Feature extracting method based on improved HHT transformation

Abstract

The invention discloses a feature extracting method based on improved HHT transformation. A wavelet packet is adopted to decompose and reconstruct measured signals to obtain narrow-band signals with different frequent bands; all the narrow-band signals are subjected to empirical-mode decomposition to obtain IMF components of all the narrow-band signals; sifting the IMF components of all the narrow-band signals to obtain final IMF components of the whole measured signals; the final IMF components are subjected to Hilbert transformation respectively to obtain instant attributes of the measured signals; according to the instant attributes of the measured signals, time-frequency statistical characteristics which can reflect the time-frequency characteristics of the measured signals are extracted, wherein the time-frequency statistical characteristics comprise the mean value of the instant amplitude of each IMF component, and the bandwidths, peak values and variances of marginal spectrums of all the IMF components. By means of the feature extracting method based on the improved HHT transformation, the precision of the time-frequency analysis of the signals is improved based on the improved HHT method for wavelet packet decomposition and virtual IMF component removal. The features for reflecting the signal properties can be more efficiently extracted, and finally, the efficiency of data mining and pattern recognition is improved.

Description

technical field [0001] The invention relates to a feature extraction method. In particular, it relates to a feature extraction method based on improved HHT transform for time-frequency analysis of nonlinear and non-stationary time-varying signals. Background technique [0002] 1. Feature extraction [0003] Feature extraction refers to obtaining one or more parameters from the original signal through mathematical transformation or statistical analysis, and these parameters can represent the characteristics of the signal in a certain aspect. These parameters are called the features of the signal, and the process of obtaining these features is feature extraction. Feature extraction is used in the fields of data mining and pattern recognition, and feature extraction is one of the key technologies. Through feature extraction, parameters that can reflect the characteristics of the data in the signal can be obtained, and the quality of the extracted features determines the perf...

AI Technical Summary

Problems solved by technology

[0010] The current existing problem of inaccurate feature extraction for nonlinear and non-stationary signals

Images