Design and experiment of jujube intelligent irrigation system based on Stacking integrated learning

In the south of Xinjiang, the rainfall is low, the climate is dry, agricultural water is scarce, and water conservation is particularly important. An intelligent irrigation system is designed to solve this problem. The system uses Alibaba Cloud server as the upper computer and raspberry pie as the lower computer, and sets up corresponding operation pages. In this paper, according to the meteorological data required in Penman-Monteith formula, the water demand of the past seven days and the meteorological data of the previous day as the input vector, and the crop water demand as the output vector, the Stacking integrated learning prediction model based on random forest, BP neural network and ridge regression is constructed. The results show that the fitting coefficient R~2 of Stacking integrated learning prediction model is 0.973, and the three types of errors of MAE, RMSE and MAPE are smaller. The prediction effect of Stacking integrated learning prediction model is stronger. In the irrigation experiment, the automatic irrigation decision is correct, and the system operates stably, which provides ideas for improving the utilization of water resources in agriculture in Xinjiang.