Adaptive decision support system for smart agricultural crop cultivation to support food safety standard

Abstract

Smart Agriculture is a trend that emphasizes the use of information and communication technology in agriculture field Smart Agriculture assists in reorient agricultural systems and guiding actions required to modify in order to ensure the food safety and support development effectively (production and quality of yields) in an ever- changing climate. Smart agriculture technology is made up of various technological information and implementations, such as smart sensing and monitoring, smart analysis and planning, could-based computing, smart control to manage cultivation production as an automatic system, decision support system, or expert system. At the moment, the existing decision support system for smart agriculture relies solely on data collected by intelligent sensors. According to experts, this is insufficient for making crop-maintenance decisions. Furthermore, existing knowledge base systems or expert systems rely solely on data captured from human experts, which is insufficient for crop maintenance decisions. As a result, improving the functionality of decision-making systems requires combining smart technologies and human experts to provide appropriate recommendations to farmers based on data collected from smart sensors and technologies and expert judgment, which is the primary goal of this research. As a result, in this thesis, an Adaptive Decision Support System (Adaptive DSS) was proposed to analyze data collected through the use of intelligent technology, both on and off the farm. Including rules for caring for plants in each growing season obtained from expert knowledge. Through a chat application, farmers will be given appropriate advice on how to care for their crops during each growing season or crop life-cycle, which includes watering, fertilizing, pest control, crop disease control, and weed control. This research is divided into three sections: Knowledge Engineering (KE), Adaptive Decision Support System (Adaptive DSS), and information and knowledge representation. The first section involves using knowledge engineering (KE) approaches to capture knowledge from experts, extract knowledge from experts, and create rules for crop management in each growth phase based on the knowledge captured by experts. A knowledge matrix is used to define cognitive conflicts between experts by matching captured and extracted expert knowledge to the body of knowledge and crop maintenance process parameters. The crop maintenance knowledge base was developed based on best practices and an analysis of expert knowledge conflicts. Furthermore, crop maintenance knowledge videos were created. These videos discuss the effects of each parameter on crops, as well as the reading and interpretation of data captured by sensor technology for use in sharing knowledge with farmers. Adaptive DSS is a system that analyzes on-farm and off-farm data, as well as the crop maintenance knowledge base, in order to provide farmers with appropriate crop maintenance recommendations analyzing using PyKE (Python Knowledge Engine). The sensors are used to collect data on farms. The open weather station website is used to collect data from off-farm sources. As the management of water resources in crop cultivation is critical due to today's limited availability of natural fresh water, water management also needs to be anticipated. In crop irrigation, soil moisture value is very important to help determine if soil water is sufficient for crop growth or not. The soil moisture value in the near future is also predicted in the Adaptive DSS using a Long-Short Term Memory (LSTM) machine learning approach (Machine Learning, ML) to provide recommendations to farmers for water management on their farm. In terms of information and knowledge representation, it functions similarly to an interface used to control and communicate between the system and farmers via a mobile chat application. Farmers can monitor their farms and control crop maintenance of their own farm crops at any time and from any location. Farmers were overwhelmingly pleased with it, giving it a 96% satisfaction rating based on the implementation results. The findings of this study show that adaptive DSS have the potential to aid decision-making in crop management processes. However, there are still limitations to the system's functionality that can be developed in the future. In future research projects, we will automatically update the system's knowledge base, which will benefit the system by improving the system's accuracy and intelligence in making suggestions when the expert's knowledge changes. In addition, soil moisture forecasting will be developed to assist the model in continuously learning and reducing parameter oscillations. Furthermore, the future application of image processing and deep learning techniques to develop the recommendation system related to mitigating the effects of disease, insect attack, and weed control, including pruning, is anticipated.