Modeling And Predictive Control Of Experimental Greenhouse Temperature Hybrid System

Greenhouse microclimate is a large time-delay nonlinear system coupled with temperature and humidity and disturbed by the outdoor environment,which can be influenced by control equipment,greenhouse structure and materials,as well as internal planting crops.Among the microclimate environmental variables,the greenhouse temperature plays the most vital role and is the main controlling factor of the greenhouse microclimate environmental system.The greenhouse temperature control equipment is normally on-off device or continuous adjustment device without position feedback unit.The state of device is generally discrete variable,while the output(indoor temperature)and the disturbance inputs(outdoor environmental factors)are continuous variables.So that the control algorithms for the continuous system are not suitable for this system,and the conventional on-off control is easy to overshoot under the large time-delay system,which may cause the equipment to start and stop frequently.In order to solve the above problems,hybrid system modeling and control methods are proposed by considering the system as a hybrid system.Based on the switching model theory,greenhouse temperature system can be divided into four subsystems:thermal insulation mode,natural ventilation mode,forced ventilation mode and wet curtain fan model,and the ARMAX model was established for four subsystems respectively.Firstly,the main correlation inputs of each subsystem model were determined,the model structure was determined by the statistical hypothesis test,and the recursive augmented least square identification model parameter with forgetting factor was adopted to identify model parameters,using experimental greenhouse data to verify the model accuracy.Based on the subsystem model,by introducing auxiliary variables,transforming the constraints of the subsystem switching and temperature control into linear inequalities,the MLD model of the greenhouse temperature system was established.It was demonstrated that the model has a certain control effect and the prediction accuracy was high by experiments.A hybrid automata was designed for greenhouse temperature hybrid system.The four subsystems of the greenhouse were abstracted to four discrete states.The switching logic of the hybrid automata was designed.The correctness of the state transitions under each discrete state was analyzed,and the deterministic and non-blocking performance of the automata was verified.Then we adopted dual-cycle temperature integration programming temperature set value,dynamically adjust the set value according to the sum of the accumulated temperature within the accumulated temperature period.The experiments verified the effectiveness of the hybrid automata control method,compared and analyzed the control effect before and after the setting of the set valueplanning,and found that the use of set value of the accumulated temperature planning can reduce the switching times of the equipment and decrease the energy consumption.Model predictive control of greenhouse temperature based on switching model was proposed,adding the set value following,the number of device switching times as well as the energy consumption to the predictive control performance index according to the demand of the greenhouse temperature.The controlled quantity is the switch sequence of each device,and its solving process is the optimization process of minimizing the performance index.Due to the unknown future time disturbance inputs in the performance index,we adopted grey prediction method for short-term prediction.Since the problem to be optimized is an NP-hard problem,the genetic algorithm was used to search the approximate optimal solution,and the optimal pruning exhaustion method was used to verify the search results of the genetic algorithm.In order to reduce equipment switching losses,the temperature integration theory was introduced to optimize the set value of model predictive control.The experiment verified the control performance of the predictive control algorithm,compared and analyzed the control effect before and after the set value planning.The experiment results show that the predictive control is suitable for the greenhouse temperature control system,and the control precision is higher than that of the hybrid automata control.