March/April 2011

ISA Certified Automation Professional (CAP) program

Certified Automation Professionals (CAPs) are responsible for the direction, design, and deployment of systems and equipment for manufacturing and control systems.

CAP question

Which of the following is an example of "Adaptive Control?"

A.        A controller uses human-like logic criteria and makes controller output changes based on quantities such as "small" and "moderate."

B.         A controller uses a quantitative process model and modifies controller tuning based on how the process responds in relation to the model prediction.

C.        A controller uses only Proportional and Integral actions in calculating controller output.

D.        A controller is comprised of layers of nodes, where the outputs from one node in a layer are multiplied by weighting factors to form the inputs to the nodes in another layer.

CAP answer

The correct answer is B, a controller uses a quantitative process model and modifies controller tuning based on how the process responds in relation to the model prediction. The essential element of "adaptive control" is the result of the control algorithm is a modification to the tuning parameters (or constants) in order to eliminate error based on changes in the process or process dynamics over time, i.e., adapting to changes in the process. "Self-tuning" controllers are based on adaptive control applied to a PID controller based on a set of tuning rules.

Answer A is not correct. This type of controller would be best described as a fuzzy logic controller, which calculates controller outputs based "rules," which use the sign of the error and the sign of the change in error.

Answer C is not correct. A PI controller is not adaptive since the controller logic itself does not adjust tuning parameters. A PI controller is a basic regulatory control algorithm.

Answer D is not correct. This example is a simplified description of an artificial neural network, which calculates changes to controller output, not changes in tuning parameters, to eliminate error.

Reference: McMillan, Gregory K., et al, Models Unleashed - Application of the Virtual Plant and Model Predictive Control - A Pocket Guide, ISA, 2004.