Artificial Intelligence Neural Network Approach to Self Tuning of a Discrete-Time PID Control System

Due to their efficiency in standard control problems, proportional-integral-derivative (PID) controllers are widely used in industrial control systems. Although this controller has been established as a control standard, tuning of its parameters and finding their optimal combination still represents a challenge, particularly under changing operating conditions, where control designer cannot rely on the invariance of the plant model. Tuning of the proportional, integral and derivative gain of a PID controller represents an optimization task, for which we propose in this work a solution based on artificial intelligence (AI) approach using radial basis (RB) function for activation of neural networks (NN) which adapt the controller gains and learn the plant model in order to account for the controller influence on the control outcome. The controller is implemented in a discrete-time system which enables real-time learning and implementation. The effectiveness of the proposed controller is tested on a benchmark example of a discrete-time model of a cantilever beam, obtained through the subspace model identification.