Active sonar echo reverberation suppression method and device based on odd-number high-order cumulant

Abstract

The invention discloses an active sonar echo reverberation suppression method and device based on odd-number high-order cumulants. The method mainly comprises the steps that echoes of single-frequency pulse signals emitted by an active sonar are received by using a vector hydrophone; performing narrowband filtering and reverberation normalization processing on the echo; performing odd-order high-order cumulant calculation on the processed signal; and filtering the signal obtained after cumulant calculation, and filtering out harmonic waves to obtain a fundamental frequency signal. According to the method and the device, noise and reverberation of the active sonar echo signal can be effectively filtered out, the signal-to-mixing ratio is improved, and engineering implementation is simple and feasible.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of underwater acoustic signal processing, in particular to an active sonar reverberation suppression method and device based on odd high-order cumulants. BACKGROUND

[0002] Noise and reverberation are the main factors limiting the performance of active sonar. A noise and reverberation suppression algorithm with good performance is often limited by platform computing power and endurance and cannot be used in engineering. A simple and effective noise and reverberation suppression method can improve the detection performance of active sonar in engineering.

[0003] According to the sonar equation, the signal-to-noise ratio of the echo of an active sonar increases with the increase of the sound source level. The signal-to-noise ratio of the target echo at a distance of 1-2 kilometers is usually 10-20 dB. Therefore, the signal-to-noise ratio is not the main reason limiting the performance of the active sonar on a mine-laying platform. The signal-to-reverberation ratio is the main reason limiting the detection performance of the active sonar on the mine-laying platform in this distance range.

[0004] When the signal-to-reverberation ratio is high, narrowband filtering, array signal processing and Doppler shift information can be used to suppress reverberation, that is, the detection of the echo of the active sonar can be realized. When the signal-to-reverberation ratio is low, array signal processing and Doppler information are usually used to suppress reverberation. When the signal-to-reverberation ratio is further reduced, a more complex reverberation suppression algorithm must be used to realize detection. There are many such methods in the academic field, such as time-frequency analysis and low-rank matrix decomposition. However, these methods have large computational complexity, or the selection of parameters depends to a large extent on human selection, and cannot be used in engineering.

[0005] In the papers “Application of Proportion Factor and Amplitude Method in Signal Processing” and “Amplitude Method - A New Gear Fault Signal Preprocessing Method”, a method of using the third power is proposed to improve the signal-to-noise ratio. This method performs a cubic operation on each receiving point to improve the signal-to-noise ratio. This method is applied to the field of mechanical fault detection, extracts the periodic components in mechanical noise, and is used to improve the signal-to-noise ratio of mechanical signals to suppress noise and enhance signals.

[0006] In the existing method, narrowband filtering and other methods can effectively suppress out-of-band noise, but the frequency of reverberation is close to that of the target echo, and it is difficult to suppress reverberation when the Doppler shift is small. The two papers mentioned above mainly aim to solve the problem of improving the signal-to-noise ratio in the field of mechanical fault detection, and do not involve reverberation. Moreover, this method uses the different characteristics of signal amplitude distribution and noise amplitude distribution, and uses the principle that the third power provides different weighting values for signals of different amplitudes to improve the signal-to-noise ratio. Although the third power calculation improves the signal-to-noise ratio, the two papers do not provide an effective solution to the harmonic components that appear. SUMMARY

[0007] This disclosure provides a simple and effective method for suppressing reverberation in active sonar echoes based on odd-order higher-order cumulants, which can be used for detecting active sonar echoes in reverberant backgrounds on mine-laying platforms. This method is computationally simple, highly reliable, and can be implemented using filters commonly used in existing engineering projects, making it highly practical for engineering applications.

[0008] This method is applicable to the signal processing of active sonar echoes from minefield platforms, and is only suitable for single-frequency pulse signals, using a vector hydrophone for signal reception. Assuming the active sonar transmits a single-frequency pulse signal with a center frequency of f0, the active sonar echo signal is processed according to the following steps:

[0009] S1, Set up a narrowband filter and store the filter coefficients to suppress out-of-band noise;

[0010] S2 uses a vector hydrophone to receive the echo of a single-frequency pulse signal emitted by an active sonar.

