University of Tasmania
whole_PolsonCranstonPhillip2003_thesis.pdf (30.85 MB)

Intelligent torque estimation and fault diagnosis for an IC race engine

Download (30.85 MB)
posted on 2023-05-27, 18:01 authored by Polson, CP
There is a remarkable improvement in the engine design and its sensory control over the last three-four decades in automotive industry. Several pressing issues such as reduced Hydro Carbons, improved fuel efficiency and optimum power for efficient engine performance are continually investigated in modem automobile companies. While the technological 'know-how' is reaching optimum in terms of the automobile body design, aerodynamics and comfort aspects, there is sufficient room for development for optimum engine performance. There is evidence that the modem automotive companies are adopting certain modem control strategies and emerging technologies such as fuzzy logic and evolutionary algorithms for better engine performance. The IC (Internal Combustion) engine performance and fault recognition is a major research issue. It is very well know that there can be more than one cause that contributes to the same effect in an IC engine. This aspect puts more pressure on the need for reliable quantitative models for fault diagnosis to identify specific cause of the same effect. On the other hand a reliable control of the air-mass flow and the air-fuel ratio plays a significant role in controlling both power and Hydro Carbon emissions into the atmosphere. Automotive companies are continually attempting to model IC engines for fault diagnosis and performance prediction using traditional modelling techniques such as heat transfer models and empirical investigation using knowledge base. An online estimation of torque and power, in absence of chassis dynamometers, in dynamic conditions is also an important aspect to find out the functional behaviour at any given time for the engine. In this thesis intelligent neural network models are proposed for prediction of torque, power and air-mass flow for a 600cc-race engine. Using extensive experimental investigation, it is shown that, the neural network architectures predict power, torque and the air-mass flow to an excellent accuracy for eventual on-line control over a range of engine operating conditions.


Publication status

  • Unpublished

Rights statement

Copyright 2002 the author - The University is continuing to endeavour to trace the copyright owner(s) and in the meantime this item has been reproduced here in good faith. We would be pleased to hear from the copyright owner(s). No access until 27 September 2007. Thesis (MEngSc)--University of Tasmania, 2003. Includes bibliographical references

Repository Status

  • Open

Usage metrics

    Thesis collection


    No categories selected


    Ref. manager