PID tuning when error input data to PID and output data from Plant is available.

I don't have a mathematical model of the plant. Currently, PID is tuned with some fixed Kp, Ki, and Kd and I want to optimize these parameters for a stable system. PID is giving output to a non-linear control valve which is used to control a temperature at the certain place by supplying amount of water. Error input to the PID is the difference between set value of temperature and actual temperature. The data of error input to PID, Control valve opening and actual output temperature are available to me. For example Set value of temperature (SV) = 980 deg C Actual output temperature (PV) = 998 deg C, Error term to PID = -18 Deg C, Control Valve Opening = 33 %, like this I have many sets of data points available

So, is it possible to tune the PID from the following data? Thanks in Advance

Answer 1

[Arkadiy Turevskiy]

Hi there,

If you do not have a model, but have input/output data from the model, you can use system identification to identify plant dynamics from input/output data. Note that you will need the transient data, not just steady state data like you describe in your question.

If you have transient data, you can follow approach described in this example .