home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!darwin.sura.net!wupost!psuvax1!psuvm!auvm!BEN.DCIEM.DND.CA!MMT
- Message-ID: <9207212243.AA04578@chroma.dciem.dnd.ca>
- Newsgroups: bit.listserv.csg-l
- Date: Tue, 21 Jul 1992 18:43:07 EDT
- Sender: "Control Systems Group Network (CSGnet)" <CSG-L@UIUCVMD.BITNET>
- From: mmt@BEN.DCIEM.DND.CA
- Subject: Re: No bomb in hierarchy
- Lines: 26
-
- [Martin Taylor 920721 18:30]
- (Rick Marken 920720.1130)
-
- >Actually, what I found in the simulation is that two control systems,
- >one with the wrong connection to one envrionmental variable, DO blow
- >up. I took this to mean that there was no way to "hide" a "positive
- >feedback system" in the hierarchy, as Martin suggested.
-
- Actually, I think they needn't blow up. I don't know whether my message in
- response to your simulation result ever got distributed, but your result is
- "intuitively" correct. If Z is the ECS with the bad connection, and Y affects
- some variable alse affected by Z, then the fact of Y controlling will tend
- to stabilize or "stiffen" that variable. Z will be acting more through any
- overlapping variables, increasing the cotnribution by the one with the
- positive feedback, and reducing Z's effective negative gain. If Y stiffens
- enough the overlapping variables, then Z will go into positive feedback.
- Whether this happens depends both on the gain of Y and the degree of overlap
- between Y and Z. Of course, if Y overlaps on the variable with the positive
- feedback, it will help Z to stabilize.
-
- I think that there are parts of this situation that require a dynamic analysis
- rather than the DC analysis Bill produced. And yes, Bill, the expressions
- get horrendous. After getting Rick's result, I went through the algenbra
- to get the result discussed above.
-
- Martin
-