[0011] S3, Narrowband Filtering and Reverberation Normalization: The received signal is narrowband filtered using the filter from step S1 to remove out-of-band noise; then active sonar echo reverberation normalization is performed.

[0012] S4, calculate the odd higher-order cumulative quantity of the signal output in step S3;

[0013] S5. Filter the signal obtained in step S4 to remove harmonics and obtain the fundamental frequency signal.

[0014] Furthermore, in step S1:

[0015] If the echo does not have a Doppler frequency shift, then the center frequency f of the narrowband filter is equal to the center frequency f0 of the active sonar transmitted signal;

[0016] If the echo has a Doppler frequency shift, adjust the bandwidth of the filter passband according to the estimated echo frequency, or set up a filter bank.

[0017] Furthermore, let the output signals in step S3 be y1(n), y2(n), ..., y N (n), where N is the number of output channels of the vector hydrophone; n represents the pulse sequence number;

[0018] The specific calculation method for S4 includes: calculating the P-order cumulant of the output signal from step S3 to obtain the sequence z(n):

[0019] z(n)=E[y(n) P ]

[0020] Where P is a positive odd number greater than 1; E[y(n)] P ] indicates that the output signal of step S3 is statistically averaged.

[0021] Further, in the step S4:

[0022] P is 3 or 5.

[0023] Further, in the step S4, the calculation method of E[y(n) P ] is:

[0024] The received signals of each vector hydrophone channel are calculated P times and averaged, and the calculation formula is:

[0025]

[0026] The odd high-order cumulant-based active sonar reverberation suppression device applying the above method mainly comprises:

[0027] A vector hydrophone module is configured to receive the echo of the single-frequency pulse signal emitted by the active sonar.

[0028] A narrowband filtering module is configured to perform narrowband filtering on the received signal to filter out the out-of-band noise, and filter out the harmonic of the signal obtained by the cumulant calculation module to obtain a base frequency signal.

[0029] A reverberation normalization processing module is configured to perform active sonar echo reverberation normalization processing.

[0030] A cumulant calculation module is configured to calculate odd high-order cumulants of the signal output by the reverberation normalization processing module.

[0031] The present disclosure uses a vector hydrophone to receive the target echo and reverberation of an active sonar, uses the characteristic that the signals of each channel of the vector hydrophone only differ in amplitude and phase factors, and uses odd high-order cumulants to suppress noise with a symmetrically distributed probability density function. As can be seen from the calculation expression, the odd high-order cumulants (the present disclosure focuses on the third and fifth order cumulants) introduce signal distortion while suppressing noise. Signal analysis shows that the signal distortion causes the appearance of the base frequency and harmonics, and when the signal-to-clutter ratio of the received signal is greater than 0, the signal-to-clutter ratio at the base frequency is improved. Therefore, only the high-order harmonics need to be filtered out using a filter to obtain the output signal with improved signal-to-clutter ratio. The output signal can continue to improve the signal-to-clutter ratio using the present method.

[0032] The main difference between the present disclosure and the prior art is that: ① the noise suppression principle is different: the prior art uses cubic operation to apply different weights to signals of different amplitudes to achieve signal enhancement and noise suppression; the present disclosure uses the characteristic that the odd high-order cumulants of symmetrically distributed signals are 0 to achieve noise suppression; for this reason, the prior art only needs to use one channel signal, while the technology disclosed in the present disclosure must use a vector hydrophone to calculate the statistical average of each channel signal to obtain the odd high-order cumulants.

[0033] 2. Application target and field are different: the prior art mainly aims at suppressing noise and the target application field is mechanical fault detection; the technology disclosed in the present disclosure mainly aims at suppressing reverberation and secondarily aims at suppressing noise; as the common knowledge in the field, noise is independent of signal while reverberation is coherent with signal, so the difficulty of reverberation suppression is significantly greater than that of noise suppression; the application field of the present disclosure is signal processing of underwater active sonar echo, which is different from the field of mechanical fault detection;

[0034] 3. The present disclosure is used for signal processing of active sonar echo, for suppressing reverberation and noise, the sound source of the transmitted signal and the hydrophone of the received signal are located at the same position in space; the prior art is used for suppressing mechanical fault noise, there is no reverberation problem, and the sound source is a mechanical structure such as a gear generating fault, the sensor receiving the signal is usually located at the position outside or on the surface of the mechanical structure, which is not located at the same position in space as the sound source.

