Research On Anti-noise Processing Method Of Grain Production Signal Based On EEMD

Grain production prediction is one of the most important key links in precision agriculture(Precision Agriculture,or PA).Through grain production information,we can obtain the information affecting crop growth accurately and quickly,and combined with other important links of precision agriculture,it makes precision agriculture develop towards a new agricultural direction,which is timing,positioning and quantitatived.In grain production,mechanical noise caused by difference of the field terrain or the vibration of the grain harvester will be mixed in the grain production signal,which will affect the accuracy of grain production.At present,researchers at home and abroad mainly focus on denoising algorithms in order to improve the accuracy of grain production prediction,although there was a certain denoising effect,these algorithms were complex and adaptiveness was not good enough in decomposing the signal,and prediction accuracy had yet to be further improved.Aiming at this problem,a wavelet threshold denoising algorithm based on Ensemble Empirical Mode Decomposition(Ensemble Empirical Mode Decomposition,or EEMD)was introduced,that is,adaptive EEMD wavelet threshold denoising algorithm,the algorithm is a new type of data-driven algorithm.It is not necessary to determine basis function and number of decomposition layers in advance.It only needs to implement self-adaptive decomposition according to characteristics of signal itself,signal is decomposed into IMF components from high frequency to low frequency,and the demarcation point of high and low frequency IMF is determined by continuous mean square error criterion.This paper innovatively combined the advantages of wavelet threshold denoising algorithm,high frequency IMF component with low SNR was de-noised to avoid the loss of useful signals caused by direct removal of high-frequency components,and finally signal was reconstructed.Test result showed that denoising effect was obvious,and production precision was effectively improved.This article first introduced the research background of grain production prediction,and current situation of domestic and foreign production precision systems;Then basic theory of wavelet and EEMD was discussed;and de-noising scheme was emphatically designed;finally,a simulation model was set up in the Matlab/Simulink to perform corresponding algorithm processing on the raw grain production signal to verify the effectiveness of the algorithm.A test platform based on STM32F103C8T6 ARM controller was built,and the hardware circuit of signal acquisition and control part,data display and output part was designed,made selection of each hardware device,and performed hardware debugging.The field test shows that the average error was 1.911%,and the maximum relative error was less than 3%.The experiment proves that the adaptive EEMD wavelet threshold denoising algorithm used in this paper has good denoising effect and can improve prediction accuracy,which fully demonstrates that production prediction system used in this paper can meet the needs of production prediction and can provide some new ideas on production prediction for the future research.