The invention discloses a fFeature extraction method for blind source separation

Abstract

The invention belongs to the technical field of communication, and particularly relates to a feature extraction method for blind source separation. The method provided by the invention is mainly characterized by comprising the following steps of: preparing materials; p; preprocessing the mixed blind source signal; O; obtaining a time-frequency map, inputting the data as training data into a neuralnetwork; D; deep learning method, fitting a neural network objective function with the following characteristics; i; in the process of minimizing the objective function; T; target function convergence, t, the Euclidean distance sum of the time frequency points of the same source signal reaches the minimum; w; when the Euclidean distance sum of different time-frequency points reaches the maximum,inputting the mixed blind source signals into the trained neural network, clustering the signals of different sources according to the output of the neural network, constructing a time-frequency masking matrix by utilizing the feature set, calculating the frequency spectrum, and obtaining separated time-domain signals. The beneficial effects of the invention are that the method can achieve the separation of mixed signals of a plurality of unknown source signals.

Description

technical field [0001] The invention belongs to the technical field of communication, and in particular relates to a feature extraction method for blind source separation. Background technique [0002] Blind source separation refers to the process of separating or estimating the original source signal using only the mixed signal detected by the sensor when the source signal and the transmission channel are unknown. Most of the current blind source separation methods are based on the sparse characteristics of the source signal in a transform domain. Under the premise of sparse assumption, some features carried by the observed signal in the time-frequency domain are analyzed, and the time-frequency points with the same feature (or similar features) are clustered, so as to realize the separation of the source signal. However, because the clustering method is particularly sensitive to outliers, and the traditional feature extraction methods cannot actually guarantee that the ex...

AI Technical Summary

Problems solved by technology

However, because the clustering method is particularly sensitive to outliers, and the traditional feature extraction methods cannot actually guarantee that the extracted features do not have outliers, so in practical applications, the estimation error of some time-frequency points will make the separation effect drop sharply.

Images