Method and system for estimating time of arrival of underwater acoustic signal based on multi-feature fusion

Abstract

The application discloses a kind of based on multi-feature fusion's underwater acoustic signal wave arrival time estimation method and system, endpoint detection algorithm is combined, based on the endpoint detection technology of multi-feature fusion's wave arrival time estimation stage, utilize the new decision basis to realize the accurate estimation of underwater acoustic signal wave arrival time.The application introduces endpoint detection technology into signal detection field.At the same time, due to the time-varying, multipath phenomenon of underwater acoustic channel, underwater acoustic channel is often represented as direct wave non-strongest.In this environment, relative to other characteristic algorithms, the application can still realize accurate wave arrival time estimation.The application has been verified by simulation experiment, and the experimental results show that the algorithm has good ability to distinguish multipath for direct wave non-strong multipath channel, and accurate underwater acoustic signal wave arrival time can be obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of underwater acoustic signal time of arrival estimation, and in particular to a method and system for underwater acoustic signal time of arrival estimation based on multi-feature fusion. BACKGROUND

[0002] Underwater acoustic signal detection has a wide range of needs, such as navigation and positioning, point measurement, network communication, etc. In a sonar system, target distance information is obtained by the time difference between the transmitted signal and the received signal, while the detected signal will be used to locate the accurate time of arrival and obtain accurate target distance information, which has high requirements for accurate detection of the probe signal and accurate estimation of the time of arrival. The challenge lies in the false alarm and time of arrival estimation error caused by the strong environmental noise in the underwater environment, channel time variation and multipath effect [1] .

[0003] Traditional time of arrival estimation algorithms usually include two stages: signal detection and time of arrival estimation. The goal of the signal detection stage is to obtain a rough time of arrival and instruct the processor to perform fine time of arrival estimation. In an active sonar system, this task is usually accomplished by combining time-domain constant false alarm rate (CFAR) and matched filter (MF) [2] . CFAR determines the detection threshold according to the signal detection false alarm rate. Combined with the prior knowledge of the replica, MF can obtain more accurate time-domain correlation peaks. However, this algorithm is prone to mismatch under the non-strongest direct arrival channel impulse response [3]. In order to solve this problem, many improved CFAR algorithms [4]，[5] and adaptive MF algorithms [6]，[7] have been proposed, but these algorithms require prior knowledge of the channel and noise, and lack universality and stability in time-varying channels.

[0004] Invention patent application CN112183582A proposes a multi-feature fusion underwater target recognition method, which uses short-time energy and GFCC features under EMD as the input of the network model, and uses a CNN-LSTM integrated time series network to complete target recognition. Since the underwater acoustic signal processing method based on neural network model requires a large number of training samples, and can only be stably executed in a specific environment. For underwater environments with severe time-varying characteristics, it has high algorithm complexity and engineering implementation difficulty. Therefore, there is an urgent need for a universal and stable algorithm that can easily implement and obtain accurate time of arrival under non-strongest direct wave underwater acoustic channels. SUMMARY

[0005] The technical problem to be solved by the present invention is to provide a method and system for estimating the arrival time of underwater acoustic signals based on multi-feature fusion in response to the shortcomings of the existing technology, so as to solve the problems of detection mismatch of matched filtering technology in scenarios where the direct wave is not the strongest and uncertainty of a single feature in estimating the arrival time of underwater acoustic signals, thereby improving the accuracy of arrival time estimation.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is: a method for estimating the time of arrival of underwater acoustic signals based on multi-feature fusion, comprising the following steps:

[0007] S1, perform multi-level spectrum analysis on the received signal, preliminarily judge the arrival of the underwater acoustic signal through the frequency domain information and peak-to-average ratio of multiple frames of received signals, and obtain the signal segment ζ0 containing the arrival time;

[0008] S2, extracting the energy characteristics, spectral entropy characteristics, zero-crossing rate and delayed autocorrelation characteristics of the underwater acoustic signal in the signal segment ζ0;

[0009] S3, fusing the energy feature, spectral entropy feature, zero-crossing rate feature, and delayed autocorrelation feature to obtain a fused feature;

[0010] S4. Using the fused features, an endpoint detection method is used to achieve accurate estimation of the arrival time.

