Exsys  
shadowTopRight

Efficiently Diagnosing and Repairing Equipment and Processes

Diagnostic systems are one of the most common uses of Exsys Corvid™.  The process human experts use to diagnose a problem are easily converted into Corvid’s rule-based representation, which has special features for handling both probabilistic and deterministic diagnostic approaches.  Corvid generates on-line interactive, web-based systems that provide customers and staff with virtual access to your top experts to resolve problems 24/7.

When equipment breaks it needs to get fixed as quickly as possible.  If it is part of a production line it can disrupt manufacturing.  If it is a product that was sold to a customer, the ability to get it fixed quickly and correctly will strongly influence the customer’s satisfaction and potential future sales. 

Typically a company has a few highly talented experts that can very quickly diagnose the cause of a problem and fix it.  They have the ability to rapidly recognize the various possible causes, and the subtle differences that enable them to know which is the underlying problem.  However, this level of knowledge is scarce since it is based on many years of experience.
 
Diagnostic reasoning is often complex.  An expert knows that a particular combination of symptoms indicate one cause, but slightly different symptoms might indicate a totally different cause.  It is very difficult to teach this type of reasoning to a novice since it is based on so many independent factors that have to be combined to reach a conclusion. However, if you ask experts “How did you know that was the cause?” they can explain it in detail and you will likely get a series of rules - “If you see ... generally it is..., but if you also see ... then that is almost never the cause”. This may fully explain the reasoning in this case, but to a novice there is too much to remember and there are many other possible situations that were not explained.

Corvid makes it practical to capture this type of expert knowledge and deliver it with the Exsys Inference Engine, which is designed for handling probabilistic rules.  Building systems using Corvid’s development tools make it easy to capture complex logic.  Each expert can systematically describe all possible factors that might influence a diagnosis, in an efficient and complete way.  Corvid builds all the files needed to implement knowledge automation systems as on-line interactive systems, which emulate one-on-one conversations with human experts.  

End users are asked questions based on the logic of the expert - drilling down where needed, but without asking irrelevant questions.  The system uses the methodology of the expert to reach specific conclusions and combines the probabilistic rules to provide the most likely cause(s) of the failure.  

Corvid handles diagnostics differently from “case-based reasoning” or search approaches, which attempt to find previous similar situations, in the hope that what worked before will work again.  This can be an appealing concept, but is fundamentally only a guess based on pattern similarity, rather than a reasoned analysis of the problem.  Guesses will sometimes be wrong because while a previous case may be “similar” it may not actually match in some important key factors.  Even when there is a good match, the actions taken previously may not be the correct or best thing to do.  Corvid will provide precise expert advice based on specific readable logic that can be understood, reviewed and maintained. 

Corvid system can also interface to sensors, databases or measuring equipment to provide information needed by the systems to automatically monitor processes and watch for developing problems.

With Exsys Corvid Knowledge Automation diagnostic systems in-house equipment will get fixed faster and more efficiently, even when the human expert is not available.  For your customers, on-line self-service systems can provide diagnosis of complex problems that would have been impossible before.

 

books  

 

"The US Air Force Oil analysis programs reported a total cost avoidance figure in millions of dollars in F-16 engine damages. A conservative estimate for the total cost avoidance generated in the USAF by (knowledge automation system) oil analysis would be over hundreds of millions of dollars." "

Machinery Oil Analysis - Methods, Automation & Benefits
Larry A. Toms

 

exsys.com

 

Download Corvid

 

Exsys

ShadowBottomRight