Multi-objective optimization parameter of diesel dual fuel using compressed natural gas at low load

The power density and fuel efficiency of a diesel engine can greatly be improved using a compressed natural gas-diesel dual fuel (CNG-DDF). At low loads, however, the effect of DDF causes a number of issues, including decreased engine performance and an increase in pollutant output (i.e., CO and NOx) during operation. As a result, a combined approach using backpropagation neural networks (BPNN), genetic algorithms (GA), and particle swarm optimization has been developed to predict and optimize the maximum CNG-DDF engine parameter outputs (cylinder pressure, maximum thermal efficiency, and minimum carbon monoxide) (PSO). To begin with, the DDF engine parameter input was carried out in order to generate experimental data by varying the values of the pilot injection time (Tpi), CNG injection timing (TiCNG), supercharger voltage (Vsp), and injection timing (Ti). Next, a three-replication orthogonal array L25 was used to generate seventy-five experimental data points. Then, using training, testing, and validation, BPNN created a fitness function using the experimental data. The best DDF engine performance was then determined by GA and PSO using the fitness function developed by BPNN. The optimization findings indicated that PSO outperformed GA, with the best cost of PSO being 25% more expensive than the GA outcome.