Sistem Monitoring Tanaman Cerdas Menggunakan Wireless Sensor Network dan Evolutionary Fuzzy Association Rule Mining

Abstract

Plant conditions monitoring requires specific knowledge in agriculture. The knowledge includes decision support to describe between good (ideal) and bad conditions from each different plant. Fuzzy rules capable to describes plants bio signal in form of fuzzy membership function that usable for decision support. To simplify the decision support process , this research proposes the design and development of the smart monitoring system of plant conditions based on wireless sensor networks (WSN) and evolutionary fuzzy association rule mining (EFARM). Plant condition monitoring is carried out through sensors input by WSN and decision support algorithm is carried out by EFARM. The proposed method aims to be carried out on a sensor network with supervised learning from training data. The dataset will only be used to create default fuzzy membership functions and rules. Detailed optimization and classification of conditions will be carried out using the evolutionary process by tree based rule extractor from Genetic Programming (GP). The Evaluation has been carried out using three raspberry pi used as EFARM processor and storage, which separated into one central processor and two partial processor for two plants, Cactus and Orchid. The simulation results show that the proposed method is able to extract rules from both plants and is able to measure significant differences between plants.