A method and related system for real-time detection of underwater targets

Abstract

The application discloses a kind of real-time detection method and related system of target in water, by processing and obtaining different detection quantities under different physical fields to the background noise original signal of multiple physical field multiple channels, according to the probability distribution calculation to different detection quantities, according to the threshold of false alarm rate set according to the calculation result, to obtain the verification standard of signal detection under different physical fields, then by optimal threshold fusion criterion single physical field multiple channel fusion detection result and basic probability distribution function are obtained;Based on the same time dimension sliding window, real-time, synchronously to the small signal under multiple physical field multiple channels is analyzed, the advantage of different physical fields, different characteristics is concentrated at the same time, shorten the algorithm execution length, improve the reliability and real-time of underwater target detection, as final detection result and real-time update, realize the full use of sound / flow / electricity multiple physical field multiple channel information, reduce false alarm, reduce detection situation, improve detection accuracy.

Description

Technical Field

[0001] This invention belongs to the field of underwater target detection, and is based on real-time underwater target detection based on acoustic / fluid / electrical decision fusion. Specifically, it relates to a real-time underwater target detection method and related systems. Background Technology

[0002] Target detection technology is a crucial technique for target detection in the ocean. Only by accurately detecting underwater targets can subsequent tasks such as parameter estimation and target identification be completed. The accuracy of target detection directly determines the efficiency and accuracy of ocean exploration.

[0003] Underwater targets possess multi-physics characteristics, such as acoustic, flow, and electric field properties. Therefore, underwater target signals can be detected based on acoustic, flow, and electric field measurements to analyze target characteristics and achieve target detection. However, information obtained from a single physical field is only partial and incomplete, susceptible to the influence of equipment quality, performance, and environmental noise. The data is often incomplete, carries significant uncertainty, and may even contain errors. For acoustic field detection, the detection range is relatively long, and the target signal is typically the radiated noise of ships, submarines, and torpedoes. However, current acoustic stealth technologies worldwide have reached advanced levels, making it prone to false alarms or missed detections, leading to misjudgments of the battlefield situation. For electric field detection, the target characteristics are more pronounced than those of acoustic field detection, typically the axial frequency electric field or power frequency electric field of underwater vehicles, but its detection range is shorter than that of acoustic field detection. For flow field detection, the detection range is relatively short, but it has high accuracy in near-field target detection, making it suitable for precise detection of targets such as those used by divers. Current underwater target detection methods cannot effectively detect underwater targets by utilizing the characteristic information of different physical fields. Existing methods have low accuracy in underwater target detection and are susceptible to false alarms due to the influence of different environmental fields, which reduces the reliability of underwater target detection. Summary of the Invention

[0004] The purpose of this invention is to provide a real-time underwater target detection method and related system to overcome the problems of low detection accuracy and low reliability in existing real-time underwater target detection based on a single physical field.

[0005] A method for real-time detection of underwater targets includes the following steps:

[0006] Based on the acoustic / flow / electric composite detection unit, the background noise raw signals of multiple physical fields and multiple channels are collected. According to the set sliding time window and step size, the background noise raw signals of each channel of the acoustic field, flow field and electric field are extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel.

[0007] For the sound field background noise signal, the time-domain energy detection quantity and EMD energy entropy detection quantity of each frame signal under each channel are calculated respectively; for the flow field background noise signal, the Hilbert energy detection quantity of each frame signal under each channel is calculated respectively; for the electric field background noise signal, a fast Fourier transform is performed on each frame signal under each channel to extract the characteristic frequency amplitude detection quantity of the electric field detection target.

[0008] Statistical analysis was performed on all the detection quantities obtained above, and a probability distribution of a physical field and the corresponding detection quantities under the corresponding channel of the physical field was obtained. Based on the probability distribution, the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel was set.

[0009] According to the set sliding time window and step size, the multi-channel signals of the sound field, flow field and electric field collected in real time are intercepted to obtain the signal of the current frame. Based on the obtained signal of the current frame, the detection quantity of each physical field corresponding to the channel is calculated. The detection quantity of each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results of the sound field, flow field and electric field.

[0010] Based on the acquired multi-channel detection results under sound field, flow field, and electric field, the multi-channel fusion detection results under sound field, flow field, and electric field are calculated respectively based on the optimal threshold fusion criterion. Based on the multi-channel fusion detection results under sound field, flow field, and electric field, the basic probability allocation function under sound field, flow field, and electric field is calculated respectively.

