Research On Feature Extraction And Information Fusion Technology Of Cabin Noise Source

For the submarine, its concealment can measure its fighting strength. At low speeds,the noise of submarine structure is the most important noise sources, effective control and reduce the noise can increase the submarine stealth, this is the first consideration in the design of submarine link. Mechanical noise is generated by the mechanical equipment in the submarine during operation, and transfer to the submarine shell to the far field radiation through the submarine structure. But because of the type of noise source in the submarine is various, the coupling between the noise source, therefore the main noise source of the cabin classification is a complex process. Through the study of identification of noise sources,obtain the contribution of each main sound source size,can provide a good basis for the noise source and noise respectively.In this paper, the data fusion technology, the sensors to collect local information in certain fusion rules for processing and analysis and integration, on the basis of making full use of the sensor signal feature information, as far as possible reduced due to redundancy to the system due to the complexity, and ultimately to achieve the consistency of the target description, so as to improve the chamber of the main noise source recognition effect.This paper will analyze and study the following contents:First of all, according to the characteristics of submarine cabin of each noise source,simulation and test data were used based on time domain, frequency domain of various feature extraction techniques for feature extraction.Secondly, the basic principle and methods of information fusion were introduced, for target recognition of D-S evidence theory is described, and the probability of obtaining method, and analyzes several improved information fusion method.Finally, experiments are carried out in the cabin and the free field research, feature extraction and analysis of cabin noise source. And through D-S evidence theory and its improved algorithm of identification and classification of two kinds of target signals,obtained satisfactory recognition rate.