Prediction of Biogas Production from Agriculture Waste Biomass Based on Backpropagation Neural Network
Arief Abdurrakhman 1,2,3*, Lilik Sutiarso 1, Makhmudun Ainuri 1, Mirwan Ushada 1, Md Parvez Islam 3

1 Faculty of Agricultural Technology, Universitas Gadjah Mada, Yogyakarta 55281, Indonesia
2 Department of Instrumentation Engineering, Institut Teknologi Sepuluh Nopember, Surabaya 60111, Indonesia
3 Faculty of Agriculture, Ehime University, Matsuyama, Japan
*Corresponding author : ariefabdurrakhman[at]mail.ugm.ac.id

Abstract

An integral aspect of sustainable agriculture involves the implementation of a meticulously planned waste management infrastructure. One strategy to achieve this objective is the utilization of agricultural waste, specifically in the form of biomass, to generate sustainable energy such as biogas. This study aims to provide valuable prediction model for biogas production with many variables which is influenced. The study identifies four variables, namely pH, moisture content, Organic Loading Rate (OLR) and temperature which significantly impact on the biogas production, especially in Indonesia. Any fluctuations in these variables can affect biogas productivity. Therefore, machine learning techniques such as adaptive backpropagation neural network is used to modeling for predition of biogas production. The configuration of the multilayer perceptron model, combined with the Backpropagation Algorithm, establishes the fundamental framework for the proposed advancements. This study explores three different types of training algorithms in the backpropagation neural network, specifically Adaptive Learning Rate, Levenberg-Marquardt, and Resilient Backpropagation. The Resilient Backpropagation approach exhibited exceptional effectiveness, as evidenced by a correlation coefficient of 0.9411 for training and 0.90423 for testing. The best results obtained for Mean Squared Error (MSE) and Mean Absolute Error (MAE) were 0.0038 and 0.0316, respectively. The Standard Deviation was computed to be 0.0615.This study highlights the potential benefits of employing Resilient Backpropagation Neural Network alghoritm to determine the appropriate operational parameters and accurately predict the biogas production.

Keywords: biogas, neural network, agriculture waste biomass

Topic: Renewable energy and biorefinery