[0011] Based on the basic probability allocation functions under the sound field, flow field, and electric field, the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field are fused according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection results of the sound field to obtain the final target detection result.

[0012] Preferably, according to the set sliding time window and step size, the original background noise signal of each channel of the sound field, flow field, and electric field is extracted into multiple frames, and the length of each frame signal is the length of the sliding time window of the corresponding channel, including:

[0013] The time dimension of the sliding window is set to K1, the time dimension of the sliding step is K2, and the sampling frequencies of the sound field, flow field, and electric field signals are Fs, respectively. 声 、Fs 流 、Fs 电 The corresponding window lengths are W 声 =K1×Fs 声 W 流 =K1×Fs 流 W 电 =K1×Fs 电 The corresponding step sizes are step声 =K2×Fs 声 step 流 =K2×Fs 流 step 电 =K2×Fs 电 ;

[0014] The original background noise signal is: X = {x1, x1, ..., x1} N}, then any two adjacent frame signals obtained after the sliding window capture are respectively: X i ={x l ,x l+1 ,…,x l+W-1}、X i+1 ={x l+step ,x l+step+1 ,…,x l+step+W-1}

[0015] Preferably, the detection quantity is calculated under different physical field background noise signals. The specific implementation process includes the following steps:

[0016] For the background noise signal: the formula for calculating the temporal energy of the i-th frame signal is as follows:

[0017]

[0018] In the formula, W is the corresponding window length, and x(i) is the amplitude corresponding to the discrete point of the frame signal time series;

[0019] For background noise signals in the sound field: the formula for calculating the EMD energy entropy detection value of the i-th frame signal is as follows:

[0020]

[0021] In the formula, e k The relative energy of the k-th eigenmode component obtained after EMD decomposition is calculated using the following formula:

[0022] e k =E(f k (t)) / E(x(t))

[0023] In the formula, E(f) k E(t) represents the energy of the k-th intrinsic mode component obtained from EMD decomposition, and E(x(j)) represents the total energy, i.e., E(x(t)) = E(f1(t)) + E(f2(t)) + ... + E(f...). n+1 (t));

[0024] For the background noise signal of the flow field: the formula for calculating the Hilbert energy detection quantity of the i-th frame signal is as follows:

[0025]

[0026] In the formula, w1 and w2 are the lower and upper cutoff frequencies of the characteristic frequency band of the flow field target, respectively, and E i (w) represents the Hilbert energy spectrum, as shown in the following formula:

[0027]

[0028] In the formula, H(w,t) is the Hilbert time spectrum of the i-th frame signal.

[0029] Preferably, in the k-th channel, for the detection quantity T m Statistical analysis was performed on different values ​​across multiple frames, and the corresponding probability density function f was obtained based on histogram statistical fitting. mk ;

[0030] For T1, T3, and T4, in the k-th channel, according to the preset false alarm rate P fa Determine the decision threshold r for the detection quantity 1k r 3k r 4k They respectively satisfy:

[0031] For T2, in the k-th channel, according to the preset false alarm rate P fa Determine the decision threshold r for the detection quantity 2k ,satisfy:

[0032] Preferably, the current i-th frame sound field signal is read in real time, and the detection quantity T of the i-th frame signal is calculated in the k-th channel. 1i and T 2i , respectively with threshold r 1k r 2k The comparison yielded the corresponding detection results as u 声ki_1 u 声ki_2 :;

[0033] The flow field signal of the current i-th frame is read in real time, and the detection quantity T of the i-th frame signal is calculated in the k-th channel. 3i , with threshold r 3k The comparison yields the following detection result for this channel:

[0034]

[0035] The electric field signal of the current i-th frame is read in real time, and the detection quantity T of the i-th frame signal is calculated in the k-th channel. 4i , with threshold r 4k The comparison yields the following detection result for this channel:

[0036]

[0037] Preferably, the number of sound field channels is C. 声 According to the optimal threshold fusion criterion, the detection result based on the sound field is u. 声i :

[0038]

[0039] In the formula, α1 and α2 are weights assigned based on the performance of the detection quantities T1 and T2.

