The Intelligent Greenhouse Monitoring System Based On Convolutional Neural Network

The previous greenhouse control systems could not properly maintain the environment conditions within the most appropriate limits for plants in the greenhouse through automatic control due to limitations of various factors.This paper integrated convolutional neural network into the intelligent control system to control the greenhouse.The essence of this article is the use of greenhouse parameter data,when dealing with greenhouse data.Due to the gradual changes of 7 greenhouse parameters,the volume of data is huge with no obvious features,which result in the difficult data processing and difficult choices of optimization algorithm.After the systematic research in the structure of convolutional neural network and training algorithm,this paper adopts cntk deep-learning framework to construct a convolutional neural network training platform.Firstly,it introduces the development history of convolutional neural network including application fields,features and etc.Then this work introduces the acquisition methods of training data and the integration of input data.After that,the whole structure of neural network and the various optimization algorithms involved are analyzed,and the network training environment is successfully configured.The training results of five kinds of network structures are listed.The most appropriate network structure is combined with the greenhouse and the results are tested,analyzed and summarized.The experiment proves that the intelligent greenhouse monitoring system based on convolutional neural network has obvious advantages and adaptability in handling complex environment control.