Research On Intelligent Greenhouse Monitoring System Based On Deep Learning

Artificial intelligence technology is the core of the development of science and technology in the 21st century and has an inestimable effect on social progress and economic development.As the fastest growing technology in the field of artificial intelligence in recent years,deep learning has achieved very good application results in the fields of computer vision,speech recognition,and natural language processing.Although the relevant technical research on intelligent control of greenhouses has been around for 50 years,many control methods have been proposed,but since greenhouse greenhouse systems are a complex system with multiple inputs,multiple outputs and strong coupling,the results of current intelligent control technologies are Not satisfactory.The research on the deep learning method with super-learning and expressive ability in the field of intelligent control of greenhouses has just begun,and it is expected to make breakthroughs in the field of intelligent control of greenhouses.This thesis studies the smart greenhouse monitoring system based on deep learning.The main work is as follows:Firstly,the research status of greenhouse greenhouses at home and abroad is summarized.The model structure of common neural network and neural network with cyclic structure(RNN)is introduced.Experiments and analysis are carried out in greenhouse greenhouse environmental factor prediction.Deep learning greenhouse greenhouse intelligent control system.Then,for the problem that the environmental factors change slowly,the data volume is large,and the switching state of the controlled device is discrete,a prediction model based on the improved LSTM structure to construct the environmental factors and the controlled state of the controlled device is proposed.The training result of the model is controlled by the greenhouse.System decision-making;for the problems of gradient disappearance,gradient explosion,over-fitting,etc.in model prediction,the strategy of using Relu activation function and adding regularization method is proposed.Experiments show that the optimized prediction model has better convergence effect than before.Finally,by transplanting the training results to the Raspberry Pi,the Raspberry Pi connects to the control box via the LAN and transmits information to each other,realizing the control of the greenhouse environment.The research based on the deep learning wisdom greenhouse monitoring system is conducive to solving the problem of difficult control and low precision in the greenhouse,and making the greenhouse environment control more efficient and effective.