Comprehensive information cross permutation entropy filtering noise reduction algorithm for ultrasonic echo signals
The useful signal components after CEEMDAN decomposition are screened through the comprehensive information cross-type multi-scale arrangement entropy algorithm, which solves the noise problem in the IMF component, realizes effective filtering and noise reduction of ultrasonic echo signals, and adapts to its concentrated frequency and weak waveforms. Features.
Patent Information
- Application Number
- CN202411139375.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-12-06
AI Technical Summary
There is positive and negative white noise in the IMF component obtained by CEEMDAN decomposition, and the useful signal components of the ultrasonic echo signal are highly similar to the excitation signal, resulting in poor filtering and noise reduction effect.
The comprehensive information cross-type multi-scale arrangement entropy algorithm is used to identify the noise components through CEEMDAN decomposition and multi-function arrangement entropy algorithm, and calculate the comprehensive information cross-over between the non-noise IMF components and the original signal and excitation signal, and filter out the high signal-to-noise ratio. Useful signal components.
It effectively removes the noise component in the ultrasonic echo signal, improves the filtering and noise reduction effect, and can identify weak ultrasonic echo signals with concentrated frequency and weak waveforms.
Smart Images

Figure CN119089115A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a mutual information permutation entropy CEEMDAN filtering denoising algorithm, belonging to the technical field of signal processing. Background Art
[0002] Empirical mode decomposition (EMD) is a new adaptive signal time-frequency processing method, which is very suitable for the analysis and processing of nonlinear and non-stationary signals. Based on the local characteristic time scale of the signal itself, EMD decomposes complex signals into the sum of finite intrinsic mode functions (IMFs). However, there is mode aliasing between the decomposed IMF components, which easily causes signal distortion after noise reduction. The ensemble empirical mode decomposition (EEMD) method can effectively eliminate the mode aliasing phenomenon existing in the EMD method, but it requires a higher number of iterations to reduce the auxiliary noise in the reconstructed signal, resulting in low efficiency. Complementary ensemble empirical mode decomposition (CEEMDAN) is based on EEMD, and replaces the added auxiliary white noise with positive and negative pairs of white noise. It can reduce the mode aliasing phenomenon while eliminating the residual auxiliary noise in the reconstructed signal and improve the computational efficiency. However, not all IMF components obtained by CEEMDAN decomposition have actual physical meanings. Some IMF components are positive and negative white noise, and the useful signal components of ultrasonic echo are relatively simple and have a very high similarity with the excitation signal.
[0003] Chen Bo et al. invented an improved empirical mode decomposition permutation entropy method in invention patent CN114176596A, which can effectively remove white noise through permutation entropy detection and windowing noise reduction processing, thereby reducing interference. This method can filter out the IMF component to remove noise, but it is not enough for the signal processing of ultrasonic echo. Ultrasonic echo signals have the characteristics of stable waveform, concentrated frequency but relatively weak echo, so it is necessary to further filter out the unique IMF component as the ultrasonic echo signal after filtering and noise reduction.
[0004] The multifunctional permutation entropy algorithm combines the multi-scale and PE algorithms, that is, performs PE optimization. The multifunctional permutation entropy algorithm can be very effective in analyzing time series; the comprehensive information cross algorithm can be used to measure the interdependence between multiple random signals, which can be regarded as the amount of information about several other random variables contained in a random variable. The present invention adopts a CEEMADAN algorithm based on the comprehensive information cross multi-scale permutation entropy, which can realize reliable filtering and noise reduction of ultrasonic echo signals. Summary of the invention
[0005] In order to solve the problem that not all IMF components obtained by CEEMDAN decomposition have actual physical meanings, some IMF components are positive and negative white noise, and the useful signal components of ultrasonic echo are relatively single and have a very high similarity with the excitation signal, a comprehensive information cross-permutation entropy filtering denoising algorithm for ultrasonic echo signals is proposed.