[0011] The present invention mainly solves the problem of estimating the arrival time of effective signals in channels where the direct wave is not the strongest. Since there are interferences such as impact noise in the underwater acoustic environment, the present invention uses multi-level spectrum analysis to detect signals. By setting the duration of each frame signal to be longer than the duration of the impact noise, the effective signal detection of multiple consecutive frames is used to avoid the interference of impact noise, and finally the time period where the arrival time is located is obtained. Since the phenomenon of non-strong direct waves is prone to exist in underwater acoustic channels, the traditional matched filtering method fails because it cannot distinguish non-strongest direct waves under single-frequency signals. In addition, the underwater acoustic environment has different fading phenomena for different signal features. The present invention analyzes different features and fuses them into one feature using certain rules. The fused feature has the advantage of optimizing and compensating for small fading features, so that it can well distinguish underwater acoustic signals where the direct wave is not the strongest, and achieve accurate arrival time estimation.

[0012] The specific implementation process of step S1 includes:

[0013] 1) Frame the received signal. Assume that the number of points per frame is N, and there are M frames of signal. The nth sampling point of the mth frame is represented by r m (n), where m = 0, 1, ..., M-1, n = 0, 1, ..., N-1;

[0014] 2) Perform spectrum analysis on the received signal of the mth frame, and the frequency domain information R of the received signal of the mth frame m(k) is:

[0015] 3) Calculate the spectrum line sound pressure level P of the received signal of the mth frame m (k): Among them, P m (k) is specifically the sound pressure level of the kth spectrum line of the received signal of the mth frame, f s is the sampling rate of the receiving end, G is the sum of the hydrophone sensitivity and the conditioning circuit gain;

[0016] 4) When the sound pressure level at the frequency of the valid signal of the continuous M0 frame exceeds the predetermined threshold value Td1, determine whether the peak-to-average ratio of the continuous signal of the M0 frame exceeds the threshold value Td2. If so, the signal is confirmed to have arrived; the peak-to-average ratio L(k0) of the continuous M0 frame signal is defined as: Among them, k0 is the frequency of the effective signal, P(k0) represents the sound pressure level of the continuous M0 frame underwater acoustic signal at the k0 frequency, P(k i ) represents the continuous M0 frames of underwater acoustic signal at k i Sound pressure level at frequency;

[0017] 5) Determine the time period ζ0 using the following formula: t0≤t<t0+M0T r ; Among them, t0 is the start time of the received signal, T r is the duration of a single frame of received signal, and t represents the time in the time period ζ0.

[0018] Step S1 of the present invention primarily employs spectrum analysis, primarily using Fast Fourier Transform technology. To combat noise interference such as underwater acoustic impact, the present invention utilizes spectrum analysis of multiple consecutive frames to achieve effective signal detection. This approach, compared to other traditional algorithms, offers lower computational complexity and ensures superior detection results with minimal computational complexity.

[0019] The specific implementation process of step S2 includes:

[0020] A) Calculate the power spectrum Y of the received signal of the mth frame m (k): N is the number of points per frame, R m (k) is the frequency domain information of the received signal of the mth frame;

[0021] B) Calculate the energy characteristic E of the received signal of the mth frame m : Among them, Y m (x) is the power spectrum of the received signal of the mth frame and the power spectrum of the xth spectral line; G is the sum of the hydrophone sensitivity and the conditioning circuit gain;

[0022] Calculate the normalized spectral probability density p of each frequency component of the m-th frame sampling signalm (k):

[0023]

[0024] C) Calculate the short-time entropy spectrum H of the received signal of the mth frame m : The short-time entropy spectrum is the spectral entropy feature;

[0025] Calculate the zero crossing rate ZCR of the received signal of the mth frame m :

[0026] sgn() is the symbol function, r m (v) is the vth sampling point of the received signal in the mth frame;

[0027] Calculate the delayed autocorrelation characteristic C when the received signal of the mth frame is delayed by one point m :

[0028]

[0029] Step S2 of the present invention mainly realizes the extraction of different signal features. Due to the randomness of the underwater acoustic channel, different underwater acoustic channels may have different fading characteristics for different features. The present invention realizes the detection of underwater acoustic channel fading through multiple features.