[0040] Preferably, the basic probability allocation function based on sound field detection is:

[0041]

[0042] A real-time underwater target detection system, comprising:

[0043] The preprocessing module is used to collect the raw background noise signals of multiple physical fields and multiple channels based on the acoustic / flow / electric composite detection unit. According to the set sliding time window and step size, the raw background noise signals of each channel of the acoustic field, flow field and electric field are extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel.

[0044] The detection quantity module is used to calculate the time-domain energy detection quantity and EMD energy entropy detection quantity for each frame of signal in each channel for the sound field background noise signal; to calculate the Hilbert energy detection quantity for each frame of signal in each channel for the flow field background noise signal; and to perform a fast Fourier transform on each frame of signal in each channel for the electric field background noise signal to extract the characteristic frequency amplitude detection quantity of the electric field detection target.

[0045] The threshold module is used to perform statistical analysis on all acquired detection quantities and fit a physical field and the probability distribution of the corresponding detection quantities under the corresponding channel of the physical field. Based on the probability distribution, the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel is set.

[0046] The multi-channel detection module extracts the real-time acquired multi-channel signals of sound field, flow field and electric field according to the set sliding time window and step size to obtain the signal of the current frame. Based on the obtained signal of the current frame, it calculates the detection quantity of each physical field corresponding to the channel. The obtained detection quantity of each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results of sound field, flow field and electric field.

[0047] The probability allocation function calculation module calculates the multi-channel fusion detection results under the sound field, flow field, and electric field based on the obtained multi-channel detection results under the sound field, flow field, and electric field, respectively, and calculates the basic probability allocation function under the sound field, flow field, and electric field based on the multi-channel fusion detection results under the sound field, flow field, and electric field, respectively.

[0048] The target detection module, based on the basic probability allocation functions under the sound field, flow field, and electric field, fuses the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection result of the sound field to obtain the final target detection result.

[0049] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described real-time underwater target detection method.

[0050] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described real-time underwater target detection method.

[0051] Compared with the prior art, the present invention has the following beneficial technical effects:

[0052] This invention discloses a real-time underwater target detection method. It processes the raw background noise signal from multiple physical fields and channels to obtain different detection quantities under different physical fields. Based on the probability distribution calculation of these different detection quantities, a false alarm rate threshold is set according to the calculation results, thereby obtaining a verification standard for signal detection under different physical fields. Then, the optimal threshold fusion criterion is used to obtain the multi-channel fusion detection result and basic probability allocation function for a single physical field. Based on the DS fusion criterion, the {acoustic / current / electrical} fusion detection result is obtained as the final detection result and updated in real time. This method fully utilizes multi-physical field and multi-channel information (acoustic / current / electrical), reduces false alarms and missed detections, and improves detection accuracy.

[0053] This invention uses a sliding window with the same time dimension to analyze small segments of signals under multiple physical fields and multiple channels in real time and synchronously, including energy, energy entropy, and characteristic frequency. It combines the advantages of different physical fields and different features at the same time, shortens the algorithm execution time, and improves the reliability and real-time performance of underwater target detection. Attached Figure Description

[0054] Figure 1 This is a flowchart of underwater target detection in an embodiment of the present invention.

[0055] Figure 2 An example diagram for determining the Hilbert energy detection threshold of the first channel based on the flow field background noise signal.

[0056] Figure 3 This is a flowchart illustrating real-time target detection based on sound field, flow field, and electric field, respectively, and obtaining multi-channel detection results.

[0057] Figure 4 The flowchart shows how to determine the multi-channel fusion detection result and basic probability allocation function of a single physical field based on the multi-channel detection results of the sound field, flow field, and electric field, respectively.

[0058] Figure 5 This is an image showing the actual test results.

[0059] Figures 6(a), 6(b), and 6(c) are the detection results based on the sound field, flow field, and electric field, respectively.

[0060] Figures 7(a) and 7(b) show the basic probability allocation functions for detecting sound, flow, and electric fields based on a single physical field, and the basic probability allocation function for the fusion of {sound / flow / electricity}, respectively.

[0061] Figure 8(a) and Figure 8(b) are the final detection results of the {current / electricity} fusion detection and the {acoustic / current / electricity} fusion detection, respectively. Detailed Implementation

[0062] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0063] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0064] like Figure 1As shown, the present invention provides a method for real-time detection of underwater targets, comprising the following steps:

[0065] S1: Acquire raw background noise signals from multiple channels across multiple physical fields based on a composite acoustic / fluid / electrical detection unit;

[0066] S2: According to the set sliding time window and step size, the original background noise signal of each channel of sound field, flow field and electric field is extracted into multiple frames, and the length of each frame signal is the length of the sliding time window of the corresponding channel.

