An ANN-entropy-FA model for prediction and optimization of biodiesel-based engine performance

Abstract

The present work incorporated full factorial experimentation for performance analysis of Nahar oil-based biodiesel in a four-stroke diesel engine. The controllable input parameters are engine revolution per minute (N), engine load (EL), and blend mixture (BM) respectively. The measured engine performance characteristics are brake thermal efficiency (BTE), brake-specific fuel consumption (BSFC), and exhaust gas temperature (ET). Prediction and optimization of engine output characteristics have been computed through the ANN-entropy-FA hybrid model. In this model, a trained Artificial Neural Network (ANN) is used to compute the objective function value during optimization using the Firefly algorithm (FA). The maximum absolute percentage error of trained ANN during the prediction of output parameters is found as less than 3%. The entropy method is employed to determine the weight of output parameters in the combined objective function. ANN-FA optimization determined minimum BSFC = 0.386 kg/kW-hr, maximum BTE = 24.60%, and minimum ET = 163.24 °C with corresponding operational input parameter setting of low engine rpm (1250 rpm), medium engine load (15.20 kg), and a high percentage of blend mixture (≈25%) respectively. A significantly low absolute % error of around 1.75% during experimental validation of ANN-FA optimized output indicates the efficacy of the proposed model.