[0035] Compared with the prior art, the beneficial effects of the present disclosure are: 1. On the basis of suppressing noise, reverberation of active sonar echo can be effectively suppressed; 2. Signal processing of underwater active sonar echo can be realized, and the signal-to-reverberation ratio can be improved; 3. The calculation is simple and the reliability is strong, which can be realized by using the filter commonly used in existing engineering, and has good engineering practicability. BRIEF DESCRIPTION OF DRAWINGS

[0036] The above and other objects, features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings, in which like reference characters refer to like parts throughout the several views, and in which:

[0037] Figure 1 is a flowchart according to an exemplary embodiment of the present disclosure;

[0038] Figure 2 is a numerical simulation result;

[0039] Figure 3 is a comparison of lake test data processing. DETAILED DESCRIPTION

[0040] Preferred embodiments of the present disclosure will be described herein below with reference to the accompanying drawings. While preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided so that the present disclosure is more thorough and complete, and the scope of the present disclosure is fully conveyed to those skilled in the art.

[0041] For the noise and reverberation suppression problem of the CW signal in the active sonar, the disclosure provides an active sonar echo reverberation suppression method based on odd high-order cumulants. The method uses the suppression effect of high-order cumulants on Gaussian noise to suppress the noise component in the active sonar echo, and at the same time, the high-order cumulants improve the signal-to-reverberation ratio of the echo at the fundamental frequency; but this method introduces signal distortion, so with the help of a filter to filter out the signal distortion introduced by high-order cumulants, the signal-to-reverberation ratio of the echo is effectively improved.

[0042] The main steps of the method include:

[0043] S1, setting a narrowband filter, storing the filter coefficients, and used for suppressing out-of-band noise;

[0044] S2, using a vector hydrophone to receive the echo of the single-frequency pulse signal transmitted by the active sonar;

[0045] S3, narrowband filtering and reverberation normalization: using the filter of step S1 to perform narrowband filtering on the received signal to filter out the out-of-band noise; and then performing active sonar echo reverberation normalization processing;

[0046] S4, performing odd high-order cumulant calculation on the signal output by step S3;

[0047] S5, filtering the signal obtained in step S4 to filter out harmonics and obtain a fundamental frequency signal.

[0048] In an exemplary embodiment, assuming that the active sonar transmits a single-frequency pulse signal with a center frequency f0, the active sonar echo signal is processed according to the following steps:

[0049] Step (1): design a narrowband filter with a center frequency f, store the filter coefficients, and use them to suppress out-of-band noise; if there is a Doppler shift in the echo, the bandwidth of the filter passband should be adjusted appropriately according to the estimated echo frequency, or a filter bank should be designed; if there is no Doppler shift in the echo, f is equal to f0.

[0050] Step (2): the active sonar transmits a single-frequency pulse signal with a center frequency f0, and a vector hydrophone is used to receive the echo;

[0051] Step (3): narrowband filtering and reverberation normalization: using the filter coefficients designed in step (1) to perform narrowband filtering on the received signal to filter out the out-of-band noise; and then performing reverberation normalization processing according to the conventional active sonar echo processing procedure. This step is a conventional processing step for echo processing in the active sonar reverberation background, and is not the core of the invention, so it is not described in detail. The output signal of this step is denoted as y1(n), y2(n), …, y N(n) (N is 3 or 4), here there are N channels because the application uses a vector hydrophone, which usually contains 3 or 4 channels of output, the value of N should be determined according to the number of output channels of the vector hydrophone used.

