Prediksi Masa Simpan Tandan Buah Segar Kelapa Sawit Berbasis Jaringan Saraf Tiruan

Abstract

Indonesia, as one of the world's main producers of palm oil, has a Crude Palm Oil (CPO) export performance index of 0.94. One of the largest palm oil producing regions in Indonesia is North Sumatra Province with CPO production of 5.3 million tons in 2021 which has a growth rate of 2.8% to 5.45 million tons in 2022. One important factor that must be considered before managing oil palm plants is the process of storing oil palm. The oil palm storage process is the process when oil palm fruit is collected before being processed or distributed. Oil Palm that has matured and fallen from the tree will usually rot in about 1 week, but there is no measuring instrument that can accurately determine when the fruit becomes rotten, making it difficult to estimate the shelf life which causes the fruit to rot during the storage and distribution process. This research creates an innovation to predict the shelf life of palm fruit by utilizing a Multi-Layer Perceptron artificial neural network. The shelf life prediction is calculated based on the results of electrical, gas and color resistivity sensor measurements on oil palm processed using an Arduino nano. The application of artificial neural networks in predict the shelf life of oil palm fruit achieves an accuracy rate of 98,62% with an MSE value of 0.006 so that the system can predict when palm fruit will rot quite accurately.