Title: Cellular implementation of the great salmon run algorithm for designing a black-box identifier applied to engine coldstart modelling

Authors: Ahmad Mozaffari; Nasser L. Azad

Addresses: Systems Design Engineering Department, University of Waterloo, Waterloo, ON, Canada ' Systems Design Engineering Department, University of Waterloo, Waterloo, ON, Canada

Abstract: In this investigation, a cellular version of a recent spot-lighted metaheuristic called The Great Salmon Run (TGSR) algorithm is developed for evolving the architecture of Artificial Neural Network (ANN). The main motivation behind the current research is to find out whether the proposed metaheuristic algorithm is able to cope with difficulties associated with designing an accurate and robust neural black-box identifier. To attest the applicability of the proposed method, the resulted strategy is applied to a real-life challenging identification problem, i.e. identifying the exhaust gas temperature (T_exh) and engine-out hydrocarbon emission (HC_raw) during the coldstart operation of an automotive engine. Generally, the coldstart operation is regarded as a highly non-linear, uncertain and transient phenomenon which in turn can be a very good problem for verifying the authenticity of the proposed hybrid identification strategy. Through the conducted experiments, it is proved that the proposed identification strategy can be used to identify the main operating parameters of coldstart phenomenon.

Keywords: cellular computing; the great salmon run; TSGR; neural black-box identifier design; vehicle cold start; automotive engines; metaheuristics; artificial neural networks; ANNs; exhaust gas temperature; hydrocarbon emissions; hybrid identification; vehicle emissions.

DOI: 10.1504/IJCAT.2016.077799

International Journal of Computer Applications in Technology, 2016 Vol.54 No.1, pp.23 - 41

Published online: 15 Jul 2016 *

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