Time-Domain Feature Extraction and Analysis of Ship Radiated Noise

Feature capability is the most important in ship_ radiated noise target recognition. In this paper, Linear Prediction Coefficient(LPC), Linear Predictive Cepstral Coefficient(LPCC) and Mel-frequency Cepstral Coefficient(MFCC) are extracted from ship-radiated noise. Contribution to the recognition on the average of each dimension are calculated by two method for the first time. Based on the contributive rate, a proper feature weighting method is putted forward to improve the feature capability. It is indicated that the three features are all good representative of ship radiated noise by mass noise examinations. After optimized, they all can be most important features of ship radiated noise recognition.