[0006] The technical solution adopted by the present invention to solve the above-mentioned problem is: the steps of the present invention include:
[0007] Step 1: CEEMDAN decomposition;
[0008] Step 2, the multifunctional permutation entropy algorithm is used to identify the noise component in the IMF component;
[0009] Step 3: Calculate the comprehensive information crossover between the non-noise IMF component and the original signal and the excitation signal.
[0010] Furthermore, the CEEMDAN decomposition in step 1 specifically includes:
[0011] Apply N pairs of positive and negative noise u to the original signal x(t) i (t), that is
[0012]
[0013] Get the signal with positive and negative noise added and Find its local maximum and local minimum points, use the cubic spline function to interpolate these local maximum and local minimum points to obtain the upper and lower envelope values, and calculate and The mean of the upper and lower envelopes and
[0014] Will and Subtract the envelope mean respectively, and we get:
[0015]
[0016] test and Whether the number of extreme points and the number of zero crossing points are equal or differ by at most one and the envelope mean at any point in the time interval is 0, if not, continue to repeat the above process until the kth cycle, and obtain the required and remember:
[0017]
[0018] and That is and The first IMF component screened out, and so on, can finally be and Decomposed into the sum of m IMF components and residuals, that is,
[0019]
[0020]
[0021] In the formula, is the newly selected IMF component, and is the irresolvable residual, and By finding the average value, we can get the kth IMF component,
[0022]
[0023] To the IMF ik By taking the ensemble average, we can finally get the m IMF components of x(t) after CEEMD decomposition.
[0024]
[0025] Furthermore, the identification of the noise component in the IMF component by the multifunctional permutation entropy algorithm in step 2 specifically includes:
[0026] Assuming the length of the IMF component is L, the time series of the IMF is processed by multi-scale coarse-graining, as shown in the following formula:
[0027]
[0028] Where s is the scale factor, It is a multi-scale time series. The core of multi-scale coarse-graining is to segment the time series and take the average value of each segment to improve the calculation accuracy. If the value of s is too large, it cannot reflect the complexity of the time series. If it is too small, it is difficult to extract segments and the analysis effect is poor.
[0029] The essence of multifunctional permutation entropy is to solve multi-scale time series The permutation entropy value of The sequence was spatially reconstructed using the coordinate delayed reconstruction method.
[0030] Take multi-scale time series Continuously sample m points to obtain a multi-scale time series The m-dimensional space reconstruction vector Where σ is the time delay, m is the embedding dimension, and the time reconstruction sequence is arranged in ascending order to obtain m! permutations, and the probability of each permutation is calculated, denoted as P i (0 <P i ≤1,i≤m! ), calculate the MPE value of the time series according to the definition of entropy, Normalize the PE value to The value range of PE^ is 0 to 1. The closer PE^ is to 0, the more random and irregular the IMF component is; otherwise, it means that the sequence has a certain regularity. By setting the PE threshold, the noise component and the useful signal component in the IMF can be distinguished.
[0031] Furthermore, the calculation of the comprehensive information crossover amount between the non-noise IMF component and the original signal and the excitation signal in step 3 specifically includes:
[0032] The IMF component is cross-calculated with the original signal x(t) and the excitation signal u(t). The cross-calculation of comprehensive information refers to the amount of information M1 of multiple other random variables contained in a random signal. The information interaction is calculated using formula (12);
[0033]
[0034] Where p(IMF,x(t)) is the joint probability density distribution of IMF and x(t), and p(IMF) and p(x(t)) are the IMF
[0035] and the marginal probability density of x(t); similarly, the mutual information between the IMF component and the excitation signal u(t) is calculated to obtain the information M2 of the IMF component and the excitation signal, and the comprehensive mutual information M * Satisfying formula (12)
[0036] M * =kM1+(1-k)M2; (0<k<1)(12)
[0037] The comprehensive information cross-value M of the IMF where the useful signal component is located * The maximum value can be extracted to remove the noise component in the ultrasonic echo signal x(t) and achieve the purpose of noise reduction.
