Research On Intelligent Control Method Of Drip Irrigation And Fertilization Based On Edge Computing

Fertilizer is one of the important guarantees for high crop yield,and the low effective utilization rate of fertilizer in China in the current season has brought a series of problems such as waste of resources and environmental pollution.Improving fertilizer utilization rate and reducing fertilizer use have become the key issues to achieve green and sustainable development of Chinese agriculture.In addition,the practical application of drip irrigation fertilization technology has problems such as uneven fertilization and lag in fertilization control.Therefore,this thesis takes drip irrigation fertilizer application device as the research object to study the intelligent control method of drip irrigation fertilizer application based on edge computing,and conduct relevant research on the agricultural data fusion and control algorithm of fertilization control process.The main work of this thesis is as follows.(1)The architecture of drip irrigation and fertilization intelligent control system based on edge computing is proposed.The functional requirements of the edge computing-based intelligent control system for drip irrigation and fertilization are analyzed in the context of the actual production requirements of drip irrigation and fertilization,etc.At the same time,the architecture of drip irrigation and fertilization intelligent control system based on edge computing is proposed to provide the theoretical basis for the research work in the subsequent chapters,taking into account the edge computing development.(2)The Adaptive Firefly Algorithm BP Neural Network(AFA-BP NN)multi-sensor data fusion algorithm for edge nodes is studied.A multi-sensor data fusion architecture for edge nodes was proposed based on the analysis of data fusion functional requirements of agricultural situation monitoring data in field.To address the problems of network transmission congestion and redundant data processing caused by a large amount of sensor data to be processed in the process of farm monitoring in large fields,the BP neural network is optimized using a firefly algorithm,and the AFA-BP multi-sensor data fusion algorithm for edge nodes is proposed.The data fusion algorithm processes environmental information from multiple field sensors to reduce the redundancy of field sensor data and data transmission volume,and provides data support for intelligent control of drip irrigation and fertilization.(3)A PID control parameter optimization Algorithm based on Partial Attraction Adaptive Firefly Algorithm(PAAFA)is proposed.In order to optimize the PID controller parameters of drip irrigation fertilizer application device and improve the PID control effect in drip irrigation fertilization,this thesis establishes a mathematical model of drip irrigation fertilizer flow control system;and a PID control parameter optimization algorithm based on partial attraction adaptive firefly algorithm(PAAFA)is proposed for finding the optimal combination of K_p,K_i and K_d parameters of PID control to reduce the control system errors caused by time lag characteristics in the fertilization process and improve the fertilizer control accuracy of the fertilizer control system.The simulation results show that the overshoot of the unit step response of PID control based on PAAFA algorithm is smaller and the adjustment time is shorter compared with other parameter optimization algorithms.(4)The algorithm validation and drip irrigation fertilization control experiments were conducted.Using the collected field sensor data and soil sample testing data,the BP neural network training,testing and result analysis of the proposed AFA-BP multi-sensor data fusion algorithm were conducted;the results showed that the data fusion accuracy of the AFA-BP fusion algorithm was improved by 3.3%compared with the BP neural network.Secondly,the drip irrigation fertilizer control test and the test results analysis were conducted in combination with the drip irrigation fertilization bench test platform;the results showed that the average relative error of PID control based on PAAFA algorithm was reduced by 3.99%,3.50%and2.42%respectively compared with GA,FA and AGA parameter optimization algorithms,which effectively improved the control accuracy of fertilization flow rate.