Smartfarm DSS Framework for Decision Support System on City Smart Farm with Deep Neural Network and Case-Based Reasoning

Smart urban farms have become one of the trends of modern cities and their ecosystems. The trend in the development of such farms is full automation and minimizing human involvement. At the same time, problematic situations that require prompt and competent intervention by specialists, often at the expert level, are not excluded on a smart farm. In particular, this is the occurrence of plant diseases or conditions for them. Such situations do not appear on a daily basis; however, they can critically affect the results of the city farm. In these conditions, it is important to supplement the software and hardware complexes of smart farms with decision support systems. The paper presents a framework for such a system. The framework implies three modes of operation: visual monitoring of plants for disease detection, monitoring of microclimate and nutrient medium parameters, and a question-and-answer system. The proposed architecture has been tested in the implementation of the DSS software prototype. The peculiarity of the framework presented in this work is that it offers an architecture and methodology for creating such DSS, where deep learning and knowledge engineering methods are organically combined the case-based reasoning method, which is effective for inferring solutions.