A deep neural network-based decision support system for intelligent geospatial data analysis in intelligent agriculture system

Abstract

Agriculture is an important field for most of the countries, as it is the primary source of food production, and has a great role in growing the economy of a country. The world’s population is increasing with a faster speed, and as a result, the need for food is rapidly increasing. Farmers’ traditional techniques are insufficient to meet the rising demand of food. The agriculture sector faces various challenges such as producing more and better products while enhancing the sustainability through the smart use of natural resources, minimizing environmental harm, and adapting to the climate change. Agricultural informatization is the inevitable trend of modern agricultural development. The purpose of introducing information technology into agriculture is to save production costs, improve production efficiency, and accelerate the development of productivity. Agricultural informatization is an intelligent, digital, and a networked-based system. Geographic information system technology is widely used in agriculture, such as precision agriculture, land resource management, crop yield estimation and monitoring, and soil and water conservation. The characteristic of expert decision-making system is the logical reasoning of knowledge, and the advantage of neural network is the acquisition of knowledge. Therefore, the study of the organic combination of neural network and expert decision-making system technology is of a great significance to the research of integrated system righteousness. In this paper, the agricultural data obtained from the expert database are displayed in the form of a tree list and are used in the process of system design. The geospatial data can be uploaded through the map loading function, find the map path, and easily uploaded by modifying the expert database. In order to realize the scalability of the platform, we transplanted the analysis objects to other provinces and cities. Further, the grey decision-making system and back-propagation neural network (BPNN) prediction models are used to predict the future indicators of agricultural data. Based on the historical data of agricultural economic indicators, the effect of predicting the future value of agricultural indicators by using grey decision-making system and neural network models is repeatedly tested. The experimental result shows that the BPNN model performed really well in terms of prediction accuracy and relative error rate as compared to the grey decision-making system.