[0052] Step (4): Calculate the P-order cumulant of the output signal of step (3) to obtain the sequence z(n), the processing method is

[0053] z(n) = E[y(n) P ]

[0054] Where P is a positive odd number and greater than 1. Taking 3 or 5 can meet most requirements; E[y(n) P ] is a symbolic notation, which means taking the statistical average of the output signal of step (3), here the received signal of each channel of the vector hydrophone is calculated to the P power and then averaged, the calculation formula is

[0055]

[0056] After this step, the symmetrically distributed noise will be suppressed, and the target echo signal will be distorted. After narrowband filtering, the noise only retains the energy at the frequency where the signal is located. According to the properties of high-order cumulants, 3, 5, 7 and higher order cumulants are 0 (provided that the signal is 0 mean and the probability density function is symmetrically distributed). Since the received signal has first passed through a narrowband filter, which makes the noise at the output end have a mean of 0, therefore, z(n) cannot completely suppress random noise by using high-order cumulants, but theoretically it can completely suppress the symmetrically distributed part of the noise.

[0057] Next, we describe how the target echo signal is distorted by the formula:

[0058] The narrowband filter output contains 3 components: the echo generated by the target (this signal is a deterministic signal, according to the characteristics of the vector hydrophone, the waveforms of this signal in each channel are the same, only the amplitudes differ), the target echo in the i-th channel is denoted as m i (n); environmental noise (this signal is a random signal), the environmental noise in the i-th channel is denoted as g i (n); reverberation generated by the active sonar (this signal is a random signal), the i-th channel is denoted as r i (n). Therefore, the i-th channel signal y i (n) can be expressed as

[0059] y i (n) = m i (n) + r i (n) + g i (n)

[0060] It is usually assumed that ambient noise follows a Gaussian distribution and is independent of the signal and reverberation. Therefore, according to the property that the P-order cumulant of Gaussian noise is 0, we know that:

[0061]

[0062] And since a single-frequency signal is used, then m i (n)=A i sin(wn), (where w = 2πf / f) s f s Let A be the sampling rate. i (where r is the amplitude of the echo signal); since it has already passed through a narrowband filter, r i The center frequency of (n) is also f, which can be denoted as f. R here i Indicates the amplitude factor. This represents the phase factor. Here, the target echo in the echo is taken as the reference, and its phase is set to 0.

[0063] For a third-order cumulant, P is 3; for a fifth-order cumulant, P is 5. Taking P values ​​of 3 and 5 as examples, we can obtain:

[0064]

[0065] As can be seen, the calculation here preserves the fundamental frequency while generating harmonics. When P=3, a third harmonic is generated; when P=5, both the third and fifth harmonics are generated. Similar results will be produced when P takes a larger odd number, so they are not calculated.

[0066] Taking P=3 as an example, we can obtain

[0067]

[0068] Among them, and The term is called the self-term, and the other terms are called cross-terms. The cross-terms change in a complex way with the reverberation phase and reverberation amplitude, so only the self-terms are used here to estimate the signal-to-mixing ratio.

[0069] For signal y i (n), whose confidence-to-mix ratio is

[0070]

[0071] When P = 3 and 5, the signal-to-mixing ratio for z(n) is approximately

[0072]

[0073] This indicates that at frequency f, the signal-to-mixing ratio of z(n) is improved, and y iThe higher the signal-to-mixing ratio (SMR) of z(n), the more significant the improvement in the SMR of z(n). This characteristic can be utilized to improve the SMR of active sonar echoes when the SMR is low. However, this also requires y i The signal-to-mixing ratio of z(n) must be greater than 0 dB; otherwise, z(n) will not be able to improve the signal-to-mixing ratio.

[0074] Step (5): Pass the signal z(n) obtained in step (4) through a filter to remove harmonics and obtain the fundamental frequency signal. The signal-to-mixing ratio (SRR) of this signal is improved. When P=3, the improvement is about 2SRR0dB, and when P=5, the improvement is about 4SRR0dB. The improvement is more significant when P is a larger value. Compared with the classic filter method, this method can significantly improve the SRR. Moreover, the calculation process of this method is simple and the amount of computation is small. It can be implemented by combining it with the narrowband filter commonly used in engineering, and has high engineering practicality.

[0075] Step (6): End.

[0076] The above processing procedure is for reference only. Figure 1 As shown.

