NEURAL MODELLING AND SLIDING MODE CONTROL OF BIODEGRADATION PROCESS IN A ROTATING BIOREACTROR

A new Recurrent Neural Network Model (RNNM) has been applied for measurement data filtering and parameters plus state estimation of hydrocarbons biodegradation process, contained in polluted slurry, treated in a rotating bioreactor. The pattern used for RNNM back-propagation learning is composed by six input variables and three output variables. The total time of learning is 200 epochs of 76 iterations each and the Mean Squared Error reached is below 1.25%. Then the RNNM is simplified and used to design a sliding mode control the two-input two-output high order nonlinear plant. The MSE% of control reached 1% at the end of the process.