Virtual Reference Feedback Tuning & Artificial Neural Network. Tecniche per il controllo e la simulazione nel trattamento dell’acqua potabile.

Water treatment is a delicate process. Drinking water contaminated by micro-organisms or by chemical residues represents a serious risk for public health.
The process of clarification is the first step in purification of water.
The use of coagulating substances permits to destabilize the charges of suspended solid particles and allows to unite them in flakes with the elimination of these residues. An excessive dosage of coagulant could leave chemical residues injurious to health; on the contrary, a dosage too low could not assure the normal demand for the water treatment.
Projecting a system of control assuring the right quantity of coagulant to be used in the mixing tank is the suggested solution.
In this document it is question to deal with the problems concerning the dosage controls and the realisation of a simulator of the clarification process of the purifying plant “Alaco” situated in Catanzaro (Calabria).
In order to avoid the very complex study of a mathematical model able to describe the system, the usage of alternative and innovative techniques, never used before for these problems, has been necessary.
The techniques that have been used, are “direct”; they are based on the knowledge of a great quantity of data selected in a real plant.
The system of control has been studied through a technique called Virtual Reference Feedback Tuning and the simulator has been realised thanks to a system of neural and artificial nets and exactly thanks to a system of Multi Layer Perceptron (MLP).