Artificial intelligence based modelling and hybrid optimization of linseed oil biodiesel with graphene nanoparticles to stringent biomedical safety and environmental standards

In this work, experiments were carried out in line with Design of Experiments (DOE) standards to assess the performance and emission features of 5% graphene nanoparticles added linseed biodiesel. The engine was operated with the blends of B10, B20, and B30 with 5% graphene nano additives (designated as B10G5, B20G5, and B30G5). To find the parameter's optimum values, the Desirability Function approach (DFA), Swarm Salp single objective, Multi Objective Bat algorithm (MOBA), Response surface methodology (RSM) and D-optimal design approach were employed. Advanced machine learning (ML) techniques were employed to anticipate these characteristics. It was found that B20G5 had a better brake thermal efficiency (BTE), when compared to the other samples (and around 11% higher than diesel fuel at full load). The emissions of Carbon monoxide (CO) and Hydrocarbon (HC) were lower for B20G5 blended fuel than for diesel (Around 23.52% lower than diesel). In comparison to Response surface methodology (RSM), the overall coefficient of determination (R2) value using Artificial Neural Network (ANN) for was high. As a result, it was revealed that the ANN was typically better than the RSM in forecasting the various factors affecting the engine performance. The optimum outcomes were achieved by single objective (Salp Swarm algorithm) and multi-objective algorithms. According to multi-objective algorithm, a B20G5 nano additive biodiesel mix at its maximum Brake power (BP) produced the highest value of BTE with the lowest Nitrogen Oxides (NOx) emissions. The comparison shows that B20G5 can be used easily without making any modifications to engines.