Near-field source positioning method of array PCA-BP algorithm with array errors

Abstract

The invention discloses a near-field source positioning method of an array PCA-BP algorithm with array errors. The method comprises the steps of Receiving K narrow-band, non-Gaussian and stable near-field sound source signals by using a uniform linear array, generating N groups of array receiving signal data in a training interval, extracting an upper triangular element of a covariance matrix of the array receiving signal data as a signal feature matrix, and performing PCA dimension reduction to obtain a dimension-reduced signal feature matrix; inputting the dimension-reduced signal feature matrix and the corresponding expected output into a BP neural network for training, and including the array error in the BP network through training, thereby obtaining a trained network estimation modelin the presence of the array error; and estimating near-field source parameters by using the trained network model. By using the array PCA-BP algorithm, errors are prevented from being corrected, thecalculation amount of the algorithm is reduced, the estimation performance of the algorithm under the low signal-to-noise ratio is improved, and good noise immunity is achieved.

Description

technical field [0001] The invention belongs to the technical field of array signal processing, and in particular relates to a near-field source positioning method using a PCA-BP algorithm with array errors. Background technique [0002] In many research fields such as passive positioning, sonar array direction finding, seismic and geological resource detection, and mobile communication, the estimation of DOA (Direction of Arrival) is a hot issue. Although some classic high-resolution algorithms can effectively estimate the direction of arrival of the signal, such as the MUSIC (Multiple Signal Classification) algorithm and the ESPRIT (Estimating Signal Parameter via Rotational Invariance Techniques) algorithm, such algorithms are very sensitive to array errors. In the presence of array errors, the performance of high-resolution algorithms will be seriously affected, so effective error correction must be carried out. Error correction often establishes different error models a...

AI Technical Summary

Problems solved by technology

Most of the existing methods use the upper triangular elements of the covariance matrix as the characteristics of the signal for network training. If this method is moved to the near-field situation, the model will be very complicated and the amount of calculation will be large

Images