[0038] The beneficial effects of the present invention are:
[0039] 1. The present invention provides a filtering processing method for ultrasonic echo signals based on CEEMDAN of comprehensive mutual information multi-scale permutation entropy, wherein a series of IMF components obtained by CEEMDAN decomposition are filtered out and removed by multi-scale permutation entropy algorithm, and the remaining IMF components are selected to perform comprehensive information correlation calculation with the original signal and the excitation signal, and the IMF component with the strongest mutual information correlation is the ultrasonic echo signal with a higher signal-to-noise ratio;
[0040] 2. By Figure 1 and Figure 2 By comparison, we can see that the comprehensive information interactive multi-scale permutation entropy CEEMDAN algorithm has a better filtering and denoising effect. It can better adapt to the characteristics of ultrasonic echo signal with concentrated frequency and weak waveform amplitude. It can identify weak ultrasonic echo signals mixed with noise signals, thereby achieving effective filtering and denoising. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 is the time domain graph of the original signal x(t);
[0042] Figure 2 It is a schematic diagram of the filtered waveform of x(t) after the mutual information multi-scale permutation entropy CEEMDAN algorithm;
[0043] Figure 3 It is a schematic diagram of the 18 IMF components obtained by decomposing the original signal x(t) through CEEMDAN;
[0044] Figure 4 This is the flow chart of the mutual information multi-scale permutation entropy CEEMDAN algorithm. Example
[0045] The 18 IMF components obtained by CEEMDAN decomposition of the original signal x(t) are as follows: Figure 2 As shown, from Figure 2 It can be seen that x(t) is decomposed into 18 IMF components by the CEEMDAN algorithm; the MPE^ calculated by the multi-scale permutation entropy algorithm of the IMF components is shown in Table 1. Through multiple experiments, the threshold of the ultrasonic echo signal can be set to 0.2. MPE^ higher than 0.2 can be considered as a noise signal, and MPE^ lower than 0.2 can be calculated by mutual information.
[0046] Table 1
[0047]
[0048]
[0049] The excitation signal is a multi-cycle sinusoidal signal modulated by the Hanning window, and the formula is:
[0050] u(t)=A×(1-cos(2πf0t / N))×sin(2πf0t)(13)
[0051] Where A is the amplitude of the excitation signal, f0 is the center frequency of the excitation signal, and N is the number of cycles of the Gaussian pulse. The signal here uses a five-cycle sine signal modulated by a Hanning window with a center frequency of 200KH. That is, A=1, f0=200KHz, N=5;
[0052] After calculation, the IMF9-IMF of the filtered noise signal 18 The mutual information M1 and M2 of the components are shown in Table 2. Taking k = 0.8, we can get the comprehensive mutual information M * As shown in Table 2,
[0053] Table 2
[0054]
[0055] Through the calculation of comprehensive information cross-algorithm, we can get IMF 10 The component has the most characteristic information with the original signal x(t). 10 The components are extracted to complete filtering and noise reduction.
[0056] The above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent replacement and improvement made to the above embodiments without departing from the content of the technical solution of the present invention, based on the technical essence of the present invention, within the spirit and principles of the present invention, still fall within the protection scope of the technical solution of the present invention.
Claims
1. A comprehensive information cross-arrangement entropy filtering denoising algorithm for ultrasonic echo signals, characterized by: The comprehensive information cross-arrangement entropy filtering denoising algorithm for ultrasonic echo signals includes CEEMDAN decomposition of the IMF mode of the original signal x(t), cross-calculation of the IMF component and the comprehensive information of the original signal x(t) and the excitation signal, and the specific steps are as follows: Step 1: CEEMDAN decomposition; Step 2, the multifunctional permutation entropy algorithm is used to identify the noise component in the IMF component; Step 3: Calculate the comprehensive information crossover between the non-noise IMF component and the original signal and the excitation signal.
