How can I decrease the overshoot on a discrete time PID controller?

How can I decrease the overshoot on a discrete time PID controller?

Decrease I

Decrease P

Decrease D

t-thanks Veeky Forums

Adjust the damping term.

Overshoot is a function of the damping term. You're going to have to play with the value(s) for a bit.

That, or analyze your Bode/Nyquist plots. You should be able to estimate your transfer function and make adjustments accordingly.

Or decrease D.

To be a bit more helpful after the memeing, can't you live adjust the PID? That will always give you the best results. Just excite your system by small increments and see and get a feel for what each parameter does.

If that does not work, read the wiki on PID controls, it also contains the formula for calculating the gains as well as an in-depth introduction to what each term does.

Basically, P keeps your signal close to the setpoint, I makes sure there is no nonzero offset between signal and setpoint, D makes your signal reach the setpoint faster and more efficiently (usually you want to start with this one to change overshoot behavior. If it's too small your system will overshoot, if it's too big it will be unstable)

I like how eloquently you just described music theory with signal theory as art.

Time signatures, notations (shift -/+ 1), bass/treble

drown it in ketchup

What? I may be a brainlet and have no idea what you're trying to tell me

Ah, sorry. Basically there are 8 notes (arithmetic has this thing called 'cast away the 9' so you can ignore that when it comes to multiplication) and 10 is just '5 x 2'.

But you still need a 'center', which is called middle C or 'c4' because the 9th note isn't a note it is 'transposition'. Meaning you go up or down an octave.

The notes you play are the 'integers', but their frequency/interval is designated by the time signature 'stated at the beginning of the piece as 4/4 or 6/8 or whatever the composer chose'.

P = forgiven errors (as in they may not be perfect but they aren't killing anyone so authenticate and not waste computation)
I = how many units/cycles have been successfully logged (repeated interval) this would be like your bass line or percussion
D = +1 or 'what other note matches this chord-key signature' which is like 'these notes form a major or minor which is a modulation of hte octave you are in'.

Each level is literally about 'how anal' you want to be to validate prior statements, because all the best values are 'the ones that would just sound and awesome and who argues over music tastes like they do politics?'

just use a neural network, PIDs are deprecated

Sorry...

Mathematically/positionally you just described:
P = Prime
I = N (you tend to modulate this further with 2 because you want 'divisibility')
D = +

Basically picking 'one thing' that has a regular interval of time and space, no matter how stupid, will still bring sense/salience/harmony to your chords.

You only allow Primes where there is no 'further division', meaning that no 'minus operator' can be applied (i.e. you aren't going to kill each other).

Starting to get an idea of what you're talk about, thanks for the clarification.

>Each level is literally about 'how anal' you want to be to validate prior statements
From experience I can tell that I > P > D in order to get a stable control if you're ramping each one from zero, not sure how that translates to your statement.

Sure, spend a shitload of time to get some fancy new shit that you don't understand and can't find any general purpose parts for instead of taking something that is simple (3 parameters really isn't that hard to keep track of) and has been proven to work for literally anything you may want to do.

I think in signal processing that is 'capacitance', the empty space something will allow you to fill and within what time/space parameters (pressure/temperature/etc)

So if you know that concept and how to min/max that shit, you're golden.

So, capacitance measurable/measured via PID

Then you +/- accordingly because you have the 'purer' signal that allows for infinite modulation. Or Prime.

There's more than one way to do it. You should first find out what causes the overshoot. Also, do you know how to do it in the continous case? If not, go look that up, then try to come up with a way to generalize it for discrete controllers. It's also a good idea to review your book (or whatever you use) on discrete control systems to be sure you grasp the concepts correctly.

>software solution when a hardware one is simpler.and cheaper
lmao csfags

>tfw using a software PID on an FPGA
Never would switch to a neural network tho, PID works perfectly fine, so why to fuck up everything for a chance of slight improvement

what is a tensor processing unit

fag

this

But a NN should be used to control the full plant, welcome to the future gramps.

Good luck with that, I'm reconstructing my setup ever 3 months.
For production NN may be better, for resarch something that works is better.