TONY NIXON 54964
I've got a rather tricky problem controlling a winch with a V6
automatic petrol driven engine. Due to the mass of the winch there is a fairly
large time lag from a throttle movement to a reaction from the winch
drum. It would be comparable to a car towing a caravan.
Has anyone got ideas on where I can look to get ideas / information for
this type of control.
Just when I thought I knew it all,
I learned that I didn't.
> I've got a rather tricky problem controlling a winch with a V6
> automatic petrol driven engine. Due to the mass of the winch there is a fairly
> large time lag from a throttle movement to a reaction from the winch
> drum. It would be comparable to a car towing a caravan.
> Has anyone got ideas on where I can look to get ideas / information for
> this type of control.
One common solution to this type of problem is what's called a "PID loop".
In a PID loop whose goal is to stabilize a position, the output (throttle
setting) is the sum of three terms (each of which has a seperate scaling
Proportional term: the distance between the object and it's target
Integral term: the above distance, integrated over time
Differential term: the velocity of the object
The proportional term causes the object to move toward its target; the
further the object is from the target, the greater the thrust toward the
target. The integral term serves to increase the thrust if the object is
not getting to its target; without it the object may reach an equilibrium
short of the target (e.g. if the object is traveling vertically, it will
settle at the point where its distance from the target causes the motor's
force to equal the object's weight). The differential term serves to con-
trol overshoot by thrusting contrary to the object's motion; without it,
the object would accelerate toward the target until it had passed it (and
the integral term might continue in the same direction even beyond that!);
because the proportional term will decrease rapidly as the object nears its
target, however, and because the differential term is in the direction opp-
osite the object's current motion, the differential term (if it has the
proper scaling factor) will prevent overshoot.
When PID loops are properly configured ("tuned") they can work extremely
well. Unfortunately, they are often extremely sensitive to the behavior
of the system in question. If an automobile used a PID loop to stop at a
traffic signal and the loop was tuned to stop properly when the car had
350lbs of passengers, the car would very likely overshoot the signal if it
had 600lbs of passengers and would approach the signal excessively slowly
("Zeno's paradox" speed) if the car only had 200lbs of people. For this
reason, many systems use "fuzzy logic".
To contrast PID with fuzzy logic, consider the traffic signal and how a
person would normally respond to it. Rather than try to figure out the
"perfect" amount of braking to apply at each point, a person will most
likely divide the stopping action into several steps. Working from last
to first, they are:
 If the car is travelling 2mph or less, let it coast until it's where I
want it to be and then hit the brake.
 If it's travelling 2-5mph, wait until I'm 5 feet before the target, and
then slow down to 2mph.
 If it's travelling 5-10mph, wait until I'm 20 feet from the target, and
then slow down to 5mph.
Such a sequence of steps will allow a car to stop at a traffic signal with
good precision, no matter how heavily or lightly loaded the car is. The
biggest problem with it, however, is that carrying those instructions out
100% literally would result in a rather jerky stop. To avoid this problem,
fuzzy logic systems replace the "hard" rules (wait until 20 ft before target,
then slow down to 5mph) with "fuzzy" rules (if the car's speed is near 10mph,
then as the car nears a point 20 feet before the target start slowing down;
the closer the car comes to the 20' point or the faster the car is moving, the
more pressure the system puts on the brake. Such a system still will not
produce as nice a response as a properly-tuned PID loop, but the response
will often be much more useful than that of a badly-tuned PID loop.
Larry G. Nelson Sr.
At 02:51 PM 2/25/97 -1000, TONY NIXON 54964 wrote:
>I've got a rather tricky problem controlling a winch with a V6
>automatic petrol driven engine. Due to the mass of the winch there is a fairly
>large time lag from a throttle movement to a reaction from the winch
>drum. It would be comparable to a car towing a caravan.
>Has anyone got ideas on where I can look to get ideas / information for
>this type of control.
>Just when I thought I knew it all,
>I learned that I didn't.
An alternative to the traditional control loops like PID would be to try
fussy logic. Microchip has a fezzy logic system called Fuzzitech or
something like that. They do a simulation of controlling a crain that works
quite well. You should look into that.
Larry G. Nelson Sr.
Dear Learned group
Anyone here ever made their own cruise control?
I ask because I have been thinking about ading one to my pickup. They aren't
too expensive as after market add ons here in the UK but there are a coupld
of features I'd find nice that the units aren't fitted with. Seems to me a
pic could handle the task so was just wondering if anyone had had a go.
Veronica Merryfield, somewhere in Cambridgeshire, UK
"The best things in life aren't things"
Do not wish to be anything but what you are, and try to be that perfectly. -
St. Francis De Sales
http://www.yadvashem.org.il - Everyone should visit
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics
More... (looser matching)
- Last day of these posts
- In 2003
, 2004 only
- New search...