EMD (Empirical Mode Decomposition) endpoint effect suppression method based on HMM (Hidden Markov Model) correction and neural network extension

Abstract

The invention discloses an EMD (Empirical Mode Decomposition) endpoint effect suppression method based on HMM (Hidden Markov Model) correction and neural network extension. The method comprises the steps of A) using a sensor to obtain a signal; B) using a neural network extension algorithm to estimate partial known data in a signal endpoint, calculating estimation errors and predicting data outside the endpoint; C) using an HMM algorithm to establish a model for the estimation errors and using the parameters of the model to predict the extension errors of the used extension algorithm; D) using the predicted error data to correct the extended data to obtain final extended data; E) performing empirical model decomposition to the extended signal and abandoning the extended data at two ends to obtain IMF (Intrinsic Mode Function) components of the original signal; and F) extracting signal features by analyzing the IMF components after endpoint effect suppression. The EMD endpoint effect suppression method based on HMM correction and neural network extension has the advantages that the neural network extension algorithm can be corrected, the errors existing in a data extension method are reduced, and the endpoint effect in empirical model decomposition is effectively suppressed.

Description

technical field [0001] The present invention relates to the technical field of signal processing, in particular to an endpoint effect suppression method based on HMM (hidden markov model, Hidden Markov) correction and Neural Network Extended Empirical Mode Decomposition. Background technique [0002] After Empirical Mode Decomposition (EMD) was proposed by N.E.Huang, it has been widely used in various signal processing fields because of its excellent time-frequency analysis capabilities. The characteristic of the EMD time-frequency analysis method is to stabilize the non-stationary signal through EMD decomposition, decompose the fluctuations or trends of different scales step by step, and obtain the Intrinsic mode function (IMF for short). It is not only suitable for the analysis of nonlinear and non-stationary signals, but also suitable for the analysis of linear and stationary signals. [0003] After more than 20 years of development, the theory of the EMD method is still...

AI Technical Summary

Problems solved by technology

[0005] The problem to be solved by the present invention is to overcome the defects of the prior art in solving the endpoint effect of empirical mode decomposition, and provide a method for suppressing the endpoint effect of EMD based on HMM correction and neural network extension

Images