[0030] The specific implementation process of step S3 includes:

[0031] Calculate the weight w(i) of the i-th signal feature: in, represents the i-th signal feature in the ζ0 time period, ζ1 represents the noise period, represents the i-th signal feature of the noise segment, ζ2 represents the signal segment, represents the i-th signal feature of the signal segment, Represents the fitting slope of the i-th signal characteristic noise and the signal mutation segment, Represents the amplitude difference between the characteristic noise of the i-th signal and the signal mutation segment, represents the variance of the characteristic noise segment of the i-th signal, represents the variance of the signal segment of the i-th signal feature, and a represents a constant; wherein the signal feature refers to energy feature, spectral entropy feature, zero-crossing rate feature or delayed autocorrelation feature;

[0032] Use the following formula to obtain the fused feature criter m : E m 、H m 、ZCR m 、C mThey are energy feature, spectral entropy feature, zero-crossing rate feature and delayed autocorrelation feature respectively.

[0033] Step S3 of the present invention primarily describes the multi-feature fusion process. This method utilizes the aforementioned means to distinguish the dominant features within the current underwater acoustic environment, thereby selecting the most advantageous features and compensating for them with the remaining features. This method can better overcome the phenomenon of non-maximum direct waves in the underwater acoustic channel and achieve superior detection characteristics.

[0034] The specific implementation process of step S4 includes:

[0035] I) traverse the fused features in the collected signal segment in time sequence and initialize p=0;

[0036] II) Initialize the number of continuous valid underwater acoustic signals L = 1, and set the minimum number of continuous useful signal segments L min , initialize the possible arrival time point m0=1;

[0037] III) The value of p is increased by 1, if criter(p) <T1,其中criter(p)代表第p帧信号融合后的特征,则判断处于非有效信号段,返回步骤II);若T1<criter(p)<T2,则判断可能处于有效信号段,L的值加1,重复步骤III);若criter(p)> T2, it is determined to be in the useful signal segment and jump to IV); T1 and T2 are the set first threshold and second threshold respectively;

[0038] IV) Update m0 = max(pL-1, 1), add 1 to the value of L, and determine L≥L min Is it true? If so, determine that m0 is the arrival time of the useful signal; otherwise, return to step III).

[0039] Step S4 of the present invention mainly describes the endpoint detection algorithm in speech signal detection. The algorithm can be adjusted by L min , T1, T2 and other parameters can effectively avoid the influence of impact noise in the sampling signal, and can achieve the effect of more accurate detection of the arrival time.

[0040] The present invention also provides an underwater acoustic signal arrival time estimation system based on multi-feature fusion, which includes:

[0041] one or more processors;

[0042] A memory having one or more programs stored thereon, which, when executed by the one or more processors, enables the one or more processors to implement the steps of the above method of the present invention.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention utilizes the significant difference between signal fusion and noise fusion characteristics to achieve accurate arrival time estimation. This feature enables the algorithm to overcome the arrival time estimation error problem caused by non-strong direct waves in underwater acoustic environments. At the same time, compared with matched filtering technology, the present invention only requires a portion of the signal duration (specifically related to the sliding window length), shortening the algorithm execution time.

[0045] 2. Based on the current underwater acoustic environment conditions, the present invention analyzes the characteristics of energy, spectral entropy, zero-crossing rate, and delayed autocorrelation features, and assigns different weights according to their performance, so that the fused feature criteria have the ability to be updated as the underwater acoustic environment changes, and can bring together the advantages of different features. Compared with the arrival time estimation algorithm of a single feature, the accuracy of underwater acoustic signal detection is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of multi-level spectrum analysis according to an embodiment of the present invention;

[0047] Figure 2 This is a flow chart of a method for estimating the time of arrival of underwater acoustic signals based on multi-feature fusion according to an embodiment of the present invention;

[0048] Figure 3 This is a simulation diagram of matched filtering according to an embodiment of the present invention;

[0049] Figure 4 This is a simulation diagram of the underwater acoustic signal arrival time estimation method based on multi-feature fusion according to an embodiment of the present invention;

[0050] Figure 5 This is a simulation comparison diagram of an underwater acoustic signal arrival time estimation method based on multi-feature fusion and an underwater acoustic signal arrival time estimation method based on a single feature according to an embodiment of the present invention; DETAILED DESCRIPTION

[0051] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0052] The following combination Figure 1 and Figure 2 An embodiment of the present invention is described.

