Page 372 - Understanding Automotive Electronics

2735 | CH 10 Page 359 Tuesday, March 10, 1998 1:27 PM

DIAGNOSTICS 10

Figure 10.16
Starting Database of
Known Facts
FPO

The procedure is repeated with the resulting conflict set R6. After
executing R6, the system returns to step 1, and finding no applicable rules, it
stops. The final fact set is shown in Figure 10.17. Note that this diagnostic
procedure has found two potential diagnoses: a break in fuel line (confidence
level 0.65), and mixture too rich (confidence level 0.70).
The previous example is intended merely to illustrate the application of
artificial intelligence to automotive diagnosis and repair.
To perform diagnosis on a specific car using an expert system, the
mechanic identifies all of the relevant features to the mechanic’s terminal
including, of course, the engine type. After connecting the data link from the
electronic control system to the terminal, the diagnosis can begin. The terminal
can ask the mechanic to perform specific tasks that are required to complete the
diagnosis, including, for example, starting or stopping the engine.
The mechanic uses the The expert system is an interactive program and, as such, has many
expert system interac- interesting features. For example, when the expert system requests that the
tively in diagnosing mechanic perform some specific task, the mechanic can ask the expert system
problems. why he or she should do this, or why the system asked the question. The expert
system then explains the motivation for the task, much the way a human expert
would do if he or she were guiding the mechanic. An expert system is
frequently formulated on rules of thumb that have been acquired through years
of experience by human experts. It often benefits the mechanic in his or her

Figure 10.17
Final Resulting
Database of Known
Facts

FPO

UNDERSTANDING AUTOMOTIVE ELECTRONICS 359