Research On The Key Technologies Of Energy-saving Control For Multi-span Greenhouse Based On Predictive State Of Environment

In recent years,China has introduced a large number of multi-span greenhouse from Netherlands.However,most greenhouses have been operating in deficit due to their high energy consumption.It can be seen that energy consumption has become a bottleneck problem restricting the development of greenhouse industry.Optimizing greenhouse environmental control system is one of effective ways to reduce energy consumption of greenhouses and has become a research hotspot in the field of energy conservation in greenhouses.At present,most greenhouse environmental control systems make energy-saving control decisions based on past and present environmental data of greenhouses.It is difficult to obtain good energy-saving effect because of the nonlinearity and hysteresis quality of greenhouses.Making energy-saving control decisions based on predictive state of environment is more conductive to realizing energy-saving of greenhouses,which considers the nonlinearity and hysteresis quality of greenhouses.Therefore,in this paper,research was conducted on the key technologies of energy-saving control for multi-span greenhouse based on predictive state of environment.The main contents of this paper are organized as follows:(1)An outdoor solar radiation forecasting model was built in this paper.The optimal combination of input parameters was selected from 32 features input through comparison experiment in this paper.The outdoor solar radiation forecasting model was trained by Deep Neural Networks.The forecasting input was built using weather forecasting service.In addition,the performance of the solar radiation forecasting model in Taian city,Shandong province(36.08~°N,116.95~°E)and Langfang city,Hebei province(39.23~°N,116.44~°E)was examined.In the verification test,different weather conditions(sunny days,cloudy days,etc.)were considered.During the test,the maximum value of the correlation coefficient(R~2)was0.98,the minimum value of Root Mean Square Error(RMSE)was 15.2W/m~2and the minimum value of Mean Absolute Error(MAE)was 7.82W/m~2.According to the comparison with solar radiation forecasting model at home and abroad,the solar radiation forecasting model based on Deep Neural Network and weather forecasting service has a higher precision.(2)A solar radiation model of multi-span greenhouse cover was proposed in this paper.In this paper,the method for calculating solar radiation on various surfaces,the method for calculating dynamic cover absorbance and transmittance and the method for calculating solar radiation absorbed by the cover and transmitted into the greenhouse were proposed.Dynamic cover absorbance and transmittance were put forward in order to improve the performance of solar radiation model of multi-span greenhouse cover.The absorbance of the cover was divided into beam-radiation absorbance,diffuse-radiation absorbance and ground-reflected radiation absorbance.Meanwhile,the transmittance of the cover was divided into beam-radiation transmittance,diffuse-radiation transmittance and ground-reflected transmittance.The solar radiation absorbed by the cover and transmitted into the greenhouse was calculated using the concept of dynamic absorbance and transmittance.In this paper,the solar radiation model of multi-span greenhouse cover was introduced into a cover temperature forecasting model of greenhouses.The solar radiation model of multi-span greenhouse cover was validated by verifying the performance of the cover temperature forecasting model of greenhouses with the performance that the maximum value of R~2was 0.98 and the minimum value of RMSE was 15.2 W/m~2.(3)A prediction model of greenhouse state was built in this paper.In this paper,the prediction model of greenhouse state was proposed based on the principle of energy balance considering five distinct forms of energy transfer:solar radiation absorption,heat convection,heat conduction,latent heat exchange and long-wave radiation.As the analytical solution to the equations is time-consuming and laborious,a computer program adopting the MATLAB standard function ode45 was written to discover a solution to the energy equations using the forth-order Runge-Kutta method.To verify the developed model,the verification test was carried out under two conditions:with crops in the greenhouse and without crops in the greenhouse.The prediction model of greenhouse state was verified by comparing the calculated value and measured value of the state variables.(4)The energy-saving control strategy was proposed based on the predictive state of greenhouses in this paper.In this paper,the prediction model of greenhouse state,theory of optimal control and feedback compensation were incorporated into the energy-saving control of greenhouses to build the energy-saving control strategy based on the predictive state of greenhouses.Finally,the energy-saving effect of the control strategy based on the predictive state of environment was verified by a check experiment.