2. The comprehensive information cross-permutation entropy filtering denoising algorithm for ultrasonic echo signals according to claim 1 is characterized in that: The CEEMDAN decomposition in step 1 specifically includes: Apply N pairs of positive and negative noise u to the original signal x(t) i (t), that is Get the signal with positive and negative noise added and Find its local maximum and local minimum points, use the cubic spline function to interpolate these local maximum and local minimum points to obtain the upper and lower envelope values, and calculate and The mean of the upper and lower envelopes and Will and Subtract the envelope mean respectively, and we get: test and Whether the number of extreme points and the number of zero crossing points are equal or differ by at most one and the envelope mean at any point in the time interval is 0, if not, continue to repeat the above process until the kth cycle, and obtain the required and remember: and That is and The first IMF component screened out, and so on, can finally be and Decomposed into the sum of m IMF components and residuals, that is, In the formula, is the newly selected IMF component, and is the irresolvable residual, and By finding the average value, we can get the kth IMF component, To the IMF ik By taking the ensemble average, we can finally get the m IMF components of x(t) after CEEMD decomposition.
3. The comprehensive information cross-permutation entropy filtering denoising algorithm for ultrasonic echo signals according to claim 1, characterized in that: In step 2, the multifunctional permutation entropy algorithm is used to identify the noise component in the IMF component, specifically including: Assuming the length of the IMF component is L, the time series of the IMF is processed by multi-scale coarse-graining, as shown in the following formula: Where s is the scale factor, is a multi-scale time series, Take multi-scale time series Continuously sample m points to obtain a multi-scale time series The m-dimensional space reconstruction vector Where σ is the time delay, m is the embedding dimension, and the time reconstruction sequence is arranged in ascending order to obtain m! permutations, and the probability of each permutation is calculated, denoted as P i (0 <P i ≤1, i≤m! ), calculate the MPE value of the time series according to the definition of entropy, Normalize the PE value to The value range of PE^ is 0 to 1. The closer PE^ is to 0, the more random and irregular the IMF component is; otherwise, it means that the sequence has a certain regularity. By setting the PE threshold, the noise component and the useful signal component in the IMF can be distinguished.
4. The comprehensive information cross-permutation entropy filtering denoising algorithm for ultrasonic echo signals according to claim 1, characterized in that: The calculation of the comprehensive information crossover between the non-noise IMF component and the original signal and the excitation signal in step 3 specifically includes: The IMF component is cross-calculated with the original signal x(t) and the excitation signal u(t). The cross-calculation of comprehensive information refers to the amount of information M1 of multiple other random variables contained in a random signal. The information interaction is calculated using formula (12); Where p(IMF,x(t)) is the joint probability density distribution of IMF and x(t), p(IMF) and p(x(t)) are the marginal probability densities of IMF and x(t), respectively. Similarly, the mutual information between the IMF component and the excitation signal u(t) is calculated to obtain the information M2 of the IMF component and the excitation signal, and the comprehensive mutual information M * Satisfying formula (12) M * =kM1+(1-k)M2;(0 <k<1) (12) The comprehensive information cross-value M of the IMF where the useful signal component is located * The maximum value can be extracted to remove the noise component in the ultrasonic echo signal x(t) and achieve the purpose of noise reduction.
Citation Information
Patent Citations
Magnetocardiogram signal denoising method for improving empirical mode decomposition permutation entropy
CN114176596A
Seismic signal random noise suppression processing method
CN108267784A
CEEMD and wavelet packet-based ultrasonic signal denoising method
CN108491355A
Electrocardiosignal denoising method based on improved EMD (Empirical mode decomposition) and threshold method fusion
CN110680308A
Acoustic emission signal feature extraction method, acoustic emission signal recognition method and device, acoustic emission signal feature extraction device and storage medium
CN113217102A
