Waste water neutralization using a fuzzy neural network controller

In this paper, the pH neutralization process is identified and controlled using a fuzzy neural network structure. Three different approaches are compared. In the first method, a generic fuzzy controller was used in which membership functions and rule base are selected intuitively. For the second and third approaches, the fuzzy system is initialized using a fuzzy error backpropagation algorithm. While in the second case study no online adaptation takes place, in the last approach the membership functions are adjusted online. The results of these methods are presented and evaluated. In order to describe the system dynamics a radial basis function network was used.