Signal Analysis for Advanced Vehicle Diagnostics

 

The University of Michigan Dearborn

A Joint Research Project with

Diagnostic R&D
Diagnostic Service Planning
Ford Motor Company

Ford Motor Company

 

This project attempts to develop and implement an advanced intelligent system that is capable of analyzing diagnostic signals to predict fault or no-fault signals.  The goal of the project the progression towards the delivery of a viable and realistic diagnostic tool(s).   Ideally this tool(s) will be delivered to the dealership technicians via the next dealer diagnostic tool WDS. 

Two major functions are involved in this system:

1.  

Recognition of Individual Signals: This function is to accurately identify good signals under a variety of actual vehicle dynamics, flagging those that are suspect.  If a signal is not good then we must inspect why it is suspect.  The goal is also to be able to identify automatically those "events" in signal recordings that lead to recognizing them as non-good (i.e. identifying them as suspect).

2.  

Multiple Signal Interactions: This function represents the interactions of signals, and recognizing the cause, effects, and relationships of signals on other signals. The goal of this function is to demonstrate the capability of recognizing whether events are symptoms or root causes, through applying formal diagnostic knowledge rules.

 

The research focus of the project includes signal analysis, automotive engineering knowledge, wavelet transform, fuzzy learning and neural networks. 

 

  ADSAS (Advanced Diagnostic Signal Analysis System)

Adsas.gif (16148 bytes)

View the Tech Day 2000 Presentation

 

 

UM-D Research Team:
Professor Yi L. Murphey
Dr. Hong Guo
Jacob Crossman
Ford Research Team:
Shane Rachedi
Robert Mills
Mark Coleman