# How can I apply different control approach without having a state space model?

Recently i am working on a hydraulic control system of a vehicle. But in this system I don't have any mathematical modeling of the system so i am not able to get any state space model for the given system. I have to control the length of Piston in Hydraulic Cylinder by giving input of desired piston position. Modeling of this system is established in Matlab/Simulink using physical modeling tools SimScape Toolbox and simFluids. As one approach i had already try to apply a feedback with P-control as show in the figure Matlab System and also the gain of P control is derived by trial and error.

Now if i want to apply state obeserver or optimal Control than how can i apply on this given system.

And is necessary to derive mathematical model of this system to desing controller?

Thank You.

• This question was also posted at the Math and Engineering Stack Exchange sites. There's a quality answer here (OP asks another answerer on the Math site about 50k4's answer), so I'm reluctant to close it here, but cross-posting is not allowed as a bulk copy-paste. If you have a question that could fit on multiple sites, then tailor it for each site.
– Chuck
Nov 7, 2016 at 18:33
• @Chuck sorry for that inconvenience. I am not aware of that situation. I will keep in mind for the next time. Nov 8, 2016 at 8:46

Many control strategegies (PID is not neceserily one of these) are computing some of their parameters from your system. If you do not want to model your system yourself (or play around with it until you identify its charecteristics) you can use Matlab to identify your system model using the system identification toolbox.

The System Identification Toolbox will give generate some inputs for your system and will record the responce for these. Based on this it will generate a simplfyied model which can be used to represent your system. It is simplifyed but in many cases it works well as a substitute.

When having this identifyed model you can freely implement any control strategy using the equations given after the system identification process.

• thanks for your response i am going to try your approach to get system model. Nov 7, 2016 at 13:52

There are different approaches to your issue. First and most likely easiest one is the approach suggested by 50k4. With a simplfied model you can tune some basic controller like PID. But there are tons of nice Design approaches, which can work totally without the model. Those controllers are usually grouped under the name "soft controllers" Some key words are Fuzzy, genetic algorithm or neural network....

Basically for every controling process you need some knowledge; in most cases you just take the mathematical description of the system. For those soft controlling approaches you take your own expert knowledge of the system... sounds a bit strange but it can work out pretty nicely.

At the end you are able to control systems where it is impossible to get a mathematical system description