Title: Neural network fault classification of transient data in an automotive engine air path

Authors: Mahavir S. Sangha, J. Barry Gomm, Dingli Yu

Addresses: School of Engineering, Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK. ' School of Engineering, Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK. ' School of Engineering, Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK

Abstract: Classification of automotive engine air path faults from transient data is investigated using Neural Networks (NNs). A generic Spark Ignition (SI) Mean Value Engine Model (MVEM) is used for experimentation. Several faults are considered, including sensor faults, Exhaust Gas Recycle (EGR) valve and leakage in intake manifold. Consideration of different fault intensities for all the sensor and component faults is a unique feature of this research. The identification of a fault and its intensity has been considered both equally important. Radial Basis Function (RBF) NNs are trained to detect and diagnose the faults, and also to indicate fault size, by recognising the different fault patterns occurring in the transient data. Three dynamic cases of fault occurrence are considered with increasing generality of engine operation: (1) engine accelerates or retards from mean speed, (2) engine runs at different steady speeds and (3) engine accelerates or retards from any initial speed. The approach successfully classifies the faults in each case.

Keywords: fault diagnosis; fault isolation; radial basis function NNs; RBF; artificial intelligence; neural networks; ANNs; fault classification; transient data; automotive engines; air paths; spark ignition MVEM; mean value engine models; fault patterns.

DOI: 10.1504/IJMIC.2008.019352

International Journal of Modelling, Identification and Control, 2008 Vol.3 No.2, pp.148 - 155

Available online: 08 Jul 2008 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article