Applying multivariate linear regression and multi-layer perceptron artificial neural network to design an energy consumption baseline in a low density polyethylene plant

Behnam Hamedi (Department of Industrial Engineering, Payame Noor University – Assaluyeh, Iran)
Alireza Mokhtar (Business School, Massey University, Auckland, New Zealand)

International Journal of Energy Sector Management

ISSN: 1750-6220

Publication date: 25 July 2019

Abstract

Purpose

The purpose of this study is to investigate and analysis of energy consumption for this industry. The core part of any energy management system (EnMS) in industry is to perfectly monitor the energy consumption of significant users and to continuously improve the energy performance. In petrochemical plants, production deals with energy-intensive processes, and measuring energy performance for recognition and assessment of potentials for saving is critical.

Design/methodology/approach

The required data are exploited for the period of March 2011-August 2016 (data set: 2,012 days). Multivariate linear regression (MLR) and multi-layer perceptron artificial neural network (ANN) methods are separately used to anticipate the energy consumption. The baseline will be assumed as a reference to be compared with the actual data to estimate the real saving values. Finally, cumulative summations (CUSUM) are proposed and applied as an effective indicator for measurement of energy performance in an LDPE.

Findings

In this study, two statistical methods of MLR and ANN were used to design and develop a comprehensive energy baseline representing the predicted amounts of energy consumption based on the recognized drivers. Although both models imply robust outcomes, when the relative errors are taken into account, performance of ANN models appears fairly superior compared to the MLR model.

Originality/value

It is highly suggested to the ISO technical committee dealing with energy management standards, to consider the proposed model for baseline development in the future version of the standard ISO 50006 as the supplementary extension for the ISO 50001 for measuring energy performance using EnB and EnPI. As for future studies, the research can be extended to investigate the uncertainty and the model could also become completed applying more advanced ANNs such as recurrent neural networks.

Keywords

Citation

Hamedi, B. and Mokhtar, A. (2019), "Applying multivariate linear regression and multi-layer perceptron artificial neural network to design an energy consumption baseline in a low density polyethylene plant", International Journal of Energy Sector Management, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/IJESM-01-2018-0012

Download as .RIS

Publisher

:

Emerald Publishing Limited

Copyright © 2019, Emerald Publishing Limited

Please note you might not have access to this content

You may be able to access this content by login via Shibboleth, Open Athens or with your Emerald account.
If you would like to contact us about accessing this content, click the button and fill out the form.