Abstract: Artillery shock wave signal test usually uses the notch algorithm to eliminate the power frequency
interference,but this method often loses the effective signal component. Therefore, an improved FastICA
algorithm is proposed to eliminate the power frequency interference. The FastICA algorithm based on negative
entropy is improved by using the Newton iteration form of fifth order convergence. The algorithm has high
accuracy of negative entropy algorithm, and has a fast convergence time and few iterations. The simulation results
show that the proposed algorithm has 0. 999 99 and 45 dB on the similarity coefficient and signal to noise ratio,
which is obviously superior to the traditional notch algorithm of 0. 996 and 21 dB. And compared with the
FastICA algorithm based on negentropy, the improved algorithm is the same as its precision,but the number of
iterations is reduced by 26. 7%. And compared with the kurtosis algorithm with faster convergence speed, the
improved algorithm has fewer iterations,faster convergence and higher stability. The algorithm has the advantages
of high precision, fast convergence speed and few iterations, and is suitable for real-time processing of shock
wave.

Key words:  shock wave signal, power frequency interference cancellation, fifth order convergence, FastICA