Study On Optimal Control Strategy For HVAC System Based On Model Predictive Method

In recent years,the energy consumption optimization control of heating,ventilation and air conditioning system(HVAC)has been focused on in the field of building energy efficiency.Since the model of the HVAC system equipment has nonlinear,pure lag and time-varying characteristics,the satisfactory results cannot be obtained in actual experiment compared with simulation.In this dissertation,the optimal control of HVAC system energy consumption is studied from two aspects.On the one hand,simulation and field experiments are combined to improve the accuracy of the model;on the other hand,the model predictive control method is applied to tolerate a modest model error,and then to achieve optimization.The main research work of this dissertation is summarized as follows.(1)In this dissertation,simplified models are established for each main component of a HVAC system,such as: zone,chiller,cooling tower,pump,fan,coil.The model predictive control method(MPC)was applied to simulate the system control.The results show that MPC can save operating cost of HVAC system,and can determine the thermal mass through time constant,thus can learn the cost saving potential of passive energy storage.(2)Building load is affected by some random variables,such as outdoor temperature,solar radiation and so on.In this paper,the power spectral density method was used to identify the transfer functions of the indoor air conditioning load under two random variables,the outdoor temperature and the solar radiation intensity.And then the identified transfer functions were applied to a heater controlled by a MPC strategy,and the optimal control of the room temperature is achieved.(3)A method combining time series method and neural network is proposed to predict the building load for the ice storage air conditioning system of a bank building.Firstly,the time series model is determined by stationary test,model order determination,parameter estimation and model checking.Then the artificial neural network is used to model the residual error of the prediction result of the time series model,since the time series model cannot describe the nonlinear characteristic of the system.The experiment results show that the combination model can improve the accuracy of load forecasting for multi-step prediction,and is proved to be applied to ice storage air conditioning system.(4)A method that takes the opening level of chilled water bypass valve as the control target without changing the original control system is proposed for a commercial building in business,and MPC with feedforward structure is chosen as the control strategy.The method forces the original control system to raise the chilled water supply temperature by controlling the bypass valve opening level at a small value close to zero.The experiment results show that the method can effectively raise the supply chilled water temperature,and then save energy,on the premise of the indoor comfort.(5)Distributed model predictive control method(DMPC)is used to solve the problem of the coupling between variable air volume(VAV)air conditioning systems and air handling unit(AHU).A one step distributed model predictive control method is proposed in this paper by simplifying the heat transfer between the zones.By simplifying the constraints of the optimization problem,the calculation amount of the algorithm and the communication between the controllers are reduced significantly,which is more conducive to the application of DMPC algorithm in the application.The experiment results show that each zone can be controlled in the comfort range,and the worst case zone can always remain at the upper temperature limit of the comfort range,which means the minimization of energy consumption.(6)To solve the nonlinear problem of the equipment models in HVAC system,a controller model based on the artificial neural network(ANN)is constructed,which solves the problem that the ordinary model predictive control method cannot achieve the optimization.The experiment results show that the artificial intelligence control algorithm can effectively achieve the optimization,which means the energy saving can be obtained under the premise of meeting the indoor comfort and system process requirements.