Recurrent Neural Kalman Filter Identification and Indirect Adaptive Control of a Continuous Stirred Tank Bioprocess

The aim of this paper is to propose a new Kalman Filter Recurrent Neural Network (KFRNN) topology and a recursive Levenberg-Marquardt (L-M) algorithm of its learning capable to estimate states and parameters of a highly nonlinear Continuous Stirred Tank Bioreactor (CSTR) in noisy environment. The estimated parameters and states obtained by the proposed KFRNN identifier are used to design an indirect adaptive sliding mode control scheme. The obtained simulation results of the real-time neural identification and control of a CSTR model, taken from the literature, exhibited fast convergence, noise filtering, and low mean squared error of reference tracking. A 20 runs comparative validation experiment in noisy environment is also done. It gives some priority of the L-M learning over the BP one. Keywords— Backpropagation learning, continuous stirred tank bioreactor, indirect adaptive sliding mode control, Kalman filter recurrent neural network, Levenberg-Marquardt learning.