Prediction and Verification of Performance and Emission Characteristics of Diesel/Natural Gas Dual-Fuel Engine Based on Intelligent Algorithm

Abstract

With the development of computer application technologies, intelligent algorithm has been widely used in various fields. In this study, a coupled Gaussian process regression and feedback neural network (GPR-FNN) algorithm is proposed, and it is used to predict the performance and emission characteristics of a six-cylinder heavy-duty diesel/natural gas (NG) dual-fuel engine. Using the engine speed, torque, NG substitution rate, diesel injection pressure, and injection timing as inputs, an GPR-FNN model is established to predict the crank angle corresponding to 50% heat release, brake-specific fuel consumption, brake thermal efficiency, and carbon monoxide, carbon dioxide, total unburned hydrocarbon, nitrogen oxides, and soot emissions. Subsequently, its performance is evaluated using experimental results. The results show that the regression correlation coefficients of all output parameters are greater than 0.99, and the mean absolute percentage error is less than 5.9%. In addition, a contour plot is used to compare the experimental results with the GPR-FNN prediction data in detail, and the results show that the prediction model has high accuracy. The results of this study can provide new ideas for the research on diesel/natural gas dual-fuel engines.