Development Of Intelligent Irrigation System Based On LSTM Neural Network

With the development of society and the innovation of science and technology,traditional agricultural farming methods are changing to modern agriculture.In view of the current low agricultural production efficiency and serious agricultural water waste,this paper designs and develops an intelligent irrigation system based on the LSTM(Long Short Term Memory Neural Network)water demand prediction model.The system uses the Raspberry Pi as the lower computer controller and the Alibaba Cloud server as the upper computer,which realizes functions such as data collection,data monitoring,water demand prediction and intelligent control of irrigation.The main findings of the paper are as follows:(1)The environmental information acquisition module is constructed.The type of industrial grade sensor is selected as needed,and the communication between the sensor and the Raspberry Part is established through the 485 bus MODBUS protocol.The information acquisition code is written by the Python language.After the Raspberry Pie,after the weather data,the meteorological data is uploaded to the database and the cloud Internet of Things platform by the HTTP protocol.(2)The LSTM neural network water demand prediction model was constructed with jujube trees as the test object.First,the daily meteorological data of Xinjiang Alar Station from 2014 to 2019,including average temperature,wind speed,air relative humidity,solar radiation,sunshine hours and The air pressure and water demand of jujube trees are the experimental data,and gray correlation analysis is carried out to calculate the degree of correlation between each meteorological factor and water demand,and then select the average temperature,wind speed,air relative humidity,and solar radiation that have a greater impact on crop water demand.Five main environmental factors,including sunshine hours,are used as the characteristic input vector of the LSTM neural network prediction model,and the water demand is the characteristic output vector of the model.The model was constructed using Python language in the Py Charm environment,and the accuracy of the LSTM prediction model was compared with the RNN model.The results showed that the fitting coefficient between the actual value and the predicted value of the LSTM model was 0.9872,which was higher than that of the RNN model,which was 0.8438,and the residual fluctuation was relatively high.Small.It shows that the model has high accuracy and can meet the system requirements.After the model is trained,an application script is written,and its code file is transplanted to the Raspberry Pi to directly call the trained model.(3)Develop the web application platform and mobile operation platform of the system Internet of Things.The system web interface and the mobile interface mainly include a real-time monitoring function of a crop growth environment and a remote irrigation switch,which is a tool interacting between people and systems.And construct a system corresponding database for storing meteorological data and irrigation data.And the alarm function is designed for the smart irrigation system.When the system is offline,or when the soil moisture is too low,the cloud will send a message prompt by the staple intelligent robot.(4)Develop system irrigation decision rules.First,the LSTM predictive model predicts the amount of water required by the sensor’s daily average meteorological data,and the prediction amount is accumulated,and when the model predicts is accumulated to the lower limit of the potential water storage,the irrigation is started,and when irrigated to the field When the upper limit of the water volume,the solenoid valve is closed,waiting for the next irrigation task.(5)After the software and hardware construction of each module of the system,the system performs functional test and irrigation test,and the results indicate that each module can work in accordance with the scheduled requirements.The system can work according to the predetermined irrigation rules in the irrigation test,and the intelligence is high.