Go to the Prior Tip "Lessons
Learned Through Process Thinking and Review"
Go to the Next Tip PMBOK Guide, Project Management Body of Knowledge
Return to MaxValue Home Page
Modelers of complex systems should read this. Few systems are as complex as a chemical plant. There are thousands of variables ... a system that defies direct assault with formulas from chemistry and physics. A better approach is to model based upon actual plant data.
This is sometimes called "linearization" for simple systems: developing formulas that predict the system's behavior. Regression is key to this approach.
In process control, we are interested in a tool for trouble-shooting, control and optimization. Some of the techniques Van Buskirk describes briefly include:
With a good model, then analysis leading to improvements can begin:
This short article has 9 illustrations (screen shots) and a bibliography. If you're modeling a complex system, this article is worth a look.
More generally, I recommend PC AI (artificial intelligence on PC's) for people interested in leading-edge topics including: expert systems, pattern recognition, natural language processing, data mining, robotics, regression, neural networks, simulated annealing, and genetic algorithms. I've subscribed for over a decade.
John Schuyler, May 2000.
Copyright © 2000 by John R. Schuyler. All rights reserved. Permission to copy with reproduction of this notice.