[0077] As can be seen from the above analysis, in this embodiment, the received signal of the vector hydrophone is processed by odd-order higher-order cumulants, thereby improving the signal-to-mixing ratio: when using third-order cumulants, the signal-to-mixing ratio is improved by about 2SRR0dB, and when using fifth-order cumulants, the signal-to-mixing ratio is improved by about 4SRR0dB; where SRR0 is the signal-to-mixing ratio of the echo after the original received signal is output through narrowband filtering.

[0078] Theoretically, z(n) in step (4) completely suppresses the noise of the symmetrical distribution by using the odd-order higher-order cumulants;

[0079] To address the distortion introduced by odd-order higher-order cumulants to deterministic signals (i.e., target echo), a solution is proposed using narrowband filters to remove higher harmonics and retain the fundamental frequency, thereby eliminating the signal distortion introduced by higher-order cumulants to the target echo.

[0080] Numerical simulation (as attached) Figure 2 ) and lake test data processing (as attached) Figure 3 The results show that, under simulation conditions, the proposed method improves the signal-to-noise ratio and signal-to-mixing ratio of active sonar echoes by more than 3 dB compared to the linear filtering method.

[0081] This disclosure applies to the processing of active sonar echo signals for mine-laying platforms. It requires the use of a vector hydrophone to receive the signal, and the active sonar transmits a single-frequency CW pulse signal.

[0082] The above technical solutions are only exemplary embodiments of the present application, and for those skilled in the art, on the basis of the application disclosed application method and principle, various types of improvements or modifications can be easily made, and are not limited to the method described in the above specific embodiments of the present application, therefore the above described method is only preferred, and does not have the meaning of limitation.

Claims

1. A method for suppressing active sonar echo reverberation based on odd-numbered higher-order cumulants, characterized in that, Includes the following steps: S1, Set up a narrowband filter and store the filter coefficients to suppress out-of-band noise; S2 uses a vector hydrophone to receive the echo of a single-frequency pulse signal emitted by an active sonar. S3, Narrowband Filtering and Reverberation Normalization: The received signal is narrowband filtered using the filter from step S1 to remove out-of-band noise; then active sonar echo reverberation normalization is performed. S4, calculate the odd higher-order cumulative quantity of the signal output in step S3; S5. Filter the signal obtained in step S4 to remove harmonics and obtain the fundamental frequency signal.

2. The method according to claim 1, characterized in that, In step S1: If the echo does not have a Doppler frequency shift, then the center frequency f of the narrowband filter is equal to the center frequency f0 of the active sonar transmitted signal; If the echo has a Doppler frequency shift, adjust the bandwidth of the filter passband according to the estimated echo frequency, or set up a filter bank.

3. The method according to claim 1, characterized in that, wherein... In step S3, the output signals are y1(n), y2(n), ..., y N (n), where N is the number of output channels of the vector hydrophone; n represents the pulse sequence number; The specific calculation method for S4 includes: calculating the P-order cumulant of the output signal from step S3 to obtain the sequence z(n): z(n)=E[y(n) P ] Where P is a positive odd number greater than 1; E[y(n)] P ] indicates that the output signal of step S3 is statistically averaged.

4. The method according to claim 1, characterized in that, In step S4: P can be 3 or 5.

5. The method according to any one of claims 1-4, characterized in that, In step S4, E[y(n)] P The calculation method for ] is as follows: The average of the received signals from each vector hydrophone channel is calculated by raising the P-th power. The formula is as follows:

6. An active sonar echo reverberation suppression device based on odd-numbered higher-order cumulants, employing the method described in any one of claims 1-5, characterized in that, include: Vector hydrophone module, used to receive the echo of single-frequency pulse signal emitted by active sonar; The narrowband filtering module is used to perform narrowband filtering on the received signal, filter out out-of-band noise, and filter out harmonics from the signal obtained by the cumulative quantity calculation module to obtain the fundamental frequency signal. The reverberation normalization processing module is used for active sonar echo reverberation normalization processing; The cumulant calculation module is used to calculate odd-numbered higher-order cumulants on the signal output by the reverberation normalization processing module.

Images