Abstract: A correlation dimension-based iteration-local projective method（CDBI-LPM）is proposed to reduce the noise in the acoustic signals from the armored vehicles and eliminate the harmful effects on feature extraction． 20 dB and 10 dB（SNR）noisy signals are obtained by adding the different Gaussian white noises to the Lorenz sequence of a typical chaos system．The noise reduction simulation results show that the proposed method not only reduces the amount of computation，but also efficiently recovers the dynamic structure of Lorenz system．The application of the method in the noise reduction of actual acoustic signals from tanks and wheeled armored vehicles shows that the noise intensities of sample signals are significantly reduced，the correlation dimensions decrease，and the disorder attractor structures of the signal series get order as well．This indicates that the proposed method could efficiently reduce the noise in the nonlinear signals.

Key words: ordnance science and technology, nonlinearity, armored vehicle, acoustic signal, noise reduction, local projection method