Question

Explain the differences between the theoretical and real Smith Predictor models.

Answer 1

Smith predictor is a predictive controller which can be used for system with large dead time. Suppose there is a process which has a delay component in it. So if we want to control the system with PID controller we will feed back the output to the input side and the output will be compared with the reference input . The difference between the output and the reference input which is called error is then fed to the input side of the PID controller. Output from the controller then give control input to the system.

But for system with large dead time i.e with a delay component in it the output from the system that is to be compared with the reference input is outdated. So the close loop is feeding back outdated information. So when there is disturbance in the system the output of the system contains both delay term and the disturbance. To remove the delay term from the output of the system smith predictor is used.

Closed loop control system with delay component

A compensator is added to the system

The Smith Predictor general structure for delay compensation

Theoretical Smith predictor creates a models that imitate output of the system with the dead time .

So, when this model output is subtracted from the actual output the output is zero. In a real smith predictor the output is not zero. As system may be subjected to various disturbance from time to time. So in a real smith predictor only the disturbance part is fed to the input side via closed loop.