EXPERT SYSTEM
TOPIC 6
ILLUSTRATIVE APPLICATIONS
Table of Content
An Intelligent Help Desk
Gates Assignment Display System (GADS)
Energy Management in Hydroelectric Plant
Intelligent Mould Costing Estimator
An Intelligent Help Desk
Problem
Help desk have been employed by private and
public organizations to provide assistance in person
or via the telephone to customers, employees or
other interested parties
Managing is becoming difficult due to the following
trends:
The volume of calls to help desks is increasing
Complexity of the required information is increasing
The quality of the personnel who provide the help is not
improving
The cost is increasing
An Intelligent Help Desk
Solution
Develop an ES that will “assist the assistors”.
An example is the help desk at Harris Corp.
headquarters (Melbourne, Fla.)
Provides a hot line for employees to troubleshoot
problems in computer hardware and networks
The system is composed of 6,000 pieces of
hardware
An Intelligent Help Desk
How the system works:
Request are made by phone
Help desk operator call an ES for help
The employee keyed in the information (e.g. he type
of equipment)
The ES ask a few questions and provide brief
procedures to help diagnose the problem
The ES provides a diagnosis and instructions on
what to do next
The consultant takes about 10 minutes
An Intelligent Help Desk
Development
The system (rule based and PC based) used
Insight2+ shell (now called level 5)
Interacts directly with the Virtual Telecommunication
Access Method network to get information about the
specific device experiencing the problem and the
current status of the device
The system includes 200 rules
An Intelligent Help Desk
Benefits
Expedites the work of the help desk operators
Allows operators to diagnose problems that
previously were directed to higher level of experts
(system programmers)
Problems that took 4 hrs of system programmers time is
now being solved in 10 minutes
The number of difficult cases transfer to system
programmers has been reduced by 80%
Also being used for training help desk workers
An Intelligent Help Desk
Future Development
Use stored data from past consultations to
personalize the current session
Use data from past consultations to induce new
rules and thus facilitate system learning over time
Use the historical data to create advisory report for
management.
Intelligent help desk provide one of the most
cost-efficient means of knowledge distribution.
Gates Assignment Display System
(GADS)
Problem
Gate assignment, the responsibility of gate
controllers is a most complex and demanding tasks.
At O’Hare airport in Chicago, for example, two gate
controllers typically plan berthing for about 400
flights a day at some fifty gates
Flights arrive in cluster of 30-40 planes in 20-30
minutes
Each flight is scheduled to maintain at its gate a
different length of time, depending on schedules
and the amount of servicing needed
Needs to juggles gates constantly due to flight
delays
Gates Assignment Display System
(GADS)
Solution
In Nov 1986, United Airlines/TI team went to work
on the problem
By end of June 1987, they had the first operational
ES in daily use at Denver airport.
The second system was implemented at O’Hare in
August 1987
The system is integrated with the United Operation
Information System
Gates Assignment Display System
(GADS)
System Capabilities
Faster to setup and rescheduled
Carries more information
Superb graphic display shows clock/time and gate numbers
Aircraft symbolized as colored bars, each bar position indicates
the gate assigned and its length indicates the length of time the
plane is expected to occupy the gate
Ground status presents in a split second by using small radio
transmitter that automatically reports to the mainframe
computer.,
Has a number of built-in reminder systems
Store historical and future views. Historical logs are kept
automatically for analysis that can improve future scheduling
Gates Assignment Display System
(GADS)
Benefits of the system
The assistant gate controllers can start scheduling
the next day operations 4-5 hours earlier
The ES is also used by zone controllers.
Super reliability where 2 Explorer computers (AI
work station from TI) for gate controllers and 5 for
zone controllers
Any seven explorers can take over the job when any of
the others has problem
If the mainframe system, shut down, all explorers shift
automatically to stand-alone mode, which allows
operation to continue.
Energy Management in Hydroelectric
Plant
Problem
A large hydroelectric plant in Spain serves more than 7
million customers
Electric provided via a large, centrally controlled
network
Decision are based on information collected from 2,500
data points
Whenever a major problem occurs in the system, more
than 100 alarms are activated in less than a minute.
The problem is to properly analyze and quickly interpret
the information collected by the alarms.
The high volume of messages and the potential
uncertainty caused by the thousands possible
combinations of signals make the situation ideal for ES
Energy Management in Hydroelectric
Plant
Solution
The utility needed a real time ES integrated with
signaling system as well as existing database
Links were done via communication lines
Developed on a TI workstation (Explorer)
Began as an initial prototype with 400 rules, 1500
objects and 200 Lisp functions
Fully implemented after 2 years
Energy Management in Hydroelectric
Plant
System capabilities
Detecting a disturbance when it happens
Locating the element of fault
Tracking the location of the disturbance
Recommending repairs and justifying them
In online mode, the ES
assesses alarms,
generates several hypothesis about fault location, select the most
likely, and send its conclusion to human operator
In offline mode, in its off-line mode, after disturbance has
been resolved and all possible information about it has
been amassed, the ES reasons about it.
The system explains the chain of reasoning it followed in selecting
its hypothesis
Intelligent Mould Costing Estimator
-MOLDCost is an intelligent mould
calculation that suggests the optimum
2-plate type of mould cost.
-To solve scarcity of human mould
costing estimator. The result is
comparable with the human calculated
cost.
-Implemented at Foundry &
Tooling Technology Centre, SIRIM
Berhad.
MOLDCost Process Flow
-Developed using Visual Basic & MS Access and runs
under Windows platform.
-Utilizing Expert Systems technology.
Customer
•Designing
•Purchasing Sales
•Machining
Mould Cost
•Mould Assembly Estimation
•Mould Testing
•Accounting
Authority Approval
MOLDCost Ver. 1.0 Screenshots
-
Back to Main Courtesy: AI Lab, SIRIM Berhad