Research On Method And Technology Of Light Environment Control Of Facility Based On Crop Photosynthetic Demand

Building micro-climate environment meeting the photosynthetic demand of greenhouse plant is the key to improving photosynthetic rate, speeding up substance accumulation, and increasing yield and quality. In northern China, lacking of illumination exists widely in greenhouses during late autumn to early spring, resulting in slow growing and developing of crops, and low fruit setting rate. The technique of artificial lighting is to regulate light environment, and it is an effective way to optimize crop photosynthetic rate. The currently available systems lack quantitative physiological basis, with simple and rough controlling method, which results in problems such as inadequate or excessive lighting. The crop photosynthesis was affected by the cross influence of several key environmental factors, which leads to light environment controlling target parameters change dynamically with these key factors. Based on the analysis of the effect of environment on photosynthetic rate, in this thesis, we study the multiple factor correlated plant photosynthetic rate modeling method, and the dynamic optimization method for regulation target value. Furthermore, we proposed a light environment controlling system based on multi-sensor information fusion and wireless sensor network architecture, which is verified by conducting a series of experiment. This study is valuable on the theory, method, and application of exploring precision management of greenhouse agriculture. The main content and conclusion of this thesis is as the following.?1? Carrying out mechanism analyzing and experiment on the keys environmental factors that affect plant photosynthetic rate. Based on the analyzing of the affection of environmental factors on photosynthesis, combination experiment on photosynthetic rate under the condition of nested temperature, photon flux density and CO₂ concentration was designed. Using Li-6400 XT photosynthetic system, the photosynthetic rate of tomato seedling were measured with respective to 10 luminous flux density gradients, 5 CO₂ concentration gradients and 6 temperature gradients, which resulting in 300 combined conditions, and 1200 groups of experiment result. These data were further utilized to analyze the affection of temperature, photon flux density and CO₂ concentration on photosynthetic rate. It is shown in the experiment result that all these three factors affect photosynthetic rate significantly, with each factor having different response curve, and being affected by other factors significantly. Therefore, multi-factor correlation should be considered for light environment regulation. In this thesis, we carry out the research on the model and controlling method of photosynthetic requirement based on the above observation.?2? Proposing the modeling method of photosynthetic rate based on genetic neural network. By this method, during the process of constructing the neural network prediction model, the initial weight matrix is optimized by the genetic algorithm, so as to make it cross the local flat area fast. For the tomato seedling photosynthetic rate testing data, the model training target could be archived with only 17 steps, with faster neural network convergence than the way to initialize weight matrix randomly. It is shown in the experiment result on photosynthetic rate model of tomato seedling that the determination coefficient of correlation analysis between measured value and predicted value by genetic neural network model is 0.9856, with straight slope being 0.994, intercept being-0.01092, and the error ranging from 0.1 to 0.4?mol/m2 s. Generally speaking, these indexes are all superior significantly to the neural network model without optimization. The above research shows that the genetic neural network modeling method can generate photosynthetic rate model with high precision, which provides a quantitative physiological model for the greenhouse light environment regulation.?3? Proposing the target value optimization method for light environment regulation based on genetic algorithm. Based on genetic neural network photosynthetic rate model, for 117 groups of various temperature and CO₂ concentration combination, the optimal light saturation points were obtained. Then, the target value model of light environment regulation was constructed, with determination coefficient R2 of the model being 0.982, so as to realize the dynamic prediction of light saturation points which takes temperature and CO₂ concentration as parameters. It is shown in the experiment result that the determination coefficient of correlation fitting between predicted value and measured value is 0.920, with slope of the fitted line being 0.989, ordinate intercept being-0.233, which showed that these two values are linear correlated, with the maximal relative error less than 6%, which shows that the proposed model had a high accuracy.?4? Proposing the optimizing method of the light environment regulation based on improved artificial fish swarm algorithm. Genetic algorithm has the defect of strong global searching ability and weak local searching ability. The traditional artificial fish swarm algorithm is improved by utilizing dynamic adjustment of vision and step, so as to improve its optimization precision, and avoiding the slow convergence speed problem of artificial fish swarm algorithm. The proposed algorithm can realize optimizing in 8 steps, with almost the same convergence speed as the genetic algorithm. The model has excellent fitting performance, with determination coefficient R2 being 0.993. It is shown in the experiment result that, the determination coefficient of correlation fitting between predicted value and measured value is 0.976, which showed that these two values are highly linear correlated, with the maximal relative error less than 2%. Hence, the target value model generated by the improved artificial fish swarm algorithm is better than the model generated by genetic algorithm.?5? Researching and developing the light environment controlling system based on wireless sensor network. Based on analyzing the light environment regulation requirement of cultivation process, the system architecture for wireless sensor network was proposed, integrating light environment controlling target value model and multi-sensor information fusion into closed-loop feedback controlling mechanism. Then, the light environment control system is developed, which consists of intelligent controlling nodes, environmental monitoring nodes and quantitative supplementary lighting nodes. A network communication protocol between nodes based on ZigBee protocol is proposed, and the hardware and software of each functional node is developed based on CC2530 processor, so as to realize real-time monitoring the environmental information inside the facilities, including the photon flux density, CO₂ concentration, temperature, etc., with the ability to automatically predict target value, and dynamic variable control of the supplementary lighting nodes.?6? Performing system integration testing and experiment result analysis. It is shown in the system performance test that the relative error of red-blue light sub-band detection was less than 5%, and the relative error of the output is less than 4% after utilizing the correction model. The overall deployment plan of the light environment controlling system was designed, and it is deployed in the Yanliang vegetable test and demonstration station. Compared with supplementary lighting system of fixed target value, the experiment result shows that the proposed system could predict the light saturation point based on real-time temperature and CO₂ concentration, and adjust the target value dynamically, so as to achieve on-demand and quantitative light regulation. Compared with the fixed target value system, the proposed system could reduce 30% of the power consumption while meeting the requirement for crops growing.