Study On The Prediction Of Mango Yield Based On The Self-attention C-BiGRU

Mango target yield prediction is of great significance for understanding the trend of mango yield,planning the development of mango production,Strengthening the defense and governance capabilities of climate disasters and promoting agricultural informatization.There are many related meteorological factors affecting mango yield,and the relationship between them and the yield is complicated,which is difficult to describe accurately with mathematical functions.A Bidirectional Gated Recurrent Unit and Convolution Neural Network based Self-attention(Self-attention C-Bi GRU)combined model with long short-term memory function is constructed to solve the problem.The model can learn the intrinsic relationship between complex meteorological data and the yield,thus a more accurate Mango yield prediction model can be obtained.Using the Data Castle Big Data Competition platform for Beijing PM2.5 meteorological data and atmospheric pollutant data from January 1,2010 to December 31,2014,this data is similar to the environmental information affecting mango production,including multiple features and the time series data is used for reliability verification test.The model has been improved to a certain extent in the prediction accuracy,and the reliability of the Self-attention C-Bi GRU model is verified.At the same time,the Self-attention C-Bi GRU model is compared with other deep learning models to verify the effectiveness of the Self-attention CBi GRU model.The Self-attention C-Bi GRU model for mango yield prediction was established by using the data of meteorological factors and mango yield collected from three meteorological stations in somewhere of Guangxi.The data are including 24 years of mango production cycle(from the 22 nd ten days of the previous year to the 21 st ten days of that year),and there are nine meteorological factors in per ten days.The experimental results show that the root mean square error between the actual yield and the predicted by the Self-attention C-Bi GRU model is 10.67,which is 37.7%,42.1%,17.6%,4.1%,5.3%,5.9% lower than that of the Support Vector Regression(17.14),the Back Propagation neural network(18.43),the Gated Recurrent Unit(12.95),the Bidirectional Gated Recurrent Unit based attention combined model(11.13),the Gated Recurrent Unit and Convolution Neural Network based combined model(11.27),the Bidirectional Gated Recurrent Unit and Convolution Neural Network combined model(11.35).The predicted value of Self-attention C-Bi GRU is in good agreement with the actual mango yield and has high prediction accuracy.