in order to obtain the reference dynamic model for the adaptive control of thechaotic optical system, a suggestion of identifying the chaotic optical system with the BPfeed forward neural network supported by the BP algorithm is made in this peper. Thefeasibility of this suggestion was demonstrated with the ctrpputer simulation throughidentifying a Bragg acousto-optic bistable &. chaotic system with a very simple 1: 4: 1 BPNN. The result of the computer simulation shows that the three-layer BP foreward NN, if trained with the BP algorithm, is indeed a fine identifier. Thus it could be used to reconstruct the dynamics of the chaotic optical system efficiently with its output time series with a satisfactory precision.