Artificial neural network modelling for predicting efficiency and emissions in mini-diesel engines: Key performance indicators and environmental impact analysis

Abstract

• ANN predicts mini-diesel engine performance & emissions (SFC, BTE. CO 2 , NO x , CO, PM). • Input data: engine speed, exhaust temp, torque, pressures, ambient temp. • MATLAB's nftool/nprtool trained & validated the ANN model. • High prediction accuracy (R>0.9, low MSE) achieved. • Potential for 12% efficiency & 40% emissions reduction via optimisation. Mini-diesel engines are widely used in the domestic and commercial sectors, with a global emphasis on improving efficiency and reducing emissions. This study involved the introduction of a novel predictive framework leveraging an advanced Artificial Neural Network (ANN) model, explicitly tailored to mini-diesel engines and addressing key challenges in Internal Combustion Engine (ICE) performance and emissions modelling. The research evaluated critical performance metrics, including specific fuel consumption (SFC), Brake Thermal Efficiency (BTE), and emissions of pollutants such as Carbon dioxide (CO 2 ), Nitrogen oxides (NO x ), Carbon monoxide (CO), and Particulate Matter (PM). The ANN model demonstrated exceptional accuracy, achieving Regression (R) values exceeding 0.9 and Mean Squared Error (MSE) as low as 0.0046 across multiple training-validation-test configurations. This approach integrates both measured and calculated variables, enhancing the robustness and reliability of predictions under diverse operating conditions. Results highlight the ANN’s capability to optimise engine efficiency by up to 12% and reduce emissions by 40%, with significant potential for real-time applications in dynamic engine control. This work bridges a critical gap in ICE predictive modelling. It sets the stage for future integration of ANN frameworks into sustainable engine designs and environmental policies, paving the way for advancements in cleaner energy technologies.