Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: yaw stabilisation



thanks for the definition guys. i had to use an LVDT to detect the change.
initially i had problems getting my controller to work correctly as i had to
estimate component values. eventually i fine tuned it and it worked
reasonably well to some extent.




----- Original Message -----
From: "Brian Whatcott" <inet@INTELLISYS.NET>
To: <PHYS-L@lists.nau.edu>
Sent: Saturday, April 20, 2002 6:19 PM
Subject: Re: yaw stabilisation


At 08:58 AM 4/20/02, you wrote:
i have designed and built a yaw stabilisation system for an airplane for
my
project at uni. i used a pid controller to achieve this. i am writing my
report for it but i cannot define a pid controller. how would i define
it.
it reads in a signal then what?

PID means Proportional, Integral, Derivative.
A controller which compares a desired output with the present output
value,
and makes corrections based on a function of the error term itself, and
its
integral and 'rate of change' forms, is a classical way to design a
controller
which can be set up to deal with a variety of different environments which
have a strong effect on the usable bandwidth of the loop.

Unfortunately, the appropriate settings for each of these three terms is
by
no means intuitively obvious, so that controllers of this kind tend to
work
sub optimally.
As a thumb-nail description, one could say a PID controller generates a
correcting drive mostly responsive to the size of the [proportional]
difference between a desired output setting, and a present output.

Particularly for small and slow changing departures from the desired
state,
this error is also accumulated over time to add in an increasing
(integral) correction.

At the other end of the control bandwidth - a fast changing error,
which otherwise might well lead to an overswinging output, is
opposed by a [differential] signal proportional to the rate of change
of the error, and set up to oppose the correction, which CAN have
this happy result: a recovery from an upset or
transient change in minimal time with minimal overshoot.

It is found that using a 'fuzzy logic' approach, which develops control
rules
from considering common sense statements of measurable control
objectives and the spans of output levels in which they should apply
and weighting them together by straight-forward centroid of area of
control
objectives often leads to superior performance in practice, where
tuning for concrete results allows much more intuitive modifications.





Brian Whatcott
Altus OK Eureka!