POHAN, Sry Dhina and Suryono, Suryono and Warsito, Budi (2020) PREDIKSI PERTUMBUHAN BIBIT TANAMAN PADA GREENHOUSE MENGGUNAKAN NEURAL NETWORK BACKPROPAGATION. Masters thesis, School of Postgraduate Studies.

	Text COVER.pdf Download (691kB)
	Text BAB I.pdf Download (427kB)
	Text BAB II.pdf Download (704kB)
	Text BAB III.pdf Restricted to Repository staff only Download (657kB)
	Text BAB IV.pdf Restricted to Repository staff only Download (810kB)
	Text BAB V.pdf Download (395kB)
	Text DAFTAR PUSTAKA.pdf Download (411kB)

Abstract

Prediksi pertumbuhan tanaman merupakan salah satu masalah penting dunia dalam rangka memenuhi ketersediaan pangan bagi penduduk. Greenhouse merupakan teknologi yang mendukung pertumbuhan tanaman. Namun, Model prediksi sebelumnya dilakukan secara manual yang menghasilkan akurasi rendah dan error yang besar. Penelitian ini mengusulkan metode neural network backpropagation berdasarkan data time series untuk memprediksi pertumbuhan bibit tanaman kangkung pada area greenhouse. Prediksi bibit dilakukan dengan membangun program komputer berbasis GUI (Graphical User Interface) menggunakan neural network backpropagation dengan arsitektur input layer, hidden layer dan output layer. Data bibit tanaman kangkung yang dikumpulkan dari hasil akusisi data secara online dari sensor yang dijadikan sebagai input yaitu suhu, kelembapan tanah, kelembapan udara, dan intensitas cahaya. Data bibit tanaman kangkung dipantau dari kamera untuk melihat selisih pertumbuhan panjang batang tanaman yang dijadikan sebagi output prediksi. Proses pembelajaran dan prediksi mendapatkan hasil yang baik berdasarkan toleransi error sebesar 0,001. Data pertumbuhan bibit tanaman kangkung yang diproses menghasilkan waktu komputasi 3,009 detik dan Mean Squared Error (MSE) prediksi sebesar 0,001 dengan waktu komputasi 0,245 detik sehingga prediksi yang dihasilkan mendekati aktual.
Kata kunci : Prediksi, Greenhouse, Neural Network, Backpropagation, Time Series, Pertumbuhan bibit.

Prediction of plant growth is one of the world's most important problems in fulfilling the food availability for the population. Greenhouse is a technology that supports plant growth. However, the prediction model was done manually which resulted in low accuracy and large errors. This research proposes a backpropagation neural network method based on time series data to predict the growth of kale seedlings in the greenhouse area. Seed prediction is done by building a computer program based on a GUI (Graphical User Interface) using a backpropagation neural network with an input layer, hidden layer and output layer architecture. Water spinach seed data collected from the results of online data acquisition from sensors that are used as input, namely temperature, soil humidity, air humidity, and light intensity. Data on kale seedlings are monitored from the camera to see the difference in plant stem length growth which is used as prediction output. The learning process and prediction get good results based on an error tolerance of 0.001. The growth data of kale seedlings that were processed resulted in a computation time of 3.009 seconds and a predictive Mean Squared Error (MSE) of 0.001 with a computation time of 0.245 seconds so that the resulting predictions were close to actual.
Keywords : Prediction, Greenhouse, Neural Network, Backpropagation, Time Series, Seeds Growth.

Actions (login required)

View Item