Diesel Engine Pollutant Prediction and Remote Vehicle Monitoring Part Ⅰ: The Prediction of Diesel Engine Smoke Emission using Neural Networks

摘要

Accurate measurement of diesel engine exhaust smoke emission is a primary phase in meeting the ever-stricter European Union regulations on emission levels and a fundamental step towards the improvement of many factors including fuel economy, atmospheric pollution levels and more importantly, human health, with the additional aim of automatic engine management systems and condition-based maintenance. However, it is difficult to measure continuously smoke levels directly and in real-time on a vehicle in transit due to the size and cost of the necessary equipment, therefore this paper (Part Ⅰ) documents a study into the feasibility of diesel engine exhaust smoke prediction based upon a variety of engine operating parameters recorded from three different engines using neural network (NN) models. In this paper two types of NN have been investigated and optimised to develop a prediction. The results show that smoke levels can be predicted by means of indirect measurements with good accuracy. Part Ⅱ of this paper describes how the NN model is used with real-time data collected remotely from a vehicle on the road to predict smoke emission levels and introduces a method of mapping these smoke levels on a city map at street level via the Internet.