GABRIEL MURARIU, CATALINA ITICESCU, ADRIAN MURARIU, BOGDAN ROSU, DAN MUNTEANU, DANIELA L. BURUIANA ASSESSMENT OF WATER QUALITY STATE DYNAMICS USING ADAPTIVE FILTERING METHODS AND NEURAL NETWORKS APPROACHING CASE STUDY - DANUBE RIVER IN GALATI AREA The identification of a temporal evolution model for complex systems has, since ancient times, been a subject of great interest. Whether it is mechanical systems for which it was essential knowledge of the final state or electrical systems, the problem of identifying evolution over time has always been extremely interesting. In the case of a complex system such as a river, whose condition is described by a set of physico-chemical parameters, the time description of the evolution of the state becomes a rather difficult problem. In this paper, two ways of identifying and predicting the parameters describing the state of such a system are presented. A LRS type algorithm and a process of approximating evolution over time considering neural networks was used for comparison. Recorded series of pH and carbonic acid values were used as study parameters. The data used covers the period 1990-1998 and consists of measurements of the water samples taken from the Danube River in the area of Galati City. The main result was to obtain a rapid convergence for the adaptive filter used. For comparison, a number of 6 neural network models were built. Finally, findings and discussion of the results are presented.
Keywords: RLS algorithm, Water Quality Index, neural network, Pearson correlation