Font Size: a A A

Microclimate Simulation Of Double-film Covered Plastic Greenhouse In South China

Posted on:2014-03-29Degree:MasterType:Thesis
Country:ChinaCandidate:X Z ZouFull Text:PDF
GTID:2250330401970344Subject:Applied Meteorology
Abstract/Summary:PDF Full Text Request
The greenhouse is a closed agro-ecosystem, the internal environmental factors of which including solar radiation, air temperature, moisture are important to crop growth, yield and quality. For South China, greenhouse covered with double films is usually used for keeping suitable growth environment, especially temperature, to protect the cultivation of crops in cold winter. Therefore, it is necessary to do research on simulation of interior microclimate in double-film plastic greenhouse.In this paper, the characteristics of interior microclimate in double-film covered greenhouse fou South China in winter were analyzed. Taking the theory of thermo-dynamics and heat-transfer into full account, the physical for different weather conditions based on the mass and energy balance between indoor and outdoor environments was built to simulate the environmental elements of double-film greenhouse. In this model, the greenhouse was divided into outside covering layer, interlayer air, inside covering layer, indoor air, plant canopy and soil layer. The physical model was performed via Eular method and the Aitken Algorithm was applied to ensure the models’convergence rate and precision. Meanwhile, the parameters of convective heat transfer were estimated by optimization theory. Futhermore, the BP neural network model using the same data with physical model was also established to predict the indoor air temperature, relative humidity and soil surface temperature, then comparing the accuracy of this two models. The main research conclusions are as follows:1. Air temperature, relative humidity, and soil surface temperature of double-film covered greenhouse performed almost the same diurnal variation trend with those of single-film greenhouse under three weather conditions, but the air temperature and soil surface temperature of double films greenhouse were higher than those of single-film greenhouse and outside. Moreover, daily mean air temperature outside was lower than that inside besides on the clear day and cloudy day. This indicated that greenhouse covered by double films could effectively reduce heat dissipation, raise ground temperature and relative humidity, which would provide adaptive growth environment for crops.2. The root mean squared error(RMSE) and relative error(RE) between simulated values of indoor air temperature were below1.0℃and6%respectively under three weather conditions. When it came to relative, the RMSE were2.18%(clear days),1.70%(cloudy days),0.98%(overcast days). The model exhibited high accuracy for soil surface temperature, with the RMSE of less than0.1℃under three weather conditions. The highest accuracy of simulated air temperature was found for clear days, followed by overcast days, while the simulated relative humidity and soil temperature performed best on overcast days.3. Based on BP neural network model, the RMSE between predicted values and measured values of indoor air temperature, relative humidity and soil surface temperature were0.76℃,1.95%and0.49℃in sunny days, and the three values were6.44%,2.19%and2.55%for RE. Furthermore, the RMSE and RE of air temperature were1.02℃and4.43%on cloudy days, and were6.15%and6.73%for relative humidity. The RMSE and RE of soil surface temperature were0.85℃and4.04%. While it came to overcast days, RMSE and1.28℃,0.41%and1.02℃for those three factors, and the RE were15.69%,0.32%and6.64%, respectively.4. Compared to BP neural network model, the physical model performed better in simulating air temperature, relative humidity and soil surface temperature of the greenhouse, whose error always kept in a low level stably. However, the error of BP neural network model fluctuated with poor stability and unsatisfied accuary.
Keywords/Search Tags:double-film, plastic greenhouse, microclimate, physical model, BP neural network, simulation
PDF Full Text Request
Related items