Abstract Growing environmental concerns and increasingly stringent emission regulations have intensified the search for cleaner combustion strategies in compression ignition (CI) engines. Hydrogen, owing to its carbon-free combustion characteristics and wide flammability range, presents a promising partial substitute for diesel in dual-fuel operation. However, the complex nonlinear interactions among injection parameters, engine load, and the resulting performance-emission trade-offs make systematic optimization of such systems challenging. This study addresses this gap by developing a hybrid artificial neural network–fuzzy logic framework to predict and optimize the performance and emission characteristics of a hydrogen-enriched diesel CI engine. A four-stroke single-cylinder engine was tested across five load conditions (20–100%) under four hydrogen injection timing strategies (DH1–DH4) alongside a baseline diesel mode (DH0). A multi-layer perceptron ANN trained on experimental measurements was used to capture the nonlinear relationships between operating inputs and engine responses, including brake thermal efficiency (BTE), volumetric efficiency, NOx, unburned hydrocarbons (UHC), and soot emissions. A Mamdani-type fuzzy inference system subsequently converted these five competing response variables into a unified multi-performance characteristic index (MPCI), enabling transparent multi-criteria optimization without subjective weighting of individual objectives. The optimal operating condition was identified as 40% load under the DH2 injection strategy (6500 μs), yielding a BTE of 18.46%, volumetric efficiency of 70.85%, NOx of 0.93 g kWh −1 , UHC of 21.5 g kWh −1 , and soot of 0.05 g kWh −1 . The proposed framework demonstrates strong predictive accuracy across all output variables (R 2 > 0.99) and offers a computationally efficient, reproducible methodology for multi-criteria engine calibration. This study supports United Nations SDG 7, SDG 9, and SDG 13 through hydrogen-assisted clean and energy-efficient CI engine technology.
Bhowmik et al. (Thu,) studied this question.