Excitation Signal Design and Modeling Benchmark of NOx Emissions of a Diesel Engine | IEEE Conference Publication | IEEE Xplore

Excitation Signal Design and Modeling Benchmark of NOx Emissions of a Diesel Engine


Abstract:

This paper focuses on two aspects of the nitrogen oxides (NOx) emissions modeling process of a Diesel engine. Firstly, a novel design of experiments (DoE) strategy with a...Show More

Abstract:

This paper focuses on two aspects of the nitrogen oxides (NOx) emissions modeling process of a Diesel engine. Firstly, a novel design of experiments (DoE) strategy with an amplitude modulated pseudo random binary signal (APRBS) within a non-convex hull is introduced. The non-convex hull is identified by a one-class regularized radial basis function (RBF) classifier. Secondly, an appropriate model architecture has to be chosen. Therefore, a modeling benchmark of six data-driven models for the NOx emissions is carried out. The investigated model architectures can be categorized in models with feedback and models without feedback. The investigated models with feedback are the long short-term memory (LSTM) model, gated recurrent unit (GRU) model, and the local model state space network (LMSSN). The models without feedback, on the other hand, include the multilayer perceptron with nonlinear finite impulse response (MLP-NFIR) model, the temporal convolutional network (TCN), and the regularized local FIR model with local model networks (NRFIR). The modeling benchmark has been carried out in such a way that it accounts for later applicability on an electronic control unit (ECU). The benchmark compares the models for stationary and dynamic conditions.
Date of Conference: 23-25 August 2022
Date Added to IEEE Xplore: 08 December 2022
ISBN Information:

ISSN Information:

Conference Location: Trieste, Italy

References

References is not available for this document.