Research On Intelligent Irrigation Control System Of Jujube Based On IoT

Although drip irrigation system can greatly improve water use efficiency,it cannot achieve intelligent irrigation in a large area.However,if drip irrigation technology is combined with intelligence,it can not only realize real-time accurate irrigation,but also realize unmanned operation and effectively reduce the labor cost.In addition,intelligent irrigation systems have achieved remarkable results,and they are widely used in greenhouses,fields and other areas to achieve monitoring,automatic irrigation and other functions.However,it is found that there is less research on the realization of intelligent agricultural irrigation system in Xinjiang,especially the research on special forest fruits such as date palm in South Xinjiang is rarely mentioned,including the prediction of meteorological indicators related to crop irrigation based on deep learning is rarely seen.Based on the above,This research developed a set of information acquisition devices for date palm orchards based on agricultural Internet of Things to achieve intelligent irrigation,and carrying out deep learning to establish prediction models based on historical information to lay the foundation for intelligent decision making for irrigation used in different growing periods of crops can provide an important technical support for the development of intelligent irrigation in Xinjiang date palm orchards.The main conclusions are as follows.(1)Based on human-computer interaction technology,a monitoring system for date palm orchards is designed.The system consists of field terminal and remote client to realize the functions of monitoring,control,real-time display and alarm.The system connects the field terminal and the remote client through serial port to complete the data sending and receiving,establishes the communication between the sensor and MCU by I2 C protocol,establishes the communication between the storage device and MCU by SPI protocol,and uses 4G module as the remote data transmission medium and Modbus RTU protocol as the remote transmission protocol.The computational analysis part of the system uses the programming software Keil 5to realize the functions of data acquisition and monitoring,analog-to-digital conversion,database comparison,computational analysis and irrigation decision,and finally verifies the accuracy of the monitoring system.(2)Based on the monitoring system,the intelligent irrigation control part of the system was designed.The irrigation process is designed based on the environment of date plantation in South China,which was divided into manual and automatic modes.The system defaults to automatic mode.In the automatic mode state,the upper and lower limits of the relative water content of the soil in the root zone suitable for the growth of jujube trees are used as the thresholds for irrigation decisions,and the system automatically controls irrigation according to its thresholds and records irrigation information and early warning at the same time.The test results show that the system can irrigate automatically according to the set thresholds.(3)A model is developed to predict air temperature,relative air humidity and evaporation.The model uses Tensor Flow’s high-level API,the Keras deep learning framework,including the use of Keras to build a Sequential model,and Sequential to build a deep neural network and implement a time series type prediction based on the LSTM algorithm.The prediction models selected the meteorological data of the National Greenhouse Data System Alar site from 2009-2018 as training features,and predicted the average temperature,average relative air humidity and evapotranspiration for 2019,respectively.The prediction results demonstrate that the loss curves of each model show a decreasing trend,and the prediction curves are close to the actual curve trend and there are no points with large data errors.This indicates that the model is more reasonable and the prediction results are more accurate.