Extremum-seeking control with anticipative action of microbial fuel cell's power

Abstract

During the last decade, the microbial fuel cell (MFC) has been considered as a promising solution to produce renewable energy while reducing the excessive consumption of electrical energy in wastewater treatment centers. One of the problems facing the use of an MFC as a battery is the fact that its internal resistance varies with various external sources of disturbances causing a variation in its optimal point of operation. In this case the use of a real-time optimization method is necessary for the battery to work at its optimum. Extremum seeking control (ESC) can be applied to optimize the system. However, in the case where important external disturbances cause rapid changes in the optimal operating point, ESC converges to a point other than the desired optimum because of its very slow convergence rate. In this paper, a method which judiciously combines an ESC routine with a neural-network based anticipatory action is proposed. The anticipatory action takes into account a measurable external disturbance, namely the inlet substrate concentration. Simulation results show that the proposed scheme leads to an improvement of the convergence rate towards the desired optimum.

Extremum-Seeking Control With Adaptive Excitation: Application to a Photovoltaic System

“…A judicious choice of ESC parameters is essential to ensure stability and compromise between accuracy and speed of convergence to the optimal operating point. Indeed, if the system is subject to large and rapid external disturbances, these performance criteria will be degraded in the case of optimization of non-linear systems with unknown dynamics as previously shown in a microbial fuel cell application [4]. Also in [13], for the same system, the effect of the choice of and on ESC performance in terms of accuracy and speed of convergence is shown.…”

“…Thus, the identification of the model parameters becomes difficult and the optimum estimate becomes imprecise. In [4],a comparative study has been carried out between two anticipative ESC schemes using respectively a neural network model and a static linear model in the anticipation loop. The performance of the two approaches was compared through the optimization of the power delivered by a microbial fuel cell for which the measurement of the inlet substrate concentration was considered as the external perturbation.…”

“…In this case, the main difficulty of the ESC technique [4] is the requirement of multiple time scales between the dynamics of the system, the frequency of excitation and the speed of adaptation. The excitation must be an order of magnitude slower than the system dynamics to separate the effect of excitation from the dynamics of the system.…”

“…Assuming that these disturbance measurements are available to the expert, they can be used to improve the performance of ESC [4]. Indeed, the modeling of the relation between the optimal operating point of the system and the external disturbances allows a real-time estimation of the location of the system optimum for each new measurement of the external disturbances.…”

“…The study shows that ESC with a rectangular excitation signal is twice as fast as a sinusoidal signal and four times faster than a triangular signal. In [11] and [4], the authors indirectly controlled the initial condition at the ESC integrator, using anticipative action to improve the speed of convergence. In this paper, we perform an adaptive control of the amplitude of the excitation signal.…”