The invention provides a method, a device, equipment and a medium for predicting hundred-
grain weight of corn, and the method comprises the following steps: firstly, converting
genotype data of each dimension into a numerical form according to
genotype frequency corresponding to the
genotype data of each dimension of a to-be-predicted corn material; then, inputting the multi-dimensional genotype value into a prediction model to obtain a hundred-
grain weight prediction value of the to-be-predicted corn material; according to the prediction model, a deep space-time network is constructed by fusing a deep residual network and a bidirectional long-
short term memory recurrent neural network, and the deep space-time network is iteratively trained by using multi-dimensional sample genotype data and hundred-
grain weight phenotype data of sample corn materials of various different strains. According to the method, the prediction model is trained on the basis of the deep space-time network, the nonlinear relation between the multi-dimensional genotype data of the sample corn material is fully learned and captured, the efficiency and accuracy of corn hundred-grain
weight prediction are remarkably improved, and then the efficiency and accuracy of corn breeding selection are improved.