Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
(Serving suggestion.)

idea: add, search, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.

user:
pass:
register,



CoolSolutions

Inventor of Alarm Generation, Notification and Evaluation System (AGNES) for IBM's Storage System Division in San Jose, Ca. filed Feb. 1997, and granted in Nov. 1999.

AGNES is a real-time data mining program with an "Auto Shutdown" feature. All tools, testers, and serial numbered sub-assembly components are traced through the manufacturing process. All the reporting test operations are electronically streamed into AGNES so that real-time running averages of multi-dimensional failure rates are maintained. Running average failure rates (usually over a running 24 hour period) and the number of widgets tested (the counts) for both the overall population as well as the individually contributing members are maintained.

Keeping track of both the failure rates and counts allows for a probability (confidence level, normalcy or "Sigma") assessment. Each time any widget fails, its history is traced (and cross referenced through relational database tables) to see if there is a trend outside normal variability of this particular failure mode. If a trend is detected, a database coincidence event record is logged (and later reviewed by a person). If the event is beyond a preset "Alarm" threshold, then AGNES can automatically issue a notification to the offending tool or tester to discontinue its operation.

This only works when there are multiple (competing and otherwise equal) members of an overall group.

The disabled tool or tester is manually restarted after reading (and hopefully performing) the suggested repair if any exist. The suggested repair can be based on previous experience, but that's another Expert System can of worms.

AGNES tends to self-optimize a given process through Pareto Analysis elimination. In other words, if the "big boulders" inhibiting the manufacturing process are removed, the system tends to become more accurate and sensitive at detecting smaller failure mode contributors, as is the goal of any feedback system.

Trust me that there is a lot more to this, like keeping track of the effects of rework. Accurate failure rate statistical feedback to competing suppliers drives continual improvement. Extremely small variations become statistically significant if the counts are large enough.

One of the unexpected results of AGNES' early data mining was to identify ("discover", if you will) seemingly unrelated cause and effect relationships (the Friday night diapers and beer relationship discovered by grocery stores).

Discovery occurs when an undeniable coincidence begs the question "Why?" and "What makes this unique?" Herein lies the cure for cancer, and the hope for a better tomorrow.

[Jan 05 2008, last modified Jan 08 2008]

   
(+4, -7) 12 Mile Deep Hole
(+1, -6)(+1, -6) False Critique Detector
(+8, -1) Kelvin Injectionado
 

back: main index