The Greenhouse Temperature System Modeling Based On Sparse FIR Model

Greenhouse microclimate system is a quite complex dynamic system with a characteristic of time delay, nonlinearity, strong interference, strong coupling and time variance. The environmental conditions determine the crops’growth status directly and have a great influence on production, quality, supply and economic benefits of greenhouse crops as well. So, it’s of great value to study the greenhouse system modeling and control to improve the level of facility agriculture.In this thesis, the Finite Impulse Response (FIR) model is used to describe greenhouse temperature system. Compared with ARMAX (Auto Regressive Moving Average Models with External Input) model, the structure identification of the FIR model does not need to determine the order and time delay. The FIR model is a linear model, but the greenhouse temperature system is a nonlinear and time-varying system. So, the impulse response sequence of the system should be identified based on as less data as possible. The FIR sequence of the system with time delay is sparse, according to the Compressed sensing theory, we can use under-sampled measurement data to reconstruct the FIR sequence of the system by solving sparse optimization problem (such as LASSO), and get the time delay property of the system. Besides, the influence of noise can be reduced by adjusting the sparsity of the signal.In order to reconstruct the FIR sequence of the system much better and obtain the time delay of the system much more accurate, this thesis begin the research from some aspects, like data selection, parameter adjustment and algorithms, and study different sparse optimization algorithm, include the Alternating Direction Method of Multipliers(ADMM), Inverse Scale Space(ISS). Finally, we get the time delay of outside temperature, outside solar radiation, cooling pad, which is10minutes,1minute and1minute, this results accord with the mechanism model of the greenhouse temperature system. The fitting of model achieve96.73%,94.14%under the situations of off and on of the Wet Curtain, so the experiment prove that the model have higher credibility.The greenhouse system is a system with a big time lag. The Predictive Control algorithm, which is based on model, is an effective method to solve the problems of the control of systems with time delay. After identifying the impulse response sequence of the system, we apply it directly to the Model Algorithm Control to predict the future temperature which determines the current state of the equipment, like the Wet Curtain. This method can reduce the switching frequency and wear of the equipment. Finally we verify the validity of the model and the control algorithm by simulation.In the final, the conclusion is given and the problems which need further study were presented.