Application of BP Neural Network and Higher Order Spectrum for Ship-radiated Noise Classification

摘要

Ship-radiated noise recognition has always been a difficulty in naval warfare. The current recognition methods are applying power spectrum density estimation to underwater signal processing technology. But due to the complexity of the circumstances and disguiser's skillful designs, it usually fails to meet the need of mine warfare. Then we need to higher order spectrum to recognize the targets. The paper analyzes the advantages and disadvantages between two methods, combines power spectrum density estimation and higher order spectrum to extract the distinguishable characteristics synthetically, then applies the BP neural network for auto-recognition. The paper also advances the improved arithmetic to the BP neural network. Through simulation test with the collected data and comparing with the result of other method of underwater targets recognition, the paper proves the effectiveness of applying BP Neural Network and Higher Order Spectrum for ship-radiated noise recognition.