Experimental Validation of Artificial Neural Network (ANN) Model for Scramjet Inlet Monitoring and Control
Azam Che Idris1, Mohd Rashdan Saad2, Mohd Rosdzimin Abdul Rahman3, Fakroul Ridzuan Hashim4, Konstantinos Kontis5

1Azam Che Idris, Faculty of Engineering, UPNM, Kuala Lumpur, Malaysia.
2Mohd Rashdan Saad, Faculty of Engineering, UPNM, Kuala Lumpur, Malaysia.
3Mohd Rosdzimin Abd Rahman, Faculty of Engineering, UPNM, Kuala Lumpur, Malaysia.
4Fakroul Ridzuan Hashim, Faculty of Engineering, UPNM, Kuala Lumpur, Malaysia.
5Konstantinos Kontis, Mechan Chair of Engineering, University of Glasgow, United Kingdom.
Manuscript received on 15 February 2019 | Revised Manuscript received on 06 March 2019 | Manuscript Published on 08 June 2019 | PP: 558-563 | Volume-7 Issue-5S4, February 2019 | Retrieval Number: E11180275S419/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: A hypersonic flight vehicle, viewed as an engineering system, must have a real-time monitoring and control of its performance, in order for it to be safe and practical for operation. The scramjet engine is the most suitable for hypersonic flow regime and its performance depends mostly on its inlet. There are multiple strategies to measure the performance of a scramjet inlet but they are limited to on-ground operations only. A number of empirical relations exist to easily calculate the scramjet inlet performance using only its internal throat Mach number but they are somewhat hit-and-miss. Using Artificial Neural Network (ANN) algorithm and data from the literature, we investigated the optimum ANN structures that can be used to model scramjet inlet performance. The optimum ANN model is then tested and validated against our own experimental measurement of our generic scramjet inlet. The optimum ANN model with 10-nodes in a single hidden layer was able to match perfectly with our experimental data.
Keywords: Artificial Neural Network, Hypersonic, SBLI, Scramjet, Shockwave.
Scope of the Article: Neural Information Processing