Work

Multi-objective optimization of marine diesel engine performance using Neural Network, Genetic Algorithm and multi-criteria decision making

Year: 2026
Type: article
Abstract: Optimizing marine diesel engine performance requires navigating complex trade-offs between fuel efficiency and emission constraints under evolving regulatory frameworks such as the International Maritime Organization Tier III. This study presents a computationally efficient multi-objective optimization framework for two stroke marine diesel engines, integrating an Artificial Neural Network (ANN), Non-dominated Sorting Genetic Algorithm II (NSGA-II) and Multi-Criteria Decision-Making (MCDM). The NN was trained on a simulation dataset generated via Diesel-RK software, capturing the effects of six control parameters (Start of Injection, Fuel Mass Fraction, Injection Delay, Common-rail Pressure and Exhaust Valve Opening/Closing) on three key objectives: Specific Fuel Consumption, Nitrogen Oxides and Particulate Matter. The ANN surrogate model replaced Diesel-RK in NSGA-II to generate a Pareto front. Three MCDM algorithms, Proximity Indexed Value, Weighted Integrated Sum Product and TriMetric Fusion were then applied to rank Pareto-optimal solutions. The methods achieved a strong consensus on optimal configurations of engine control parameters. In addition, solution re-evaluation with Diesel-RK showed close numerical agreement with the surrogate predictions. Beyond the methodological development, this study presents a numerical example implementation of the proposed framework through a newly developed Decision Support System (DSS) guiding operators from surrogate-based objective prediction and NSGA-II Pareto generation to MCDM ranking and final operational recommendation. The software demonstration bridges the gap between computational optimization and actionable decision-making.
Language: English
Open Access status: gold
Estimated APC paid: $3,040
OpenAlex ID: W7236400919
FWCI:
Cites: 39
Cited by:
Related to: 10