Reaserch On Prediction Of Rice Yield In Hunan Province Based On BP Neural Network

Rice production in China has a long history and plays an important role in the development of Chinese civilization,which has always been attached great importance by the state.Document No.1 of the Central Committee of the People’s Republic of China in2022 proposed to ensure stable and increased yields,which put forward higher requirements for accurate implementation of policies to master the law of rice yield change.There are many factors that affect the rice yield.Mining the influence degree of these factors on the rice yield and predicting the rice yield can provide decision basis for macro-regulation and control of stable and increased production and grain reserve,and build a precaution and warning safety line for rice production.This study takes rice yield in Hunan Province as the research object.In view of the good non-linear system mapping ability of BP neural network,the method of optimizing BP neural network was used to build the rice yield prediction model.The main work is as follows:(1)The TIPSO-BP prediction model is constructed.First,a BP network prediction model was built by using grey correlation analysis to screen the important factors affecting rice yield.Then,the TIPSO algorithm is optimized by introducing a hybrid strategy of dynamic inertia weights and Tent chaotic mapping.Finally,TIPSO-BP model is constructed by using TIPSO algorithm to optimize BP network model.The single-strategy optimization model and the BP network model are used to validate and analyze.The results show that tipso-bp model has the highest prediction accuracy,and the Mean Absolute Percentage Error(MAPE)is 77% lower than BP neural network model.(2)The Sparrow Particle Swarm Optimization(SPSO)is proposed.To solve the problem of slow convergence speed and low convergence accuracy of the Particle Swarm Optimization,the sparrow particle swarm optimization is proposed by improving the sparrow search algorithm and combining with the optimized particle swarm optimization.The algorithm first improves the finder’s search method in the sparrow search algorithm,then explores the optimal solution by using the improved finder’s behavior,then further explores the optimal solution by using the particle swarm behavior,and finally explores the new optimal solution by using the investigator’s behavior to avoid falling into the local optimal solution.By comparing the results of six other optimization algorithms on the basis function,it is proved that the sparrow particle swarm algorithm has better convergence accuracy and speed than the particle swarm optimization.(3)The TSPSO-BP prediction model is constructed.The TSPSO-BP prediction model is built by SPSO optimization TIPSO-BP model.The prediction error,prediction accuracy and convergence curve before and after optimization are compared and analyzed.The results show that the Mean Absolute Percentage Error(MAPE)of TSPSO-BP model is 62% lower than that of TIPSO-BP model and 91% lower than that of BP neural network model.