A method for active sonar pulse blind detection

Abstract

The present application relates to underwater acoustic signal processing technology field, especially to a kind of active sonar transmitting pulse blind detection method.The present application includes the following steps: extracting a section of active sonar echo data, the echo data contains multiple scanning periods, and the echo data contains transmitting pulse signal and target echo signal;Time-frequency analysis is carried out on echo data by short-time Fourier transform, to obtain the spectral distribution of segmented sequence;All echo signals are selected using the difference between signal and noise in time-frequency characteristics;Eliminate target echo, and retain transmitting pulse direct wave;Short-time Fourier transform is carried out on each signal screened out, to obtain the spectral sequence of signal, and finally the signal parameters of all transmitting pulse signals screened out are calculated.The present application provides a kind of active sonar transmitting pulse blind detection method, so that active beam data without transmitting parameter information can also be subjected to matched filter processing, the signal-to-noise ratio of echo signal is improved, so that the detection probability of target is improved.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic signal processing technology, and is used for active sonar transmit pulse detection without any prior information. It obtains information such as the frequency, bandwidth, pulse width, and transmission period of the transmitted pulse, so that the received signal can be further processed by matched filtering, target detection and recognition, etc. In particular, it relates to a blind detection method for active sonar transmit pulses. Background Technology

[0002] Active sonar is an important underwater detection device. It determines the target's position, velocity, and size by emitting specific types of sound waves and then receiving and analyzing the reflected echoes. Active sonar generally uses matched filtering to improve the signal-to-noise ratio of the echo signal. However, matched filtering algorithms rely on complete prior knowledge of the target signal. In certain situations, prior information about the emitted signal is unavailable. For example, the sonar system itself may not have packaged the emission parameters with the echo data for storage, or the emission parameters stored during sonar recording may be incorrect, or some sonars may lack recording equipment and rely on a recorder for data storage. This data lacks emission parameters, making it impossible to obtain gain through matched filtering; only simple energy detection is possible, significantly reducing the target detection probability. While research on active sonar signal processing in China is quite advanced, there is a lack of specific research on active sonar signal processing without emission parameters. In reality, a large amount of active sonar data without emission parameters remains to be analyzed.

[0003] Therefore, there is an urgent need to find a precise and fast method to extract transmitted signal parameters from active sonar echo data, so that these data without prior information can also be processed by matched filtering to improve the signal-to-noise ratio of the echo signal, thereby increasing the probability of target detection. Summary of the Invention

[0004] The purpose of this invention is to provide a blind detection method for active sonar transmitted pulses. Since the raw signals received by active sonar generally have a low signal-to-noise ratio, the target cannot be directly detected by energy detection or signal-to-noise ratio detection. The first problem to be solved by this invention is how to detect all target echoes from active sonar echo signals with low signal-to-noise ratio. Furthermore, it is also necessary to distinguish between real target echoes and transmitted pulse direct waves from all target echoes. Finally, the transmitted pulse frequency, bandwidth, pulse width and other parameters are correctly extracted from the selected transmitted pulse direct waves.

[0005] To address the aforementioned technical problems, this invention provides a method for blind detection of active sonar emission pulses, comprising the following steps:

[0006] Step S1: Extract a segment of active sonar echo data, which contains multiple scan cycles and includes the transmitted pulse signal and the target echo signal.

[0007] Step S2: Perform time-frequency analysis on the echo data using short-time Fourier transform to obtain the spectral distribution of the segmented sequence;

[0008] Step S3: Select all echo signals by utilizing the differences in time-frequency characteristics between the signal and noise;

[0009] Step S4: Eliminate the target echo and retain the direct wave of the transmitted pulse;

[0010] Step S5: Perform a short-time Fourier transform on each selected signal segment, and use the signal spectrum data to calculate the signal parameters of all selected transmitted pulse signals.

[0011] Preferably, step S2 takes the following specific steps:

[0012] Step S2-1: First, decompose the signal into a series of small segments of length L, assuming a total of N segments. Then, window each segment of the signal:

[0013] y(t) = x(t)·ω(t-τ0);

[0014] Where ω(t) is the window function;

[0015] Step S2-2: Perform a Fourier transform on the windowed signal;

[0016]

[0017] This yields the spectral distribution S(ω,τ) of the segmented sequence.

[0018] Preferably, step S3 takes the following specific steps:

[0019] S3-1: First find the maximum value of S(ω,τ) in each segment, and record the position of the maximum value in each segment as Index(i), i=1,...,N;

[0020] S3-2: Then find the segments where Index(i) changes continuously for more than a certain number of times, and record the starting position Start1(i), i = 1, ..., M and the ending position End1(i), i = 1, ..., M of these segments;

[0021] S3-3: Check the selected start positions one by one. If the distance difference between two positions is less than the maximum pulse width range, they are considered to be the same pulse. The start point is the start point of the first one, and the end point is the end point of the second one. The start position of all selected data segments is denoted as Start2(i), i = 1, ..., P, and the end position is denoted as End2(i), i = 1, ..., P.

