home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.ai.neural-nets:4785 comp.realtime:1536 comp.theory.dynamic-sys:402 comp.theory.self-org-sys:139 alt.control-theory:17
- Path: sparky!uunet!pipex!doc.ic.ac.uk!uknet!comlab.ox.ac.uk!sharkey
- From: sharkey@robots.ox.ac.uk (Paul Sharkey)
- Newsgroups: comp.ai.neural-nets,comp.realtime,comp.theory.dynamic-sys,comp.theory.self-org-sys,alt.control-theory
- Subject: Re: Use of hysteresis in control problems
- Keywords: artificial neural networks,control
- Message-ID: <1993Jan11.101044.7127@oberon.robots.ox.ac.uk>
- Date: 11 Jan 93 10:10:44 GMT
- References: <1993Jan2.191649.874@noose.ecn.purdue.edu> <1993Jan5.200214.4270@news.acns.nwu.edu> <1iilfcINN6q0@news.cerf.net>
- Distribution: inet
- Organization: Robotics Research Group, Engineering Science Dept, Oxford, UK.
- Lines: 20
- Originator: sharkey@oberon.robots
-
- As mentioned in a previous post hysteresis is widely used to prevent
- a bang-bang controller chattering ie. either on full blast or off.
-
- (Simple control stuff follows for those who want to skip)
-
- Hysteresis is also very improtant in simple PID controllers
- (proportional, integral + derivative). The integral part of this type
- of control integrates the error to drive that error to zero. However
- if the error is very large (imAgine a step increase in demand) the
- integral action would saturate and therefore have an adverse effect
- (called integral windup). To prevent this happening, take a guess.
-
- Paul
-
- -------------------------------------------------------------------
- Paul M Sharkey Office: +44 (0)865 273 150
- Reactive Vision Laboratory Lab: +44 (0)865 273 168
- Robotics Research Group Fax: +44 (0)865 273 908
- Dept Engineering Science
- Parks Road, Oxford OX1 3PJ Email: sharkey@uk.ac.oxford.robots
-