Neural network model for profile temperature of nickel kiln

Leo Agung Arie Purnomo, Totok Ruki Biyanto

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


Reduction Kiln plays a very important role in the nickel smelting process. As multi-variable processing and nonlinear system, nickel kiln is difficult to be maintain in the best performance. While research on this area is very rare for this industry is such specific and unique. In this paper, it is proposed time-series neural network model for nickel kiln, with an approach based on Nonlinear Auto Regressive with eXogenous input (NARX). This model provides prediction of input-output data set to achieve certain operating condition, especially profile temperature alongside the kiln. Input-output data set of real plant was collected from Kiln 5 in PT Vale Indonesia for along 2018. The data set was reviewed and chosen for the neural network model. It was achieved that regression of the model is 0.999 which is proving that there is strong relationship between input and output. While mean squared error (MSE) is 16.4077, which means that the model is able to capture the characteristic of the plant and temperature of 6 points alongside the kiln, yet open for future research.

Original languageEnglish
Title of host publicationAdvanced Industrial Technology in Engineering Physics
EditorsAgus Muhamad Hatta, Katherin Indriawati, Gunawan Nugroho, Totok Ruki Biyanto, Dhany Arifianto, Doty Dewi Risanti, Sonny Irawan
PublisherAmerican Institute of Physics Inc.
ISBN (Electronic)9780735418189
Publication statusPublished - 29 Mar 2019
Event2nd Engineering Physics International Conference 2018, EPIC 2018 - Surabaya, Indonesia
Duration: 31 Oct 20182 Nov 2018

Publication series

NameAIP Conference Proceedings
ISSN (Print)0094-243X
ISSN (Electronic)1551-7616


Conference2nd Engineering Physics International Conference 2018, EPIC 2018


Dive into the research topics of 'Neural network model for profile temperature of nickel kiln'. Together they form a unique fingerprint.

Cite this