[0053] In the embodiment of the present application, the underwater acoustic signal is collected to a signal processing machine by using digital-to-analog conversion hardware, and the following estimation method is executed by the signal processing machine.

[0054] The embodiment of the present application provides an underwater acoustic signal wave arrival time estimation method based on multi-feature fusion, comprising the following steps.

[0055] (1) Multi-stage spectrum analysis is performed on the received signal, the arrival of the underwater acoustic signal is determined through the frequency domain information of the continuous multiple frames of received signals exceeding a threshold Td1 and the peak-to-average ratio of the continuous multiple frames of received signals exceeding a threshold Td2, and a signal segment ζ0 containing the wave arrival time is obtained, wherein the thresholds Td1 and Td2 are mainly set according to the actual underwater acoustic environment, and are generally set as Td1 = 110 and Td2 = 0.

[0056] (2) Single feature extraction is performed on the signal segment ζ0. The energy feature, spectral entropy feature, zero-crossing rate feature and delay autocorrelation feature of the underwater acoustic signal in the signal segment ζ0 in (1) are extracted.

[0057] (3) According to the underwater acoustic environment, the advantages and disadvantages of the single feature in (2) in the current environment are judged by using a specific index, and the normalization weight of the four features is given, and finally the multiple features are fused into a new feature (i.e. the fused feature).

[0058] (4) According to the new feature criterion, the wave arrival time is accurately estimated by using an endpoint detection algorithm.

[0059] In the embodiment of the present application, in step (1), the basic principle of multi-stage spectrum analysis is: mainly using the frequency domain information of the underwater acoustic effective signal frequency which is obviously larger than the amplitude of the noise at the frequency to realize signal detection, in order to prevent the interference of impact noise, the frequency domain of the continuous M0 frames of signals is used as a criterion in the embodiment of the present application, which greatly improves the accuracy of rough signal detection.

[0060] S1, frame processing is performed on the received signal, assuming that the number of points of each frame is N, and the received signal is divided into M frames, and each frame sampling signal is represented as r m (n), that is, r m ,m=0,2,…,M-1.

[0061] S2, spectrum analysis is performed on the mth frame of received signal, and the frequency domain information R m (k) is:

[0062]

[0063] S3, the spectral line sound pressure level of the mth frame of received signal is calculated:

[0064]

[0065] In the formula, f sis the sampling rate of the receiver, and G is the sum of the hydrophone sensitivity and the conditioning circuit gain.

[0066] S4. When the sound pressure level at the frequency of the continuous M0 frame valid signal exceeds the predetermined threshold value Td1, and the peak-to-average ratio of the continuous M0 frame signal exceeds the threshold value Td2, the signal is confirmed to have arrived. The peak-to-average ratio of the continuous M0 frame signal is defined as:

[0067]

[0068] Where k0 is the frequency of the effective signal, P(k i ) represents the continuous M0 frames of underwater acoustic signal at k i The sound pressure level at the frequency. Then the time period ζ0 is:

[0069] t0≤t<t0+M0T r (4)

[0070] Where t0 is the start time of the received signal, T r The duration of a single frame of received signal.

[0071] In step (2), a single feature of the underwater acoustic signal is extracted. Although the frequency, amplitude, etc. of the underwater acoustic signal may be distorted after passing through the underwater acoustic channel, it is found from experiments that there is still a certain feature in the signal that is relatively easy to distinguish between signal and noise. Then, feature extraction is performed on the sampled signal:

[0072] S1. Calculate the power spectrum of the received signal of the mth frame:

[0073]

[0074] S2. Extract the energy characteristics of the underwater acoustic signal. Calculate the energy characteristics of the received signal of the mth frame:

[0075]

[0076] S3, calculate the normalized spectral probability density p of each frequency component of the m-th frame sampling signal m (k):

[0077]

[0078] S4. Extract the short-time spectrum entropy characteristics of the underwater acoustic signal. Calculate the short-time entropy spectrum of the m-th frame received signal:

[0079]

[0080] S5. Extract the zero-crossing rate feature of the underwater acoustic signal. Calculate the zero-crossing rate of the received signal of the mth frame:

[0081]

[0082] Where sgn is the sign function, that is:

[0083]

[0084] S6. Extract the delayed autocorrelation characteristics of the underwater acoustic signal. Calculate the delayed autocorrelation of the received signal of the mth frame when it is delayed by one point:

[0085]

[0086] In step (3), multi-feature fusion is performed. The fading of signal features in the time-varying underwater acoustic channel is also random, so the time of arrival estimation using a single feature has the characteristics of poor estimation effect and instability. Based on this situation, the embodiment of the present invention uses the mathematical characteristics such as the slope and variance of each feature under the sampled signal to judge the quality of the feature, and then determines its feature weight to complete the multi-feature fusion operation.

