Application of a neural observer to phenols ozonation in water: simulation and kinetic parameters identification

Presented in this study, a dynamic neural network (DNN) is employed to estimate the states dynamics of the phenols-ozone-water system. A new technique based on the dynamic neural network observer (DNNO) with relay (signum) term is applied to estimate the decomposition dynamics of phenols and to identify their kinetic parameters without any mathematical model usage. The decomposition of phenols (phenol (PH), 4-chlorophenol (4-CPH) and 2,4-dichlorophenol (2,4-DCPH)) and their mixture by ozone, realized in a semi-batch reactor, is considered as a process with uncertain model ("black-box"). Only one parameter monitoring, namely, the ozone concentration in gas phase in the reactor outlet, is measured during ozonation. The variation of this variable is used to obtain the summary characteristic curve for the phenols ozonation. Then, using the experimental decomposition dynamics of phenols and of their mixture, obtained by HPLC method, the proposed DNNO is applied to estimate the ozonation constants of phenols at the different pH 2-12. A good correspondence between the decomposition dynamics and the estimated ones by DNNO is obtained.