How Can I Use Matlab/Simulink to Optimize Control Strategies From Real Data

Hellow,

I am a mecatronics engineer and work in mineral processing, i have PH Control loop (adding Milk of Lime). I Have real data with current conditionds ( PID Control Loop). The control is not working well the PH is not in control and i need to correct that by tunning or changing the control strategy, i wonder if somebody could share with me his experience to face this problems with Matlab/Simulink help. Thanks. Luis.

[Arkadiy Turevskiy]

Hi Luis, My advise is to contact your local MathWorks representative. They will arrange a webex or a visit and will show you in detail how to apply our tools for solving this problem.

A short version of how you could go about fixing the problem is as follows:

1. Create a model of your process. Collect input /output data, and use it to create a model using System Identification Toolbox. Alteratively, if you know the equations decribing dynamics of the plant, create a model using those equations, and calibrate model parameters from test data using Design Optimization.
2. Now that you have a model that captures your process dynamics, retune your PID loop with the PID Tuner. If PID controller does not do a good enough job, design an arbitrary compensator using SISO Design Tool.

Hope this helps.

Arkadiy

[Ryan G]

Is PID the best method here? My impression would be the reaction time required to change PH and measure it would be to slow.

I would think something like a bang-bang control, similar to house heating, would work better. Add some Lime, wait, check ph, determine if more lime is needed and repeat.