[0087] S1. Determination of weights of different features:

[0088]

[0089] Where ζ1 represents the noise segment, which is a certain time before ζ0, and ζ2 represents the signal segment, which is a certain time after ζ0. Represents the fitting slope of the characteristic noise of the i-th signal and the signal mutation segment, Represents the amplitude difference between the characteristic noise of the i-th signal and the signal mutation segment, represents the variance of the characteristic noise segment of the i-th signal, The variance of the characteristic signal segment of the i-th signal, a represents a constant.

[0090] S2. Multi-feature fusion, new criterion for the m-th frame signal:

[0091]

[0092] In step (4), endpoint detection technology is combined to achieve arrival time estimation based on multi-feature fusion.

[0093] S1: Perform endpoint detection, traverse the fused features in the acquired signal segment in time sequence, and initialize p = 0;

[0094] S2: Initialize the number of continuous valid underwater acoustic signals L = 1, set the minimum number of continuous useful signal segments L min , initialize the possible arrival time point m0=1;

[0095] S3: The value of p is increased by 1, if criter(p) <T1,其中criter(p)代表第p帧信号的新判据(即融合后的特征),则判断处于非有效信号段,返回步骤S2;若T1<criter(p)<T2,则判断可能处于有效信号段,L的值加1,重复步骤S3;若criter(p)> If T2 is reached, it is determined to be in the useful signal segment and the process jumps to S4; T1 and T2 are the set first threshold and second threshold respectively;

[0096] S4: Update m0 = max(pL-1, 1), add 1 to the value of L, and determine L≥L min Is it true? If so, m0 is the arrival time of the useful signal; otherwise, return to step S3.

[0097] From the above analysis, we can see that the present invention selects the optimal feature based on the current underwater acoustic environment and uses the remaining features as compensation. By using multiple features to distinguish between noise and signal in environments where the direct wave is not the strongest, a more accurate arrival time can be obtained.

[0098] The effectiveness of the method of the embodiment of the present invention is verified by simulation experiments below.

[0099] Table 1 Simulation channel parameters

[0100]

[0101] Simulation channel parameters such as

[0102] As shown in the table, this channel is a multipath channel where the direct wave is not the strongest, and the arrival times of each path are 1.2s, 1.8s, and 2.2s respectively. Figure 3 The channel simulation results of the matched filter and the embodiment of the present invention are shown. The matched filter algorithm estimates the arrival time by detecting the maximum peak. Therefore, the matched filter will use 1.8s as the arrival time and mistakenly regard the secondary arrival as the direct arrival. In addition, it is difficult to distinguish different paths from the matched filter results. However, from Figure 4 It can be seen that the fusion characteristics of the signal and the noise are significantly different. Furthermore, the fusion characteristics of different paths also differ significantly. In this case, the embodiments of the present invention, combined with the endpoint detection algorithm, can overcome the problem of the direct wave not being the strongest, and obtain more accurate time of arrival estimation results.

[0103] Figure 5The simulation results of the embodiment of the present invention and the arrival time estimation algorithm based on a single feature are shown. The results of the single-feature arrival time estimation algorithm vary greatly with the change of the signal-to-noise ratio. However, the embodiment of the present invention varies stably within the set signal-to-noise ratio range. This is because the embodiment of the present invention can optimize the arrival time estimation by extracting the advantages of different features through weights. However, the degree of optimization of the algorithm has a certain tolerance. When the signal-to-noise ratio is between [2dB, 7dB], the RMSE of the arrival time estimation algorithm based on delayed autocorrelation is the smallest. The embodiment of the present invention cannot completely approach the optimal value. This is because, considering stability, the embodiment of the present invention will not give up the influence of other features, and will also give other disadvantageous features a very small weight. In addition, the experimental results show that the embodiment of the present invention has a low error level. It is worth noting that the error of the embodiment of the present invention is the smallest in the signal-to-noise ratio range of [-4dB, 2dB] and greater than 7dB.

