So, we do not speak about scailing, but scaling. Your question is so general, bring the real problem to be discussed. What is the input, what is the type of control (PID...) and controller, sensor type, controlled object, output of the system... You have to be more precise!
In general, in feedback control, we use the method called normalization. First, you should create a math model (typically a set of linear or nonlinear differential equations). You define the parameter set, usually rated values of the system, which you are going to use as a reference for normalization. Next, you perform the normalization by dividing your sensed signals by corresponding references in your math model. This should bring all your physical signals (voltage, current, temperature, motor speed etc.) to the values close to or within zero to one (or -1 to +1) region. If you have analog or digital scaling by your equipment, typical for current and voltage sensing circuits, as well as A/D conversion, in place, you should include it in you calculations. Now, you can use these values for scaling of your other signals and component values, such as impedance normalization, power normalization, PWM signals etc. This way, you will have the entire system represented by p.u (per unit) non-dimensional math model, which you can apply to any other similar system, regardless of the power level. The only thing you need to do is to rescale it back using corresponding set of rated values. For more info, take a look at the following links:
I watched the debate. We should always encourage the person who is interested in developments in science and asks questions. On the other hand, everyone must strive to ask questions about linear systems, input-output relations or other specific conditions. Then it will be achieved cooperation between question and answer.