A Hybrid Model Of A Direct Expansion Air Conditioning System For The Simultaneous Control Of Indoor Air Temperature And Humiditv

With improvement of the living standards,the living environment has become increasingly demanding comfort.Present household air conditioning usually solely control indoor air temperature,while at low latitudes where have high humidity in summer,solely control of temperature will lead to uncomfortableness of occupant.It's necessary to develop household air conditioning which can control temperature and humidity simultaneously.With the improvement of variable frequency technology,researchers proposed to simultaneously control indoor air temperature and relative humidity by varying speeds of compressors and supply fans in theory.While the heat and mass transfer in air conditioning system are highly coupled,the coupling of physical process leads to coupling of control loops.So the development of temperature and humidity simultaneously control algorithm is quite a challenging work.In this work,the model of direct expansion air conditioning system for predicting cooling dehumidification ability should be built first,the main content of this thesis is the modeling process for simultaneously control of temperature and humidity of direct expansion air conditioning system.In this thesis,components physics-based model of direct expansion air conditioning system are built first.Evaporator model based on enthalpy potential method is built,the average predicting errors of ?s and ?l(Sensible cooling capacity and latent cooling capacity)are 4.4%and 0.8%respectively.During the modeling,the fin efficiency equation based on enthalpy potential method is simplified to two simpler empirical equations.However,after the combination of components,the number of iterations is greatly increased.Physics-based model of direct expansion air conditioning system is complicated,and response time of the model is long,so empirical modeling is tried next.Artificial neural network(ANN)modeling is the most popular empirical modeling method.Researchers used ANN to predict ?s and ?l of direct expansion air conditioning system under an indoor air temperature and humidity successfully.However,one has to consider that empirical models,do not allow to accurately extrapolate beyond the range of the data used for estimating the model's parameters.The predicting results of the ANN-based model developed are validated when the operating point of the direct expansion air conditioning system drift from the training point.The average predicting errors of ?s and ?l are 1.3%and 18.3%respectively,the predicting accuracy for output latent cooling capacity was unacceptable.Therefore,the empirical-based modeling approach shouldn't be solely applied for simultaneously control of indoor air temperature and humidity.For the defects of physics-based model and ANN-based model,the thesis try to combine two modeling method,which use physics-based model to decouple the physical process and ANN-based model for quickly and accurately predicting.For the heat and mass transfer coupling in evaporator,physics-based modeling approach is applied for the evaporator while ANN-based modeling approach is applied for the refrigeration system except evaporator.Several variables are chosen as inputs and outputs to combine two sub models,and the errors of sub models are analyzed.Finally,the errors of hybrid model when the operating point drift from the training point is analyzed,the average predicting errors of ?s and ?l are 5.8%and 2.8%respectively,which are better than ANN-based model under same condition,and the high accuracy of predicting results showed that the hybrid modeling approach of developing ANN-based sub-model for components of direct expansion air conditioning system except evaporator and developing physics-based sub-model for evaporator is effective.It is worth mentioned that the hybrid modeling approach is not only suitable to be applied to develop control oriented model for simultaneously control of indoor air temperature and humidity using a DX A/C system,but also could be used for simulating a system contained coupling phenomenon,i.e.,applying the physical-based sub-model for the component having coupling phenomenon and the empirical sub-model for all the other components.