[0104] in addition Figure 5 The embodiment of the present invention is compared with the time of arrival estimation algorithm based on energy entropy ratio. The time of arrival estimation algorithm based on energy entropy ratio combines the two features of energy and spectral entropy by using a simple ratio. Figure 5 It is known that across the entire SNR range, the embodiment of the present invention outperforms the energy-entropy ratio-based time-of-arrival estimation algorithm. This is because, unlike the embodiment of the present invention, the energy-entropy ratio combines features using fixed weights, whose weights do not change with the underwater acoustic environment. The embodiment of the present invention, based on measurements of the specific environment, evaluates the strengths and weaknesses of different features using specific criteria and dynamically assigns them weights. This eliminates features that compromise the algorithm's performance while complementing features that offer advantages.

[0105] Another embodiment of the present invention provides an estimation system corresponding to the method of the above embodiment, which includes a terminal device, which can be a processing device for a client, such as a mobile phone, laptop computer, tablet computer, desktop computer, etc., to execute the method of the above embodiment.

[0106] The terminal device of this embodiment includes a memory, a processor, and a computer program stored in the memory; the processor executes the computer program in the memory to implement the steps of the method of the above embodiment.

[0107] In some implementations, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage.

[0108] In other implementations, the processor may be a central processing unit (CPU), a digital signal processor (DSP), or other general-purpose processors, which are not limited herein.

[0109] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal translation scripting language JavaScript, etc.

[0110] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0112] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0113] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

[0114] References

[0115] [1]Wang Z,Member S,Zhou S,et al.Clustered Adaptation for Estimationof Time-Varying Underwater Acoustic Channels[J].IEEE TRANSACTIONS ON SIGNALPROCESSING,2013,60(6).

[0116] [2]Robey F C,Fuhrmann D R,Kelly E J,et al.A CFAR adaptive matchedfilter detector[J].IEEE Transactions on Aerospace and Electronic Systems,1992,28(1):208-216.

[0117] [3]Tan H P,Diamant R,Seah W,et al.A survey of techniques andchallenges in underwater localization[J].Ocean Engineering,2011,38(14-15):1663-1676.

[0118] [4]Acosta G G,Villar S / .Accumulated CA-CFAR Process in 2-D for OnlineObject Detection From Sidescan Sonar Data[J].IEEE Journal of OceanicEngineering,2015,40(3):558-569.

[0119] [5]Wang L,Wang D,Hao C.Intelligent CFAR Detector Based on SupportVector Machine[J].IEEE Access,2017,5:26965-26972.

[0120] [6] Diamant, Roee. Closed Form Analysis of the Normalized Matched Filter With a Test Case for Detection of Underwater Acoustic Signals [J]. IEEE Access, 2016: 1-1.

[0121] [7] Jay S, Guillaume M, Blanc-Talon J. Underwater Target Detection With Hyperspectral Data: Solutions for Both Known and Unknown Water Quality [J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2012, 5(4): 1213-1221.

