This study was conducted to create a decision-making model that can economically approach the initially set goals of increasing production, controlling the timing of production, and the quality of the final fruit through big data analysis of environme...
This study was conducted to create a decision-making model that can economically approach the initially set goals of increasing production, controlling the timing of production, and the quality of the final fruit through big data analysis of environmental factors and crop growth in smart farms at the time of the paradigm shift from conventional experience-oriented agriculture to digital-oriented agriculture. For this purpose, environmental, growth, and management data were collected from tomato farmers in Gyeongnam Province since 2018, and based on this data, factors related to production were analyzed and a prediction model was created. As a result of the analysis, considering that it takes 7 weeks for tomatoes to bloom and be harvested, we extracted the main factors affecting yield by combining environmental, growth, and yield data for 7 weeks per plant, and then predicted yield per 3.3m2 using a deep neural network (DNN) based on these factors. In addition, by applying the ensemble technique, it seems that five major growth factors such as stem thickness and leaves can be predicted by the current week's environment. Based on the results of this analysis, it is expected that digital agriculture in the Gyeongnam region can be implemented through the development of models that improve productivity and profitability by integrating and managing smart farm measurement big data and applying the analyzed results in smart farm farms, and through on-farm demonstration.