Hi all,
I'm a novice to control systems, and have a question about the delayed system which has delayed feedbacks. In such a case, what is the general solution to control a system? Any reference is appreciated.
Thanks, Won
A common way to improve on simple PIC control is to add a Smith Predictor (that models the plant without the delay, and stands in for the plant till the delay times out). This works best on plants that do not experience much in the way of sudden disturbances, but only slow changes, relative to the delay. But it does allow setpoint changes to settle faster than with just a PID control.
There is no general solution. You are asking to predict the future. The Smith Predictor helps a bit but fundamentally your best bet is to eliminate deadtime in your system.
Walter.
I agree with Walter on all counts. There has been quite a bit of work on using various modeling methods, such as neural networks, to determine the model of a system. You can use the model of the eventual results (if it is accurate) as the feedback portion of your controller, though I would want to have something in the model to allow for corrections to reality now and then. This would essentially be a virtual sensor. There is no guarantee that you will be able to get a good neural network for this particular task.
Michael
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.