Research And Realization Control System Of Greenhouse Environment Based On Muti Sensor Fusion

The greenhouse intelligent monitoring technology is an economical efficient agriculturecontrol system,the current greenhouse control methods are mostly cable control modes ofsingle-factor switch,while greenhouse is strong coupling, nonlinear,serious delayed,soexisiting monitoring mode can't meet the needs of multi-point dense monitor in the mostgreenhouses.Aiming at greenhouse monitoring application demand in industrializedagriculture, in this paper,multi-sensor fusion technology is introduced into the greenhouseintelligent monitoring system, multi-factor greenhouse environment control systemsupporting vector machine is established and greenhouse wireless monitoring nodes based onZigBee are designed,which provide technical support for the intelligent control of thegreenhouse environment.The main research contents and results are as follows:(1) The structure and algorithm of multi-sensor fusion are studied and the advantagesand disadvantages of various fusion algorithms are analyzed as well. Then, this thesis focuseson the basic principle of support vector machine.Aiming at greenhouse cucumber specificneeds of greenhouse environmental parameters and regional characteristics of the greenhousein the northwest,the data fusion mode of multi-sensor system is established.(2)Wireless monitoring nodes are designed on condition that TI's CC2530ismicro-processing chip and the greenhouse environment remote monitoring system isanalyzed, combined with Specific application requirements of greenhouse environmentalmonitoring in Facility Agriculture, each monitoring node can gather multiline environmental-factor message through the I2C bus and analog quantity,and control up to four electricalequipment by electromagnetic relays (such as shade net, rolling machines, wet screens,heating equipment, etc.).(3) Recognition methods of greenhouse environment control classification arestudied. Recognition methods based on support vector machine are established. Differenttraining samples and test samples are selected on identification model for training and testing,different kernel functions are selected on support Vector Machines model. Test shows that recognition rates is as high as100%when training samples are180(the number of eachcontrol samples are20) and test samples are72(the number of each control samples are8)and kernel function are linear as well,which can meet the requirements of intelligentgreenhouse environment control.(4) The WSN optimization design is conducted,and ZigBee2007is adopted as theWireless Transport Protocol, according to the greenhouse application characteristics, from sixaspects of the device type, network channel, PANID,network topology, data transmissionmode and OSAL tasks ZigBee2007is optimized to implement the communication amongwireless monitoring nodes, root nodes and terminal nodes.The design realizes the wirelesstransmission of monitoring data and control information, making the monitoring and controlequipment distribution more flexible,lowering input costs, and improving the stability andreliability of the WSN.(5) Multi-factor greenhouse environment control prototype system is developed onMatlab(R2010a). Through testing, the system is able to identify categories of greenhouseenvironmental control accurately and quickly, which can control greenhouse environmenteffectively.