Optimisation of operating parameters of DI-CI engine fueled with second generation Bio-fuel and development of ANN based prediction model

Honne oil methyl ester which is derived from non-edible Honneoil was blended with petroleum diesel fuel and tested on the DI-CI engine. The experiments were conducted at different levels of operating parameters, viz. compression ratio, static injection timing, fuel injection pressure, load and blend. This study aims to determine optimal combination of engine operating parameters With objective of attaining better performance and lower emission. The multi-objective optimisation based on Genetic algorithm is performed which lead to multi pareto optimal solution. The performance parameters were BSEC, BTE and EGT. The emittants were CO2, CO, HC, NOx and smoke. The regression models for performance parameters and emittants as a function of the operating parameters are developed using Minitab. These models were used as a fitness function for optimisation. ANN model based on multi-layer perception was developed to predict the performance and emissions using the experimental data. Four different transfer functions were tried to develop the model and one which yields lowest mean absolute percentage error and highest prediction accuracy is chosen. Preprocessing of data is done by normalizing input data and logarithmic transformation of target data. Levenberg-Marciardt back propagation training algorithm trainlm is used with feed forward multi layer neural network. (C) 2016 Elsevier Ltd. All rights reserved.