[0022] Preferably, step S4 takes the following specific steps:

[0023] S4-1: Calculate the mean amplitude of the first and last T points of the selected data segment, denoted as SigL(i) and SigR(i), i = 1, ..., P, where:

[0024]

[0025]

[0026] S4-2: Calculate the amplitude difference between the preceding and following data based on the average amplitude of the preceding and following T points.

[0027] delta(i) = 20 * log 10 (SigR(i) / SigL(i)),i=1,...,P;

[0028] Echoes with a difference of less than 3dB are considered target echoes and removed from the sequence; the selected transmitted pulse signals are reorganized, with the starting position denoted as Start3(i), i = 1, ..., Q, and the ending position denoted as End3(i), i = 1, ..., Q.

[0029] Preferably, in step S5, the signal spectrum sequence is S(i,j,k), i=1,...,Q; j=1,...N,k=1,....,K, where N is the number of FFT points and K is the number of signal segments; the modulus value is calculated for the spectrum result of each segment k=1,...K of each signal segment, and the position corresponding to the maximum modulus value in the j=1,...,N sequence is taken as max_index(i,k). Assuming the signal sampling rate Fs, the signal parameters of all transmitted pulse signals can be calculated based on the above data; wherein the signal parameters of the transmitted pulse signals include the signal center frequency, signal bandwidth, signal pulse width, and transmission period.

[0030] Preferably, the center frequency of the signal is calculated using the following formula:

[0031]

[0032] Preferably, the signal bandwidth is calculated using the following formula:

[0033] B(i)=[max_index(i,K)-max_index(i,1)]*Fs / N,i=1,...,Q.

[0034] Preferably, the signal pulse width is calculated using the following formula:

[0035] T(i)=[End3(i)-Start3(i)] / Fs, i=1,...,Q.

[0036] Preferably, the emission period is calculated using the following formula:

[0037] R(i)=[Start3(i+1)-Start3(i)] / Fs, i=1,...,Q-1.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1. This invention utilizes the differences in time-frequency characteristics between signals and noise to establish a signal coarse screening model. Compared to traditional energy detection or signal-to-noise ratio detection models, it reduces the requirements for signal-to-noise ratio, is more suitable for detecting raw signals, and has low computational complexity and strong robustness.

[0040] 2. This invention utilizes the reverberation difference between the transmitted pulse direct wave and the target echo to establish a transmission pulse fine-screening model. This model breaks through the traditional limitation of distinguishing different types of targets based on the characteristics of the signal itself, and creatively distinguishes them by the difference characteristics of the adjacent data of the transmitted pulse direct wave and the target echo. This method has low computational cost and is stable and easy to implement. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the original signal of an active sonar with unknown transmission signal parameters, representing a segment of tracking beam data.

[0042] Figure 2 For the present invention Figure 1 A schematic diagram of the time-frequency results after performing a short-time Fourier transform on the data.

[0043] Figure 3 This is a schematic diagram of the emission pulse detection results of the present invention.

[0044] Figure 4 This is a flowchart illustrating the principle of the present invention. Detailed Implementation

[0045] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer from the following description. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0046] like Figure 1-4 As shown, this embodiment of the invention provides a method for blind detection of active sonar emission pulses, comprising the following steps:

[0047] Step 1: Extract a segment of the active sonar echo signal, which contains multiple scan cycles, such as... Figure 1 As shown, the data contains both transmitted pulse signals and target echo signals.

[0048] Step 2: Perform time-frequency analysis on the echo data using short-time Fourier transform.

[0049] (1) First, the signal is decomposed into a series of small segments of length L, assuming a total of N segments. Windowing is then applied to each segment:

[0050] y(t) = x(t)·ω(t-τ0);

[0051] Where ω(t) is the window function;

[0052] (2) Perform a Fourier transform on the windowed signal;

[0053]

[0054] This yields the spectral distribution S(ω,τ) of the segmented sequence.

[0055] Step 3: Select all echo signals by utilizing the differences in time-frequency characteristics between the signal and noise. After performing a short-time Fourier transform on the original signal, the location of the maximum frequency point of the signal within the continuous segmented data changes continuously, while the noise is chaotic.

[0056] (1) First find the maximum value of S(ω,τ) in each segment, and record the position of the maximum value in each segment as Index(i), i=1,...,N.

[0057] (2) Then find the segments where Index(i) changes continuously for more than a certain number of times, and record the starting position Start1(i), i = 1, ..., M and the ending position End1(i), i = 1, ..., M of these segments.

[0058] (3) Check the selected start positions one by one. If the distance difference between two positions is less than the maximum pulse width range, they are considered to be the same pulse. The start point is the start point of the first one, and the end point is the end point of the second one. The start position of all selected data segments is denoted as Start2(i), i = 1,...,P, and the end position is denoted as End2(i), i = 1,...,P.

[0059] Step 4: Eliminate the target echo and retain the direct wave of the transmitted pulse. The data before the transmitted pulse is the tail end of the previous range, and the amplitude is generally small. After the transmitted pulse, there is reverberation, and the amplitude is larger. The amplitude difference between the beginning and end of the target echo is not large. Based on this characteristic, the target echo is eliminated.

