ENHANCING CROP YIELD PREDICTION IN PRECISION AGRICULTURE THROUGH SUSTAINABLE BIG DATA ANALYTICS AND DEEP LEARNING TECHNIQUES

Abstract

Sustainable agricultural production can be planned and managed with the use of meteorological data collected by a farming Internet of Things (IoT) system. However, forecasting future trends with accuracy is challenging. Since complex nonlinear relationships with several components are a constant feature of data, in this research using a deep learning predictor with a sequential two-level decomposition structure, in which the data of weather had been split into 4 components sequentially, while recurrent gated (GRU) served as the component sub-forecast throughout training.. As a result we found, the agricultural IoT system may make more precise weather predictions. Finally, medium and long-term prediction findings were produced by GRUs' results combination. The experiments for the proposed model were validated using data of weather through Internet of Things system (IoT) in Ningxia (China), to obtain the planting of wolfberries. The results of tests of prediction discloses the suggested predictor could obtain temperature & humidity predictions accurately and satisfying requirements of precision production in agriculture.