Research On Crop Water Demand Calculation Model Based On BP Neural Network And Internet Of Things

Water resources is an important factor closely related to human life and economic production.However,there are not many freshwater resources available for human use on the earth,which cannot meet the demand for water consumption brought about by population growth and economic development.Contradiction in water distribution.As a large agricultural country,China's annual agricultural water consumption ratio has been entrenched at the top of the country's total water consumption,reaching 61.2%in 2019.At the same time,there is a lack of efficient use of water resources in agricultural irrigation,and the utilization rate is only about50%.Therefore,China was vigorously promote water-saving irrigation as a basic national policy of modern agriculture.However,the research on water-saving irrigation technology in my country focuses on the improvement of irrigation efficiency.The lack of research on deep-level crop water requirements(ETc)cannot be regarded as real water-saving irrigation.In practical applications,to estimate the water requirement of a specific crop,first calculate the reference crop evapotranspiration(ET₀),and then revise it in conjunction with the appropriate crop coefficient.Therefore,the calculation of ET₀ has an important position in water-saving irrigation technologyAiming at the problem that the ET₀ calculation process is complicated,involveing many factors,and the existing calculation formulas cannot be applied to different locations.A computational model(NDPSO-BP)based on the combination of neural network and improved particle swarm optimization(NDPSO)is proposed in this paper.First,the average influence value method(MIV)and SPSS software is used to reduce the dimensions of the affected variables,and the first three main factors of independent distribution are selected as input;then the adaptive nonlinear decreasing weight update factor and Levy Flight are used.Combine the improved NDPSO algorithm to determine the best weights and thresholds of the BP neural network,and enhance the prediction fit of the BP neural network.The experimental results show that under the same experimental data and experimental platform,the predicted value of the NDPSO-BP model proposed in this paper is better than that of the PSO-BP model and the BP model;at the same time,the predicted value of the algorithm at different locations is better than the two basic algorithm,suitable for promotion and application in different locationsOn the other hand,Based on the advantages of Agricultural Internet of things,a crop information collection system based on STM32 platform is designed in this paper.The information collection terminal monitors the environmental information of crop growth through temperature,humidity and wind speed sensors,and transmits it to the information processing terminal through the Zig Bee protocol.Based on the optimal weight threshold of the trained neural network;A real-time crop water demand calculation model was built on the the information processing terminal,and transmits relevant data to the Gizwits cloud platform through the 4G module,users can view cloud information through their mobile phone to helping make timely and accurate decisions.