PID controller

[ccrause]

I recently wanted to apply PID control to an electric heater, but to my surprise I couldn't locate something already written in Pascal. I whipped up a PID implementation with some of the more common variations of parameters and control forms here.  Anyone is welcome to use the code.  Tips and tricks to improve the general behaviour of the controller would be appreciated.

[avra]

Very nice, thanks!

I used PID controllers in both industrial and embedded world, but prefer Fuzzy Logic for more complex stuff when I want to avoid PID overshooting/undershooting and smoothly reach set point, when rules based expert knowledge is available for something I want to automate, and when multiple inputs and multiple outputs are needed. All these can not be achieved with conventional PID. Of course, if you already have tuning experience and none of the above requirements then PID is quickest to implement.

[ccrause]

I used PID controllers in both industrial and embedded world, but prefer Fuzzy Logic for more complex stuff when I want to avoid PID overshooting/undershooting and smoothly reach set point, when rules based expert knowledge is available for something I want to automate, and when multiple inputs and multiple outputs are needed. All these can not be achieved with conventional PID. Of course, if you already have tuning experience and none of the above requirements then PID is quickest to implement.

Indeed PID is not the answer to all control problems 

For the situations where PID control are appropriate, it should be easy and straight forward to configure the class for the problem at hand. My aim is to include sensible default rules rules for things such as output limiting, integral windup and derivative filtering so that a casual user of the code can just focus on tuning the main controller constants.

[jamie]

Very nice, thanks!

I used PID controllers in both industrial and embedded world, but prefer Fuzzy Logic for more complex stuff when I want to avoid PID overshooting/undershooting and smoothly reach set point, when rules based expert knowledge is available for something I want to automate, and when multiple inputs and multiple outputs are needed. All these can not be achieved with conventional PID. Of course, if you already have tuning experience and none of the above requirements then PID is quickest to implement.

You need to apply a little D in the PID!

[jwdietrich]

You might want to use the CyberUnits Bricks library. It includes classes for proportional, integral and derivative elements. The main purpose of this class library is simulation of biological feedback loops in living organisms, but it should be usable for the implementation of technical control systems as well.

[ccrause]

You might want to use the CyberUnits Bricks library. It includes classes for proportional, integral and derivative elements. The main purpose of this class library is simulation of biological feedback loops in living organisms, but it should be usable for the implementation of technical control systems as well.

Thanks jwdietrich, I was not aware of this library.