Exploring the performance and emission characteristics of a dual fuel CI engine using microalgae biodiesel and diesel blend: a machine learning approach using ANN and response surface methodology

Alternative fuels in internal combustion engines have gained significant attention to environmental sustainability and energy security. The study employs a machine-learning (ML) approach, integrating artificial neural networks (ANN) and response surface method (RSM), to analyze the engine characteristics. The experimental data used to train the ANN and RSM model was obtained by employing different combinations of input parameters obtained by the Design of the experiment tool. Four input parameters load 25–100% ((1.3, 2.6, 3.9, and 5.2 kW) loading condition, speed (1200, 1500, and 1800 RPM), compression ratio (17.5 and 18.5), and biodiesel–diesel blends (Diesel, SM20, SM40, SM60, SM80 and SM100) were used. The results show predictability for ANN with training and test regression coefficients (R2) of 0.975 and 0.948 whereas RSM with R2 of 0.992. Optimized results for RSM and ANN, BTE (29.4% and 29.1%), BSFC (0.0.3201 and 0.334 kg/kWh), IMEP (2.83 and 2.69 bar), and CO2 (922.72 and 940.87 g/kwh), NOx (964 and 937 ppm). When compared with experimental data, the error was about 5%. It can be inferred that RSM and ANN may be used to model processes with high predictability and that optimization can be carried out using various techniques depending on the process or problem.