Design And Implementation Of The Greenhouse Monitoring System With The Ability Of Enhanced Learning

With the development of information technology,the automated control technology is widely used in the greenhouse monitoring systems.However,there are still some unclear issues in the traditional greenhouse monitoring systems such as low level of intelligence,poor portability and adaptability and high complexity.Therefore,it is of great significance to develop a greenhouse monitoring system with adaptive control functions.This thesis designs and realizes a novel greenhouse monitoring system with the ability of enhanced learning,which is based on the full investigation of the realistic requirement of the greenhouse planting and overcomes the disadvan tages in traditional greenhouse monitoring systems.The main research contents are: 1)greenhouse Monitor the design and implementation of the upper computer and the lower computer;2)Enhance the application of the learning algorithm in the greenhouse greenhouse monitoring system.Firstly,according to the functional requirements of the system's lower computer,four software and hardware solutions for the data acquisition end,field control terminal,solar electronic insecticidal device and video monitorin g terminal of the lower computer are proposed.The data acquisition end uses a wireless mesh network(Mesh network)technology to build a sensor network,which utilizes the power supply technology and the intermittent sleep mode of the sensor to reduce the operating power consumption of the data acquisition end;the field control terminal uses a programmable logic controller(Programmable Logic Controller(PLC)is the main controller,which realizes the environment receipt forwarding and control command receiving function by using the serial port server.The solar electronic insecticidal device tracks the solar timing through the light source tracking algorithm,improves the solar-electric energy conversion efficiency,and realizes effective control of pests.Secondly,the design and implementation process of the monitoring cloud platform and application terminal of the host computer are elaborated.Using MVC design pattern and browser/server software design architecture,the implementation process of user modules such as user management,equipment management,historical data query,basic information management and data reporting in greenhouse greenhouse system is elaborated.In-depth study of the theory of enhanced learning algorithms,found that enhanced learning algorithms can achieve control optimization goals with a small amount of experience,and avoid the "state explosion" problem.In the implementation of the enhanced learning algorithm,a value-based approach is adopted,that is,the value of the action is judged by the bonus value,and then the action with the largest total value of the expected reward accumulation is selected in the action group.Through the formal description of the state space and action space,the environmental state-the calculation of the value of the device action and the selection and update of the action strategy,the adaptive control of the greenhouse is realized.The PC-side webpage was designed by Web front-end technology,and the mobile APP was designed by Android technology,which realized the visual display of system data and the remote control of the device.Finally,the functional test of the proposed greenhouse monitoring system was conduced.The test results show that the greenhouse monitoring system works well and meets the requirements of the greenhouse users.Subsequently,the designed greenhouse monitoring system was used in practical situations,and the experimental results show that comparing with the methods of rule base and BP(Back Propagation)neural networks,the proposed method in this thesis will shorten the growth time of crops by 2-3 days and increase the yield by 10%.