The agricultural internet of things has been gradually popularized in agricultural planting management, but most of the application scenarios are mainly focusing on guiding growers to fertilize and irrigate crops from its acquired field data. Environmental and meteorological data of the crop’s entire growth season can also reflect the microscopic and macroscopic environment of crop growth, and hence influencing the final yield. Therefore, the environmental data (soil moisture content and soil temperature) and meteorological data (precipitation and air temperature) of a sugarcane field collected from a field IoTs system were used to build up yield predict models. The collected data was ranged from the year of 2008 to 2017, and the original BP neural network and an improved BP neural network based on genetic algorithm (GA-BP) were adopted to predict the yield. Comparing the prediction results of these two models, it was found that the prediction accuracy of the GA-BP neural network model was significantly higher than that of the BP neural network model, with its R2 reaching 0.9894, and its average relative prediction error was only 0.64%, which was also obviously lower than that of BP neural network model (5.66%). The result indicated that the GA-BP neural network was effective and feasible in predicting sugarcane production.