[0067] S3: For the sound field background noise signal, calculate the time-domain energy detection quantity and EMD energy entropy detection quantity for each frame of signal in each channel; for the flow field background noise signal, calculate the Hilbert energy detection quantity for each frame of signal in each channel; for the electric field background noise signal, perform a fast Fourier transform on each frame of signal in each channel to extract the characteristic frequency amplitude detection quantity of the electric field detection target.

[0068] S4: Perform statistical analysis on all the detection quantities calculated in step S3, and fit a physical field and the probability distribution of the corresponding detection quantities under the corresponding channel of the physical field. Set the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel according to the probability distribution.

[0069] S5: Read the multi-channel signals of the sound field, flow field and electric field to be detected in real time, use the corresponding sliding time window of the corresponding length in step S2 to obtain the signal of the current frame, calculate the corresponding detection quantity respectively and make a judgment according to the corresponding threshold in step S4 to obtain the multi-channel detection results of the sound field, flow field and electric field.

[0070] S6: Based on the multi-channel detection results obtained under the sound field, flow field, and electric field, calculate the multi-channel fusion detection results under the sound field, flow field, and electric field respectively based on the optimal threshold fusion criterion, and calculate the basic probability allocation function under the sound field, flow field, and electric field respectively based on the multi-channel fusion detection results under the sound field, flow field, and electric field.

[0071] S7: Based on the basic probability allocation functions under the sound field, flow field, and electric field, the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field are fused according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection results of the sound field to obtain the final target detection result.

[0072] According to the set sliding time window and step size, the original background noise signal of each channel of the sound field, flow field, and electric field is extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel. The specific implementation process is as follows:

[0073] (2-1) Set the time dimension of the sliding window to K1 and the time dimension of the sliding step to K2 to facilitate real-time decision fusion of sound, flow, and electricity. The sampling frequencies of the sound field, flow field, and electric field signals are Fs, respectively. 声 、Fs 流 、Fs 电 The corresponding window lengths are W 声 =K1×Fs 声 W 流 =K1×Fs 流 W 电 =K1×Fs 电 The corresponding step sizes are step 声 =K2×Fs 声 step 流 =K2×Fs 流 step 电 =K2×Fs 电 ;

[0074] (2-2) The original background noise signal is: X = {x1, x1, ..., x1} N}, then any two adjacent frame signals obtained after the sliding window capture are respectively: X i ={x l ,x l+1 ,…,x l+W-1}、X i+1 ={x l+step ,x l+step+1 ,…,x l+step+W-1}

[0075] Step S3 calculates the detection quantity under different physical field background noise signals. The specific implementation process includes the following steps:

[0076] (3-1) For the background noise signal: The formula for calculating the temporal energy detection quantity of the i-th frame signal is as follows:

[0077]

[0078] In the formula, W is the corresponding window length, and x(i) is the amplitude corresponding to the discrete point of the frame signal time series;

[0079] (3-2) For the background noise signal: The formula for calculating the EMD energy entropy detection value of the i-th frame signal is as follows:

[0080]

[0081] In the formula, e k The relative energy of the k-th eigenmode component obtained after EMD decomposition is calculated using the following formula:

[0082] e k=E(f k (t)) / E(x(t))

[0083] In the formula, E(f) k E(t) represents the energy of the k-th intrinsic mode component obtained from EMD decomposition, and E(x(j)) represents the total energy, i.e., E(x(t)) = E(f1(t)) + E(f2(t)) + ... + E(f...). n+1 (t));

[0084] (3-3) For the background noise signal of the flow field: the formula for calculating the Hilbert energy detection quantity of the i-th frame signal is as follows:

[0085]

[0086] In the formula, w1 and w2 are the lower and upper cutoff frequencies of the characteristic frequency band of the flow field target, respectively, and E i (w) represents the Hilbert energy spectrum, as shown in the following formula:

[0087]

[0088] In the formula, H(w,t) is the Hilbert time spectrum of the i-th frame signal;

[0089] (3-4) For the electric field background noise signal: the characteristic frequency amplitude detection quantity T4(X) of the i-th frame signal i )=y f In the formula y f The amplitude corresponding to the characteristic frequency obtained after FFT transformation is denoted as f, where f is the characteristic frequency. If multiple characteristic frequencies or characteristic frequency bands exist, they can be replaced with the sum of their amplitudes.

