Kirankumar, K.R.Kumar, G.N.Kamath, N.Gangadharan, K.V.2026-02-032024Energy, 2024, 306, , pp. -3605442https://doi.org/10.1016/j.energy.2024.132521https://idr.nitk.ac.in/handle/123456789/20873In the present study, performance, emission, and vibro-acoustic studies were conducted on a spark ignition (SI) engine fueled with gasoline and an n-propanol blend at variable compression ratio (CR), speed, and propanol blend fraction (PBF). Experimental data were used to model an artificial neural network (ANN) trained with a genetic algorithm (GA). ANN predictive responses were employed to establish regression relationships between brake power (BP), brake specific fuel consumption (BSFC), brake thermal efficiency (BTE), oxides of nitrogen (NO<inf>x</inf>), carbon monoxide (CO), hydrocarbon (HC), resultant vibration acceleration (RVA), and sound pressure level (SPL) with operating parameters using response surface methodology (RSM). These models served as objective functions in the non-dominated sorting genetic algorithm-3 (NSGA3), a multi-objective optimization (MOO) technique, to optimize responses and obtain non-dominated solutions. These solutions were filtered using a modified technique for order preference by similarity to the ideal solution (TOPSIS) to obtain a compromised optimal solution. ANN-GA model outcomes showed high accuracy, with coefficient of determination (R2) and root mean square error (RMSE) values ranging from 0.979 to 0.993 and 0.0381 to 0.0643, respectively. NSGA3 coupled with modified TOPSIS identified optimal operating conditions at 1271.77 RPM, a CR of 11.96, and a PBF of 33.26 %. © 2024 Elsevier LtdAccelerationBrakesCarbon monoxideEnginesGasolineMean square errorMultiobjective optimizationNeural networksSurface propertiesThermal efficiencyAlgorithm-3Ideal solutionsModified technique for order preference by similarity to the ideal solutionMulti-objectives optimizationN-propanolNon-dominated sorting genetic algorithm-3Non-dominated sorting genetic algorithmsPerformanceResponse-surface methodologySpark-ignition engineGenetic algorithmsartificial neural networkdiesel engineexperimental studyfuel consumptiongenetic algorithmoptimizationparameter estimationresponse surface methodology