The establishment of intelligent systems for classifying marine vessels based on their acoustic radiated noise is of major importance in the sonar systems. This paper presents a new method for classification of marine vessels. In this method, the acoustic radiated noise of ships is modded by an AR (Autoregressive) model with appropriate order and Coefficients of this model are used for classiRcation of ships. A Probabilistic Neural Network (PNN) is used as the classifier and the AR model coeftlcients are used as the input vector to this classifier. The performance o f this method is examined by using a bank of real data files.The results of evaluating the proposed method with real data show that this method is successful in classifying ships into three separate classes (heavy ships, medium ships, and boats). 1.INTRODUCTIONThe cIassification and recognition of marine vessels using their radiated noise is one of the major needs of sonar systems. Using the experience of sonar operators for cIassifymg detected targets is a method that has been commonly used for many years but it is obvious that using an intelligent system for this purpose has important advantages.Due to the militxy applications of the ship and submarine classification methods, and the difficulty of real data measurements, the literature on this subject is rare and usually classified. Thus, the methods presented by the existing few papers on this subject are usually evaluated using simulated data or very few real data files.The non-stationary nature of radiated noise of marine vessels, the random nature of some of the parameters and also the difficulty in the simulation of radiated noise leads to a lack of reliability in the classification algorithms which have been examined by red data with low variety or with data driven by simulators.This paper suggests a new method for classification of ships using their acoustic radiated noise. The proposed method uses an autoregressive model fur the ship radiated noise. The coefficients of this model are estimated and used as a feature vector for classifying ships. The classifier is a Probabilistic Neural Network (P").The performance of the proposed method will be examined using a variety of real data files.In part 2, the characteristics of ambient noise and radiated noise wilt be briefly described. . In part 3, the existing methods for shi noise classification are briefly discussed and criticized. Ia the 4 part, we will explain our suggested method and present the evaluation results. i ! 0-7803-91 03-9/05/$20.00@2005 IEEE KTHE BACKGROUND NOISE AND ACOUSTIC RADIATED NOISEUnderstanding the nature and characteristics of ambient noise and the acoustic radiated noise from ships is of great importance for selection of the discriminating features and the classification algorithm. Therefore, we present a brief explanation about sources of ambient and radiated noises and their spectral characteristics. A. Ambient noise LI-31The major sources of ambient noise are considered to be: seismic disturbances, biological organis...
Shipping noise and wind are the dominant sources of ocean noise in the frequency band between 20 and 500 Hz. This paper analyzes noise in that band using data from the SPICEX experiment, which took place in the North Pacific in 2004–2005, and compares the results with other North Pacific experiments. SPICEX included vertical arrays with sensors above and below the surface conjugate depth, facilitating an analysis of the depth dependence of ambient noise. The paper includes several key results. First, the 2004–05 noise levels at 50 Hz measured in SPICEX had not increased relative to levels measured by Morris [(1978). J. Acoust. Soc. Am. 64, 581–590] at a nearby North Pacific site three decades earlier, but rather were comparable to those levels. Second, at 50 Hz the noise below the conjugate depth decreases at a rate of −9.9 dB/km, which is similar to the rate measured by Morris and much less than the rate measured by Gaul, Knobles, Shooter, and Wittenborn [(2007). IEEE J. Ocean. Eng. 32, 497–512] for the CHURCH OPAL experiment. Finally, the paper describes the seasonal trends in noise over the year-long time series of the measurements.
In this paper, we present an orthonormal version of the new information criterion (NIC) algorithm for fast estimation and tracking of signal subspace using a twolayer linear neural network (NN). The fast orthonormal NIC referred as to FONIC algorithm guarantees the orthonormality of the weight matrix at each iteration. The proposed FONIC algorithm has a linear complexity which makes it efficient in real time applications. The FONIC algorithm provides a fast on-line learning of the optimum weights for the two-layer linear. Simulation results show better performance of FONIC algorithm than the NIC algorithm.Index Terms-new information criterion, subspace tracking, adaptive algorithm, neural network learning.I.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.