Prediction of emission performance in a diesel engine fuelled with bio-diesel based on double-hidden layer BP neural network

摘要

Based on BP neural network relevant theories, using the fuel consumption, the load and the diesel blended rate as input parameters and measured CO, HC, NO_x and soot emission data from bench tests of 180FA diesel engine under various operating conditions as training samples, a double-hidden layer BP neural network model for emission performance in a diesel engine fuelled with bio-diesel was established. The results show that the prediction results of CO, HC, NO_x and soot emissions have a good agreement with their experimental ones, and correlation coefficients (R) are very high. It is further shown that the predicted values of HC and CO emissions increase as fuel consumption rate increase, and the predicted values of NO_x and soot emissions decrease with the increase of fuel consumption rate.