Adaptive Modelling of Fed-batch Processes with Extreme Learning Machine and Recursive Least Square Technique

Kazeem Alli, Jie Zhang

2020

Abstract

This paper presents a new strategy to integrate extreme learning machine (ELM) with recursive least square (RLS) technique for the adaptive modelling of fed-batch processes that are subject to unknown disturbances. ELM has the advantage of fast training and good generalization. ELM is effective in modelling nonlinear processes but faces problems when the modelled process is time varying due to the presence of unknown disturbances or process condition drift. The RLS can adapt to current process operation by recursively solving the least square problem in the considered model. The RLS estimation algorithm nullifies the model plant mismatches caused by the unknown disturbances through correction of parameter estimates at each iteration. The offline trained output layer weights of an ELM are used as the initial values in parameter estimation in RLS, which are being updated after each batch run using RLS. The proposed strategy is tested on an isothermal semi-batch reactor. The results obtained show that the proposed batch to batch adaptive modelling technique is very effective.

Paper Citation

in Harvard Style

Alli K. and Zhang J. (2020). Adaptive Modelling of Fed-batch Processes with Extreme Learning Machine and Recursive Least Square Technique. In Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART, ISBN 978-989-758-395-7, pages 668-674. DOI: 10.5220/0008980506680674

in Bibtex Style

@conference{icaart20,
author={Kazeem Alli and Jie Zhang},
title={Adaptive Modelling of Fed-batch Processes with Extreme Learning Machine and Recursive Least Square Technique},
booktitle={Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART,},
year={2020},
pages={668-674},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0008980506680674},
isbn={978-989-758-395-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART,
TI - Adaptive Modelling of Fed-batch Processes with Extreme Learning Machine and Recursive Least Square Technique
SN - 978-989-758-395-7
AU - Alli K.
AU - Zhang J.
PY - 2020
SP - 668
EP - 674
DO - 10.5220/0008980506680674