Toggle Main Menu Toggle Search

Open Access padlockePrints

Predicting molten steel endpoint temperature using a feature-weighted model optimized by mutual learning cuckoo search

Lookup NU author(s): Professor Qiangda Yang, Dr Jie ZhangORCiD



This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND).


A feature-weighted neural network model for the prediction of the endpoint temperature of molten steel (MSET) in a ladle furnace (LF) is proposed in this paper. Accurate prediction of MSET is essential for promoting product quality, reducing production costs and enhancing productivity. Considering that different features have different impacts on the MSET during the process of LF refining, a weight is applied to each feature before feeding the feature to neural networks. A mutual learning cuckoo search (MLCS) algorithm is proposed to simultaneously determine the feature weights and network parameters of the proposed prediction model. The search of each cuckoo in the basic cuckoo search algorithm and many of its variants is performed independently, which may decrease the algorithms’ performance. The proposed MLCS algorithm introduces two new search strategies, the mutual learning-based search strategy and the bottom reinforcement learning-based search strategy. The superior performance of MLCS is first confirmed with 20 benchmark optimization problems. Then, MLCS is applied to optimize the feature weights and network parameters in the feature-weighted MSET prediction model. Application to modeling the production data from a 300 t LF in an iron & steel plant in China demonstrates the effectiveness of the proposed feature-weighted neural network model.

Publication metadata

Author(s): Yang Q, Zhang J, Yi Z

Publication type: Article

Publication status: Published

Journal: Applied Soft Computing

Year: 2019

Volume: 83

Print publication date: 01/10/2019

Online publication date: 31/07/2019

Acceptance date: 28/06/2019

Date deposited: 05/08/2019

ISSN (print): 1568-4946

ISSN (electronic): 1872-9681

Publisher: Elsevier


DOI: 10.1016/j.asoc.2019.105675


Altmetrics provided by Altmetric


Funder referenceFunder name