| With the development of digital agriculture technology, it has become a key issue that how to monitor the agricultural crop conditions accurately and in real-time. Recent years, the development of ZigBee technology grows fast in the farmland monitoring field. With the help of the wireless sensor network technology, establishing an agricultural environmental monitoring system, which possesses the characteristic of positioning and low power consumption, has far-reaching significance and application value. The system can monitor temperature, humidity, light intensity, carbon dioxide concentration, soil moisture, soil pH value of farmland and other data through a variety of sensors, meanwhile, in order to support mobile monitoring, I also add positioning function for the sensor nodes, so the growing environment of each place can be obtained, these could provide a precise regulation for agricultural management and could improve the yield of crops.The system adopts the NXP company’s JN5148 as the basic platform, realize the development of ZigBee Pro protocol, and do some research on the following aspects:Firstly, the issues of power supply and low power consumption for the mobile monitoring sensor nodes were discussed. The system solves these problems from two aspects:on the one hand, a solar battery board, a charging circuit and a protection circuit were combined into a charging management module, which is used to charge the lithium battery and power to the system; on the other hand, I researched four most probable operation modes for each sensor node, and respectively studied and compared the average operating current in a whole working cycle according to the actual usage of each node, so the most energy-efficient mode can be obtained, at last Icalculated each node’s continuous working time under the worst weather conditions, in which the solar panels can not charge to each node.Secondly, the multi-protocol outputs of the coordinator node were discussed, which achieves three ways of RS-232, RS-485 and Ethernet to upload collected data from each sensor node. Among them, the RS-485 mode achieved the MODBUS RTU protocol. By embedding the TCP/IP protocol stack in JN5148, Ethernet communication was implemented, which avoid the traditional need of relying on the gateway devices with ARM or DSP to achieve the conversion of ZigBee PRO and Ethernet protocols.Thirdly, the mobile network of sensor nodes was discussed. The system designs a intelligent networking mechanism for the sensor nodes, which allows sensor nodes to identify network problems, and the nodes can automatically delete the saved network parameters and re-networking, it makes WSN networks have higher self-healing capabilities.Forthly, the localization of sensor nodes was discussed. In this system, TOF and RSSI are assembled in ranging, due to the presence of randomness in ranging, which are not conducive to the convergence of neural network, a second-order Butterworth low-pass filter was used to preprocess the data, and then machine learning methods are applied in the data subsequent processing, BP neural network, RBF network, SVM network were respectively added in specific processing, by comparing the fitting curve I found SVM had the best effect, finally trilateration was used for positioning.Finally, the completed work has been summarized in this thesis, the inadequacies were pointed out, and the improvement direction and improvement goals were proposed for the next phase of work. |
PDF Full Text Request