Title: Firefly algorithm assisted optimised NN to predict the elongation of API X65 pipeline steel

Authors: Masoud Rakhshkhorshid; Sayyed Hojjat Hashemi

Addresses: Department of Mechanical Engineering, University of Birjand, P.O. Box 97175-376, Birjand, Iran ' Department of Mechanical Engineering, University of Birjand, P.O. Box 97175-376, Birjand, Iran

Abstract: Selection of a proper neural network structure is a complicated work that is often done through the trial-and-error method. In this research, a three layer neural network with feed forward topology and back propagation algorithm is considered as a basic neural network model. The number of neurons in hidden layer, the activation function, the training algorithm and the normalisation procedure are considered as the parameters of the optimisation problem. A proper fitness function is defined and the process of finding the best combination of theses optimisation parameters, based on firefly algorithm, is presented. The proposed procedure is used to find an optimised neural network to predict the elongation of API X65 pipeline steel. The optimised neural network is used to investigate the effects of Ni and microalloying elements on the elongation of the tested steel. The results are in general agreement with the other published works.

Keywords: neural networks; firefly algorithm; API X65; high strength steels; low alloy steels; pipeline steel; steel elongation; nickel; microalloying.

DOI: 10.1504/IJMMNO.2013.056536

International Journal of Mathematical Modelling and Numerical Optimisation, 2013 Vol.4 No.3, pp.238 - 251

Received: 23 Jan 2013
Accepted: 30 May 2013
Published online: 26 Jul 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article