[0090] Step S4 involves statistical analysis of all detected quantities and fitting a physical field and the probability distribution of the corresponding detected quantities under the corresponding channel of that physical field. The specific implementation process of setting the threshold of the false alarm rate corresponding to the corresponding detected quantity under the corresponding channel based on the probability distribution includes:

[0091] (4-1) In the k-th channel, for the detection quantity T m Statistical analysis was performed on different values ​​across multiple frames, and the corresponding probability density function f was obtained based on histogram statistical fitting. mk ;

[0092] (4-2) For T1, T3, and T4, in the k-th channel, according to the preset false alarm rate P fa Determine the decision threshold r for the detection quantity 1k r 3k r 4k They respectively satisfy:

[0093] (4-3) For T2, in the k-th channel, according to the preset false alarm rate P fa Determine the decision threshold r for the detection quantity 2k ,satisfy:

[0094] The specific implementation process of step S5 includes:

[0095] (5-1) Read the current i-th frame sound field signal in real time, and calculate the detection quantity T of the i-th frame signal in the k-th channel. 1i and T 2i , respectively with threshold r 1k r 2k The comparison yielded the corresponding detection results as u 声ki_1 u 声ki_2 :

[0096]

[0097]

[0098] (5-2) Read the current i-th frame flow field signal in real time, and calculate the detection quantity T of the i-th frame signal in the k-th channel. 3i , with threshold r 3k The comparison yields the following detection result for this channel:

[0099]

[0100] (5-3) Read the current i-th frame electric field signal in real time, and calculate the detection quantity T of the i-th frame signal in the k-th channel. 4i , with threshold r 4k The comparison yields the following detection result for this channel:

[0101]

[0102] The specific implementation process of step S6 includes:

[0103] (6-1) The number of sound field channels is C 声 According to the optimal threshold fusion criterion, the detection result based on the sound field is u. 声i :

[0104]

[0105] In the formula, α1 and α2 are the weights assigned based on the performance of the detection quantities T1 and T2;

[0106] The basic probability allocation function based on sound field detection is:

[0107]

[0108] m声i (0, no target) = 1 - m 声i (1. Have a goal)

[0109] (6-2) The number of flow field channels is C. 流 According to the optimal threshold fusion criterion, the detection result based on the flow field is u. 流i :

[0110]

[0111] The basic probability assignment function based on flow field detection is:

[0112]

[0113] m 流i (0, no target) = 1 - m 流i (1. Have a goal)

[0114] (6-3) The number of electric field channels is C. 电 According to the optimal threshold fusion criterion, the detection result based on the electric field is u. 电i :

[0115]

[0116] The basic probability allocation function based on electric field detection is:

[0117]

[0118] m 电i (0, no target) = 1 - m 电i (1. Have a goal)

[0119] Step S7, based on the fundamental probability allocation functions under the sound field, flow field, and electric field, fuses the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field according to DS evidence theory to obtain the flow field / electric field fusion result. The specific implementation process of fusing the flow field / electric field fusion result with the multi-channel fusion detection result of the sound field to obtain the final target detection result includes:

[0120] (7-1) Based on the DS evidence theory, the flow field and electric field detection results of the i-th frame are fused to obtain the {flow / electricity} fused detection result:

[0121]

[0122] In the formula m {流 / 电}i (1, with target) is the basic probability allocation function for {current / electricity} fusion detection obtained through DS fusion, as shown below:

[0123]

[0124] In the formula This is the normalization coefficient.

[0125] (7-2) Based on the DS evidence theory, the {flow / electricity} fusion detection result and the sound field detection result of the i-th frame are fused to obtain the {sound / flow / electricity} fusion detection result:

[0126]

[0127] In the formula m {声 / 流 / 电}i (1, with target) is the basic probability allocation function for {acoustic / fluid / electric} fusion detection obtained through DS fusion, as shown below:

[0128]

[0129] In the formula These are the normalization coefficients;

[0130] (7-3) Update the detection results of the current i-th frame in real time.

