Research And Prediction On The Control Of Greenhouse Microclimate Environment Prediction Based On CFD Model

With the advance of agricultural modernization,facility agriculture has become the basic indicator of the development of agricultural modernization.The modem greenhouse could provide the suitable growth environment for the crops,and realize the high efficiency and high quality production of the crops.As one of the main means to regulate the environment of crop growth,the environmental regulation of the greenhouse is one of the key technologies to realize the modem greenhouse.The object to establish a precise mathematical model is the main form of traditional control method,the establishment of a reasonable and accurate model of the greenhouse environment is the important prerequisite for the implementation of greenhouse environment control.Because of the complexity of the greenhouse environment,in the process of the establishment of the model often contains uncertain parameters,so it was necessary to identify the parameters of the model.Based on computational fluld dynamics modeling technology,although tihe greenhouse microclimate environment can be simulated,the CFD calculation is large,the calculation time is long,so it needs to be calculated in parallel.This paper took the plastic greenhouse in Liyang region of Jiangsu as the research object,establishment of CFD model for summer,autumn,winter greenhouse microclimate environment based on the computational fluid dynamics(CFD)modeling method,the parameter identification of CFD model was realized on the MATLAB and FLUENT collaborative platform,And four kinds of parallel structures are built,and the parallel eflficiency computing platform is compared.And based on this the control method of greenhouse microclimate environment prediction based on CFD model was explored.The main research contents are as follows:(1)On the basis of reasonable simplification of the greenhouse,collected data and established under natural ventilation condition in summer,natural ventilation condition in autumn and closed condition in winter the CFD model of the microclimate environment in the greenhouse.Calculated under natural ventilation condition in summer,natural ventilation condition in autumn and closed condition in winter the CFD model of greenhouse under natural ventilation condition in the FLUENT platform,which were single core,single machine multi-core,multi machine multi-core three operating conditions,with the experimental results the distribution of temperature and humidity in the greenhouse was in accordance with the actual,and in the single machine environment,the computation time was 300s;in the single machine multi-core and multi machine multi-core parallel environment,the computation time was 90s.Therefore,in parallel environment,the computation time was greatly reduced;the establishment of this model was ready for the subsequent model parameter identification and predictive control.(2)On the basis of MATLAB parallel environment and program design,implemented Parallel Particle Swarm Optimization(PPSO)algorithm on greenhouse CFD model parameter identification.Among them,MATLAB as the PPSO algorithm runned platform,FLUENT as a greenhouse CFD model computed platform,used MATLAB and I/O FLUENT file operation function,established a shared file mechanism based on the MATLAB and FLUENT collaboration platform,CFD model parameters identification and subsequent tests were completed on this platform.We set up a platform for computing the four structures and under the same identification accuracy comparied and calculated performance of parameter identification of CFD model in greenhouse environment.The four kinds of computing structures were structure 1 single core PSO+ single machine multi core FLUENT which calculation time consuming 9.6h,structure 2 single core PSO+ multi stage multi core FLUENT which calculation time consuming 30.1h,structure 3 single multi core PPSO+ multi core FLUENT which calculation time consuming 11h and structure 4 Multi machine multi core PPSO+ multi core FLUEN which calculation time consuming 5.7h.The structure 4 effectively improved the computational efficiency,and solved the problem that the CFD model of the greenhouse could identify the time consuming problem,so the structure 4 was the optimal computing structure.(3)Besides establishment of CFD prediction model of greenhouse environment with the optimal parameter identification,the forecast results and the measured results were compared with the CFD prediction model under natural ventilation condition in summer,natural ventilation condition in autumn and closed condition in winter.the prediction results of CFD model were compared with the measured results.The error of predicted value and measured value of greenhouse temperature and humidity is small.The predicted results of the greenhouse prediction model were in good agreement with the actual results under the optimal parameter values.So the identification of the optimal parameters was reasonable,the establishment of the greenhouse environment CFD prediction model was effective.(4)On the base of collaboration platform of MATLAB and FLUENT,predictive control model(MPC)method based on CFD model was explored and studied.Selection of CFD model in greenhouse under the condition of wet curtain fan ventilation,wet curtain and fan for control equipment selection of greenhouse,and the prediction time domain was set to 3,and the sampling time was set to 6.Compared with the model results in greenhouse CFD under the same conditions with the proportional integral derivative control(PID),MPC control algorithm under the greenhouse temperature calculation results and reference temperature absolute deviation was 1.27?,the relative humidity calculation results and reference relative humidity absolute deviation was 18%.PID control algorithm under the greenhouse temperature calculation results and reference temperature absolute deviation was 1.83?,the relative humidity calculation results and reference relative humidity absolute deviation was 18.89%.So the results of using MPC control algorithm were stable and better than PID control.Greenhouse temperature and humidity distribution was in line with the actual situation.