【導(dǎo)讀】農(nóng)業(yè)大棚溫室智能化自動控制。摘要確定控制溫室作物生長歷來使用約束優(yōu)化或應(yīng)用人工智能技術(shù)解決了。軌跡的問題已被用作經(jīng)濟利潤的最優(yōu)化研究的主要標準以獲得足夠的作物生長。的氣候控制設(shè)定值本文針對溫室作物生長的問題通過分層控制體系結(jié)構(gòu)由一個。高層次的多目標優(yōu)化方法在解決這個問題的辦法是找到白天和夜間溫度參考軌。跡氣候相關(guān)的設(shè)定值和電導(dǎo)率fertirrigation相關(guān)設(shè)定值的目標是利潤最大化。果實品質(zhì)水分利用效率這些目前正在培育的國際規(guī)則結(jié)果說明選擇從那些獲得。工業(yè)的溫室在過去的八年中示出和描述。關(guān)鍵詞農(nóng)業(yè)分層系統(tǒng)過程控制優(yōu)化方法產(chǎn)量優(yōu)化。是最重要的主要受周圍環(huán)境的氣候變量光合有效輻射-PAR溫度濕度二氧化碳。聯(lián)營公司的成本主要是燃料電力和化肥減少減少殘留物主要是殺蟲劑和離子在。一個MO優(yōu)化問題可以定義為尋找決策變量的向量它滿足約束條件和優(yōu)。一旦前收割工作已經(jīng)產(chǎn)生這兩種替代品的有效期為多收獲收入取決于番茄果實

　　

【正文】 nal setpoints are defined and steady state models of greenhouse climate and tomato crop summarized in EqsAlthough several techniques have been evaluated to solve the MO optimization problem Liu et al 2020 in this case a goal attainment algorithm has been used sequential quadratic programing SQPbased Priorities for each objective are determined by using weights that are sequentially modified in each iteration The constraints are defined by imum and minimum values of temperature and EC obtained from experts knowledge that indicate optimal growing temperatures for tomato and by analyzing local data from historical series The resulting constraints are changing throughout time with a yearly pattern designed on the basis of the last twenty years collected data 3 Multilevel hierarchical control architecture The dynamics involved in the greenhouse production process present different time scales as described above namely internal greenhouse climate fast crop dynamics ie transpiration photosynthesis and respiration and slow crop development ie crop growth and fruit changes Hence a multilayer hierarchical control architecture has been proposed and used 31 Crop growth control layer Taking into account the longterm objectives market prices harvesting dates and required quality and the longterm predictions of the growth state using the modified Tomgro model Ram237。rezArias et al 2020 for the estimation of yield and profits the optimization is performed to calculate the setpoint trajectories of the inside greenhouse temperature and the EC along the considered control horizon typically 65 days for a short season 260 decision variables or 120 days for a long season 480 decision variables Models for irrigation have also been developed for control and optimization purposes The longterm weather prediction which is logically one of the elements with a higher degree of uncertainty and is performed using a software tool that accesses the weather predictions given by the Spanish National Institute of Meteorology for the next eight days forward generates patterns based on several indexes clarity imum mean and minimum temperatures and solar radiation and searches within a local historical database for a climatic sequence that better fits the generated patterns In this way taking the selected sequence as a short term weather prediction the estimation for the rest of the crop cycle is generated starting from this short sequence and using a data window from the historical database The associated high degree of uncertainties is reduced through the receding horizon approach and modifications performed in the second layer 32 Setpoint adaptation layer In this layer the setpoints to be sent to the lower layer for the next day are modified and updated in order to avoid unfeasibility problems and allow reaching the reference values These modifications are performed considering the trajectories generated in the upper layer the short term weather prediction that has a lower degree of uncertainty the current state of the crop and the short term grower goals considering hisher skill and the crop status this being a necessary degree of freedom to let the grower interact with the hierarchical control system Then this information is used within the models described above in order to simulate the greenhouse behavior and to evaluate if the provided setpoints can be reached The optimization process is repeated modifying the constraints diminishing or increasing the setpoints according to the simulation results When the setpoints are reachable they are sent to the lower layer 33 Climate control and nutrition layer Using the temperature and EC setpoints from the upper layers the controllers pute the adequate control signals for the actuators The control algorithms developed include a wide range from feedforward control adaptive control predictive control and hybrid control This list of references is evidently limited and many important papers on temperature control are not mentioned due to space constraints 4 Conclusion In this work an MO optimization problem has been proposed and tested for greenhouse crop growth management obtaining tradeoff solutions of three objectives imization of economic benefits fruit quality and wateruse efficiency This optimization scheme has been integrated into a hierarchical control architecture that allows the automatic generation of setpoints for diurnal and nocturnal temperatures and EC through a whole crop cycle using a receding horizon strategy The obtained results show logical trajectories both in short and long crop cycles The work summarizes research performed on modeling simulation control and optimization of greenhouse crop production during eight years providing real results in an industrial greenhouse 畢業(yè)設(shè)計