Abstract: The problem of the active control of combustion instabilities in aeroengines is investigated. A function model is derived between the oscillation pressure and the mean fuel-air ratio in the combustion chamber of an aeroengine under unstable combustion, which implicates an extremum relation between the oscillation amplitude of the pressure in the combustion chamber and the mean fuel-air ratio. Moreover, a novel extremum seeking algorithm based on the annealing recurrent neural network is proposed in this article. By means of this method, the control of the minimal oscillation amplitude of the pressure in the combustion chamber can be realized by adaptively seeking the optimal mean fuel-air ratio of this extremum model. This method can effectively restrain the phenomenon of combustion instabilities in an aeroengine. By comparing the simulation results, the effectiveness of the proposed method is validated.

Key words: annealing recurrent, neural network, extremum seeking algorithm, aeroengine, combustion instabilities, active control