Modeling And Control Of Temperature System In A Greenhouse By Deep Flow Technique Of Nutrient Solution

The main tasks of greenhouse climate control are to make full use of natural resources and to control microclimate including temperature, humidity, solar radiation, concentration of CO₂, water and nutrient component of crop rhizosphere to create the optimum environment for the crops. It is one of the important tools for manipulating crop growth, increasing the yield and improving quality of greenhouse crop, regulating production cycle and improving economic benefit. Greenhouse microclimate has the characteristics of strong nonlinearity, large time delay, severe coupling, strong external noise disturbances, time-varying etc., so it's a very complex dynamics system. Greenhouse climate control has been dealt with traditionally by several disciplines: crop ecophysiology, horticulture, greenhouse physics, control engineering, instrument and computer science.Greenhouse microclimate modes are essential for improving environment management and control efficiencies. Physical models can provide good predictions over short future time horizons, but they have the drawbacks of high cost and too many parameters. Non-linear models which are built by observing input and output data have good performances in simulation and prediction of greenhouse microclimate, but they require long computation time, which restricts their application to real-time implementation. At present, model structure, cultivation pattern, weather outside greenhouse, interference of the greenhouse growers and the like are not considered in linear models built by observing input and output data, which leads to low precision of models. Therefore, it is of theoretical and applicable importance for modeling of greenhouse microclimate.Most of greenhouse climate control equipments in China are on-off control. There is no displacement feedback. So it is hard to use modeling and control theroy based on continuous control devices. The drawback of the conventional ON/OFF control algorithm is frequent switch actions of equipments, which leads to equipment spoilage and energy waste. To handle with the above problems, this dissertation mainly worked on modeling and control of air temperature system by deep flow technique nutrient solution based on hybrid system.In this dissertation, the achievements are:(1) The mechanism of greenhouse microclimate was analyzed utilizing laws of physics such as heat transfer, mass transfer, and the principles of crops physiology such as transpiration. Then modeling of the greenhouse inside temperature system by deep flow technique of nutrient solution was implemented according to energy-balance method. After that the mechanism models of the inside air temperature system were analyzed. Furthermore some prior information was obtained which was helpful to the subsequent experiment modeling research.(2) Exploring whether on-line model could perform well and predict the temperature in an unheated, naturally ventilated greenhouse. Linear auto-regression moving average with extra input (ARMAX) model was used, and statistic hypothesis test, mechanism analysis and model fits analysis were used together to select the model structure. An intelligent supervisory segment was devised to monitor and fix problems appearing during the on-line modeling process. Moreover, the residuals passed white noise. Experiments were carried out in different seasons. It is concluded from various simulations that auto-regression with extra input (ARX) models produce better results than physical models and off-line models. Computation times for ARX models are much lower, so it's suitable for real-time implementation.(3) Environment management and control of muskmelon and tomato cultivated by deep flow technique were studied. Different growth periods of the crop have different demands on the growing environment, comprising temperature, humidity, radiation and nutrient solution. According to different requirements, the greenhouse microclimate and roots environment of the crops (muskmelon and tomato) were controlled by the greenhouse measurement and control system developed by the authors. The results indicate that the circulatory system of nutrient solution and the measurement control system for greenhouse based on CAN bus is designed reasonably.(4) Model predictive control of greenhouse temperature based on mixed logical dynamical systems was proposed. The auxiliary variables, the on-off statuses of the window and the constrained operations for temperature control were combined into the discrete-time state space model of temperature, thus the mixed logical dynamical system of greenhouse temperature was constructed. Mechanism models and identified models of the inside air temperature system were validated, respectively. Mixed logical dynamical systems with model predictive controllers were applied for window temperature system in a greenhouse. Quadratic index predictive control strategy was designed based on management experience. Simulations and experiments on the temperature control system have been made to validate the efficiency of the algorithms.At the end of this thesis, all research work was summarized, and future research upon modeling and control of the greenhouse microclimate system was prospected.