[0060] (1) Calculate the mean amplitude of the T points before and after the selected data segment, denoted as SigL(i) and SigR(i), i = 1, ..., P, where:

[0061]

[0062]

[0063] (2) Calculate the amplitude difference between the data before and after based on the average amplitude of the T points before and after.

[0064] delta(i) = 20 * log 10 (SigR(i) / SigL(i)),i=1,...,P;

[0065] Echoes with a difference of less than 3dB are considered target echoes and removed from the sequence. The selected transmitted pulse signals are then reorganized, with the starting position denoted as Start3(i), i = 1, ..., Q, and the ending position denoted as End3(i), i = 1, ..., Q.

[0066] Step 5: Perform a Short-Time Fourier Transform (SFT) on each selected signal segment. Similar to Step 2, obtain the signal's spectral sequence S(i,j,k), i=1,...,Q; j=1,...N, k=1,....,K, where N is the number of FFT points and K is the number of signal segments. Calculate the modulus of the spectral result for each segment k=1,...K of each signal segment, and take the position corresponding to the maximum modulus in the j=1,...,N sequence, denoted as max_index(i,k). Assuming the signal sampling rate Fs, calculate the signal parameters of all selected transmitted pulse signals.

[0067] The formula for calculating the center frequency of the aforementioned signal is as follows:

[0068]

[0069] The formula for calculating signal bandwidth is as follows:

[0070] B(i)=[max_index(i,K)-max_index(i,1)]*Fs / N,i=1,...,Q;

[0071] The formula for calculating the signal pulse width is as follows:

[0072] T(i)=[End3(i)-Start3(i)] / Fs, i=1,...,Q;

[0073] The formula for calculating the launch period is as follows:

[0074] R(i)=[Start3(i+1)-Start3(i)] / Fs, i=1,...,Q-1.

[0075] The above description is merely a description of preferred embodiments of the present invention and is not intended to limit the scope of the present invention in any way. Any changes or modifications made by those skilled in the art based on the above disclosure shall fall within the protection scope of the claims.

Claims

1. A method for blind detection of active sonar emission pulses, characterized in that, Includes the following steps: Step S1: Extract a segment of active sonar echo data, which contains multiple scan cycles and includes the transmitted pulse signal and the target echo signal. Step S2: Perform time-frequency analysis on the echo data using short-time Fourier transform to obtain the spectral distribution of the segmented sequence; Step S3: Select all echo signals by utilizing the differences in time-frequency characteristics between the signal and noise; Step S4: Eliminate the target echo and retain the direct wave of the transmitted pulse; Step S5: Perform a short-time Fourier transform on each of the selected signal segments, and use the signal spectrum information to calculate the signal parameters of all the selected transmitted pulse signals; Step S2 involves the following specific steps: Step S2-1: First, decompose the signal into a series of small segments of length L, assuming a total of N segments. Then, window each segment of the signal: ； in, It is a window function; Step S2-2: Perform a Fourier transform on the windowed signal; ； This yields the spectral distribution of the segmented sequence. ; Step S3 involves the following specific steps: S3-1: First find each segment The maximum value is recorded as the position of the maximum value in each segment. ; S3-2: Then search For segments that change continuously more than a certain number of times, determine the starting positions of these segments. and end position Write it down; S3-3: Detect each selected starting position. If the distance difference between two positions is less than the maximum pulse width range, they are considered to be the same pulse. The starting point is the first starting point, and the ending point is the second ending point. The starting positions of all selected data segments are denoted as follows: The end position is denoted as ; Step S4 involves the following specific steps: S4-1: Calculate the mean amplitude of the T points before and after the selected data segment, denoted as... and ,in: ； ； S4-2: Calculate the amplitude difference between the preceding and following data based on the average amplitude of the preceding and following T points. ； Echoes with a difference of less than 3dB are considered target echoes and removed from the sequence; the selected transmitted pulse signals are then reorganized, with the starting position recorded as... The end position is denoted as .

2. The active sonar emission pulse blind detection method as described in claim 1, characterized in that, The spectral sequence of the signal in step S5 is as follows: ,in For FFT points, The number of signal segments; for each segment of each signal segment Calculate the modulus value from the spectral results, and take the maximum modulus value. The corresponding position in the sequence is denoted as Assuming the signal sampling rate Based on the above data, the signal parameters of all transmitted pulse signals can be calculated; The signal parameters for transmitting pulse signals include the signal center frequency, signal bandwidth, signal pulse width, and transmission period.

3. The active sonar emission pulse blind detection method as described in claim 2, characterized in that, The center frequency of the signal is calculated using the following formula: 。 4. The active sonar emission pulse blind detection method as described in claim 2, characterized in that, The signal bandwidth is calculated using the following formula: 。 5. The active sonar emission pulse blind detection method as described in claim 2, characterized in that, The signal pulse width is calculated using the following formula: 。 6. The active sonar emission pulse blind detection method as described in claim 2, characterized in that, The launch period is calculated using the following formula: 。

Images