US 11,741,555 B2
Crop yield estimation method based on deep temporal and spatial feature combined learning
Tao Lin, Zhejiang (CN); Renhai Zhong, Zhejiang (CN); Jinfan Xu, Zhejiang (CN); Hao Jiang, Zhejiang (CN); Yibin Ying, Zhejiang (CN); and Kuan-Chong Ting, Zhejiang (CN)
Assigned to ZHEJIANG UNIVERSITY, Zhejiang (CN)
Appl. No. 17/777,053
Filed by ZHEJIANG UNIVERSITY, Zhejiang (CN)
PCT Filed Oct. 29, 2020, PCT No. PCT/CN2020/124870
§ 371(c)(1), (2) Date May 16, 2022,
PCT Pub. No. WO2021/098472, PCT Pub. Date May 27, 2021.
Claims priority of application No. 201911134971.4 (CN), filed on Nov. 19, 2019.
Prior Publication US 2022/0405864 A1, Dec. 22, 2022
Int. Cl. G06Q 50/02 (2012.01); G01W 1/02 (2006.01); G06N 3/049 (2023.01); G06N 3/084 (2023.01); G06Q 10/04 (2023.01); G06N 3/045 (2023.01)
CPC G06Q 50/02 (2013.01) [G01W 1/02 (2013.01); G06N 3/045 (2023.01); G06N 3/049 (2013.01); G06N 3/084 (2013.01); G06Q 10/04 (2013.01)] 6 Claims
OG exemplary drawing
 
1. A crop yield estimation method based on spatio-temporal deep learning, comprising the following steps:
step 1): obtaining and preprocessing a historical crop yield data and a historical meteorological data of a region, preprocessing the historical meteorological data to obtain meteorological parameters, preprocessing the historical crop yield data to obtain a detrended yield, and taking respectively the meteorological parameters and the detrended yield as input data and output data of a crop yield spatio-temporal deep learning model;
step 2): constructing the crop yield spatio-temporal deep learning model, and optimizing hyperparameters of the crop yield spatio-temporal deep learning model;
step 3): forming a training set sample by taking the meteorological parameters obtained in step 1) as the input data and the detrended yield obtained in step 1) as the output data so as to train the crop yield spatio-temporal deep learning model, obtaining optimal parameters of the crop yield spatio-temporal deep learning model after multiple rounds of the training, and then obtaining a trained model; and
step 4): inputting the meteorological parameters of a crop yield to be estimated into the trained model, and outputting an estimation result to obtain a crop yield estimation result.