The ability to infer smoke emission levels from routinely monitored engine operating parameters, without the need for specialist or cumbersome instrumentation, would bring with it the possibility of mapping quantitative particulate emission for an engine in a vehicle operating under normal on-road conditions. This paper (Part II) follows on from its companion paper (Part I) in which a method of predicting vehicle emission characteristics using a neural network (NN) model was described. The NN was trained on various engine and vehicle parameter data and optimised to produce a model to predict diesel engine smoke emission levels with sufficient accuracy. This paper (Part II) describes how the NN model can be used to predict remotely vehicle emission characteristics in real-time using the General Packet Radio Service (GPRS) network with the Internet and mapping the emissions on a city map at street level. The real-time smoke emission prediction methodology is predominantly directed at fleet operators. The paper describes the equipment and process used and demonstrates what impact this can have on overall fleet emissions monitoring, identifying key benefits to the fleet operator. Important spin-off factors resulting from the system are also discussed.
|Number of pages||5|
|Journal||International Journal of COMADEM|
|Publication status||Published - 1 Oct 2004|