Claims

1. A method for estimating the arrival time of an underwater acoustic signal based on multi - feature fusion, characterized in that: It includes the following steps: S1. Perform multi - level spectrum analysis on the received signal, and preliminarily judge the arrival of the underwater acoustic signal through the frequency - domain information and peak - to - average ratio of consecutive multi - frame received signals, and obtain a signal segment ζ0 containing the arrival time; S2. Extract the energy feature, spectral entropy feature, zero - crossing rate and delayed autocorrelation feature of the underwater acoustic signal in the signal segment ζ0; S3. Fuse the energy feature, spectral entropy feature, zero - crossing rate feature and delayed autocorrelation feature to obtain a fused feature; S4. Use the fused feature and adopt an endpoint detection method to achieve accurate estimation of the arrival time; The specific implementation process of step S1 includes: 1) Frame the received signal. Assume that the number of points per frame is N, and there are M frames of signal. The nth sampling point of the mth frame is represented by r m (n), where m = 0, 1, ..., M-1, n = 0, 1, ..., N-1; 2) Perform spectrum analysis on the received signal of the mth frame, and the frequency domain information R of the received signal of the mth frame m (k) is: N is the number of points per frame; 3) Calculate the spectrum line sound pressure level P of the received signal of the mth frame m (k): Among them, P m (k) is specifically the sound pressure level of the kth spectrum line of the received signal of the mth frame, f s is the sampling rate of the receiving end, G is the sum of the hydrophone sensitivity and the conditioning circuit gain; 4) When the sound pressure level at the frequency of the valid signal of the continuous M0 frame exceeds the predetermined threshold value Td1, determine whether the peak-to-average ratio of the continuous signal of the M0 frame exceeds the threshold value Td2. If so, the signal is confirmed to have arrived; the peak-to-average ratio L(k0) of the continuous M0 frame signal is defined as: Among them, k0 is the frequency of the effective signal, P(k0) represents the sound pressure level of the continuous M0 frame underwater acoustic signal at the k0 frequency, P(k i ) represents the continuous M0 frames of underwater acoustic signal at k i Sound pressure level at frequency; 5) Determine the time period ζ0 using the following formula: t0≤t <t0+M0T r ; Among them, t0 is the start time of the received signal, T r is the duration of a single frame of received signal, t represents the time in time period ζ0; The specific implementation process of step S4 includes: I) Traverse the fused features in the collected signal segment according to the time sequence, and initialize p = 0; II) Initialize the number of continuous valid underwater acoustic signals L = 1, and set the minimum number of continuous useful signal segments L min , initialize the possible arrival time point m0=1; III) Increment the value of p. If criter(p) < T1, where criter(p) represents the fused feature of the p - th frame signal, it is judged that it is in a non - effective signal segment, and return to step II); If T1 < criter(p) < T2, it is judged that it may be in an effective signal segment, increment the value of L by 1, and repeat step III); If criter(p) > T2, it is judged that it is in a useful signal segment, and jump to IV); T1 and T2 are the set first threshold and second threshold respectively; IV) Update m0 = max(pL-1, 1), add 1 to the value of L, and determine L≥L min Is it true? If so, determine that m0 is the arrival time of the useful signal; otherwise, return to step III).

2. The method for estimating the time of arrival of underwater acoustic signals based on multi-feature fusion according to claim 1 is characterized in that: The specific implementation process of step S2 includes: A) Calculate the power spectrum Y of the received signal of the mth frame m (k): R m (k) is the frequency domain information of the received signal of the mth frame; B) Calculate the energy characteristic E of the received signal of the mth frame m : Among them, Y m (x) is the power spectrum of the received signal of the mth frame and the power spectrum of the xth spectral line; G is the sum of the hydrophone sensitivity and the conditioning circuit gain; Calculate the normalized spectral probability density p of each frequency component of the m-th frame sampling signal m (k): C) Calculate the short-time entropy spectrum H of the received signal of the mth frame m : The short-time entropy spectrum is the spectral entropy feature; Calculate the zero crossing rate ZCR of the received signal of the mth frame m : sgn() is the symbol function, r m (v) is the vth sampling point of the received signal in the mth frame; Calculate the delayed autocorrelation characteristic C when the received signal of the mth frame is delayed by one point m :

3. The method for estimating the time of arrival of underwater acoustic signals based on multi-feature fusion according to claim 1 is characterized in that: The specific implementation process of step S3 includes: Calculate the weight w(i) of the i-th signal feature: in, represents the i-th signal feature in the ζ0 time period, ζ1 represents the noise period, represents the i-th signal feature of the noise segment ζ1, ζ2 represents the signal segment, represents the i-th signal feature of signal segment ζ2, Represents the fitting slope of the characteristic noise of the i-th signal and the signal mutation segment, Represents the amplitude difference between the characteristic noise of the i-th signal and the signal mutation segment, represents the variance of the characteristic noise segment of the i-th signal, represents the variance of the signal segment of the i-th signal feature, and a represents a constant; wherein the signal feature refers to energy feature, spectral entropy feature, zero-crossing rate feature or delayed autocorrelation feature; Use the following formula to obtain the fused feature criter m : E m 、H m 、ZCR m 、C m They are energy feature, spectral entropy feature, zero-crossing rate feature and delayed autocorrelation feature respectively.

4. A system for estimating the time of arrival of underwater acoustic signals based on multi-feature fusion, characterized in that: It includes: One or more processors; A memory, on which one or more programs are stored. When the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the method according to any one of claims 1 to 3.

Images