Fuel consumption analysis in dynamic states of the engine with use of artificial neural network

Abstract

The paper presents the construction of fuel consumption dynamic characteristic determined with use of artificial neural network (ANN). The characteristic is based on the data obtained during measurements carried out on the engine dynamometer. The momentary fuel consumption as a non-linear function of two variable parameters, engine speed and torque, has been presented. Article discusses the way of determining such a characteristic which can be used for both SI and CI engines for entire range of engine speed and load. Described characteristic enables analysis of engine properties in dynamic states and allows computing mileage fuel consumption of the car with a given engine in combination with the specific transmission in any virtual driving cycle. An important advantage of the ANN method in case of general fuel consumption characteristic in dynamic operating states is that standard measurement equipment can be used and the solution to a complex problem is reduced to programming issues.