DYNAMIC MODELING OF THE ACTIVATED-SLUDGE PROCESS - IMPROVING PREDICTION USING NEURAL NETWORKS

A procedure has been developed to improve the accuracy of an existing mechanistic model of the activated sludge process, previously describe d by Lessard and Beck [Wat. Res. 27, 963-978 (1993)]. As a first step, optimization of the numerous model parameters has been investigated u sing the downhill simplex method in order to minimize the sum of the s quares of the errors between predicted and experimental values of appr opriate variables. Optimization of various sets of parameters has show n that the accuracy of the mechanistic model; especially on the predic tion of the dissolved oxygen (DO) in the mixed liquor, can be easily i mproved by adjusting only the values of the overall oxygen transfer co efficients, K(L)a. Then, in a second step, neural network models have been used successfully to predict the remaining errors of the optimize d mechanistic model. The coupling of the mechanistic model with neural network models resulted in a hybrid model yielding accurate simulatio ns of five key variables of the activate sludge process.