of optimized energy-saving control for central air conditioning cold station system and is also an important link in building energy-saving technologies.In this paper,we use a BP neural network to predict the cooling load of a central air conditioner.Predict the minute-by-minute cooling load of the central air conditioning system for the next day based on historical sample data.the historical cooling load data and meteorological factors of the sample data are selected as training samples to obtain accurate forecasting values of the cooling load of the central air-conditioning system.In the light of obtaining cooling load of the central air-conditioning system and the energy consumption model of the central air-conditioning cold station system,by controlling the optimal operating point of the central air-conditioning cold station system running under the current load demand,the optimal energy-saving goal is realized.For obtaining the optimal operating point of the central air-conditioning cold station system under different cooling loads,use the central airconditioning cold station system’s energy consumption as the goal,and the static energy consumption model of the central air-conditioning cold station system as the objective function.established the constraints of the objective function,At the same time using particle swarm optimization algorithm,ideal working point of central air-conditioning cold station system under different cooling load requirements is calculated and optimized energy-saving control is performed.Since the central air-conditioning cold station is a system with strong disturbances,Therefore,when the external influence factors change,the system need to estimate and eliminate the disturbance.In this paper,it uses an automatic disturbance elimination controller to achieve optimized energy-saving control of the central air conditioning cold station system.And compare with the control performance of PID controller to get the best control method,the active disturbance rejection controller can effectively eliminate the overshoot in the control process of the central air-conditioning cold station system through the differential tracker.In the same time,this paper uses the expanded state observer to predict and remove the external and internal disturbances of the central air-conditioning cold station system,has strong robustness.In this paper,simulation is carried out by building an active disturbance rejection controller model in Simulink Model.The best working point for the central air-conditioning cold station system to work under different cooling load requirements,the purpose is to realize the optimized energy-saving control of the central air-conditioning cold station system.