Development of Artificial Intelligence for Variable Rate Application Based Oil Palm Fertilization Recommendation System

Abstract

The need for fertilization knowledge that meets good cultivation principles is the background for the need for a platform that can help provide advice on fertilization implementation. The conventional method through leaf and soil analysis to determine the dose of fertilization has many obstacles in its implementation. The availability of experts to provide advice is also not available at all times. One way to transfer knowledge to non-experts is to use Artificial Intelligence (AI) in the form of a dynamic expert system. The purpose of this research is to create a dynamic expert system that can provide advice and apply fertilization quickly, cheaply, accurately, and available at all times. In building this AI system, a knowledge base, working memory, inference engine, and interface are needed. The implementation of dynamic expert system development consists of four stages, namely (1) literature study, (2) laboratory analysis, (3) construction of an inference engine, and (4) interface creation. Based on the literature study that has been carried out, it is known that there are three domains of knowledge of oil palm fertilization, namely the domain of soil, plants and climate. Each of these knowledge domains consists of attributes, sub-attributes, and facts of knowledge which are then arranged in the form of mathematical rules. The relationship between the three knowledge domains is used as the basis for making fertilization rules. The result of this research is an application with Artificial Intelligence to provide information on nutritional needs according to plant age, population, production, land area and location. The application provides recommendations for the type, frequency, amount of fertilizer, and time of application.