Adaptive RBFNN versus self-tuning controller: An experimental comparative study

In this paper a comparative study evaluates two different techniques: neural adaptive control and self tuning (ST) control. The neural adaptive control is based on a new hybrid learning technique using an adaptive learning rate for the on-line learning of a Gaussian Radial Basis Function Neural Network (RBFNN) type. In the self tuning structure the control parameters are updated from a pole placement design via the estimation of the process model. A selective forgetting factor method is applied to both control schemes: in the RBFNN to on-line estimate the second layer weights and in the ST to estimate the process parameters. These two techniques are applied to a laboratory-scaled pilot plant, an airstream heating (PT 326), in order to evaluate their performance. Experimental results show the effectiveness of both the proposed methods for set-point tracking and disturbance rejection.