[0131] Figure 2 This is an example diagram for determining the Hilbert energy detection threshold of the first channel of the flow field based on the background noise signal. By performing statistical analysis on the detection quantities of multiple frames of background noise signals, a probability distribution can be fitted, and then the corresponding threshold can be determined according to the given false alarm rate.

[0132] To verify the advantages of the real-time underwater target detection method proposed in this invention (acoustic / current / electrical), an experiment was conducted to validate the method. Figure 5 Figure 6(a) shows the real detection results of a certain experiment. No target was detected in frames 0-50, and the target appeared in frames 51-100. Figure 6(a) shows the real-time detection results based on the sound field. False alarms occurred in frames 5 and 35, and a missed detection occurred in frame 67. Figure 6(b) shows the real-time detection results based on the flow field. False alarms occurred in frames 17 and 41, and a missed detection occurred in frames 63 and 79. Figure 6(c) shows the real-time detection results based on the electric field. False alarms occurred in frame 17, and a missed detection occurred in frames 51, 54, and 88. Figure 7(a) shows the basic probability allocation function m for real-time target detection based on the sound field, flow field, and electric field, respectively. 声 (1), m 流 (1), m 电 (1) As can be seen from the figure, outliers will appear when false alarms or missed detections occur; Figure 7(b) shows the basic probability allocation functions m based on {current / electricity} fusion detection and {acoustic / current / electricity} fusion detection, respectively. {流 / 电} m {声 / 流 / 电}As can be seen from the figures, after flow / electricity fusion, the number of outliers is reduced, but a few outliers still exist. After sound / flow / electricity fusion, all outliers disappear, and the detection results are more reliable. Figure 8(a) is the detection result of flow / electricity fusion. As can be seen from the figure, false alarms and missed detections are reduced, but false alarms still exist in the 17th frame. Figure 8(b) is the detection result of sound / flow / electricity fusion (final detection result). As can be seen from the figure, false alarms and missed detections have all disappeared, and accurate real-time detection has been achieved for each frame.

[0133] In one embodiment of the present invention, a real-time underwater target detection system is provided, comprising:

[0134] The preprocessing module is used to collect the raw background noise signals of multiple physical fields and multiple channels based on the acoustic / flow / electric composite detection unit. According to the set sliding time window and step size, the raw background noise signals of each channel of the acoustic field, flow field and electric field are extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel.

[0135] The detection quantity module is used to calculate the time-domain energy detection quantity and EMD energy entropy detection quantity for each frame of signal in each channel for the sound field background noise signal; to calculate the Hilbert energy detection quantity for each frame of signal in each channel for the flow field background noise signal; and to perform a fast Fourier transform on each frame of signal in each channel for the electric field background noise signal to extract the characteristic frequency amplitude detection quantity of the electric field detection target.

[0136] The threshold module is used to perform statistical analysis on all acquired detection quantities and fit a physical field and the probability distribution of the corresponding detection quantities under the corresponding channel of the physical field. Based on the probability distribution, the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel is set.

[0137] The multi-channel detection module extracts the real-time acquired multi-channel signals of sound field, flow field and electric field according to the set sliding time window and step size to obtain the signal of the current frame. Based on the obtained signal of the current frame, it calculates the detection quantity of each physical field corresponding to the channel. The obtained detection quantity of each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results of sound field, flow field and electric field.

[0138] The probability allocation function calculation module calculates the multi-channel fusion detection results under the sound field, flow field, and electric field based on the obtained multi-channel detection results under the sound field, flow field, and electric field, respectively, and calculates the basic probability allocation function under the sound field, flow field, and electric field based on the multi-channel fusion detection results under the sound field, flow field, and electric field, respectively.

[0139] The target detection module, based on the basic probability allocation functions under the sound field, flow field, and electric field, fuses the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection result of the sound field to obtain the final target detection result.

[0140] In another embodiment of the present invention, a terminal device is provided, the terminal device including a processor and a memory, the memory being used to store a computer program, the computer program including program instructions, and the processor being used to execute the program instructions stored in the computer storage medium.The processor may be a Central Processing Unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), or field-programmable gate arrays (FPGAs). Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., are the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to realize the corresponding method flow or corresponding function; the processor described in this embodiment of the invention can be used to operate a real-time underwater target detection method, including the following steps: acquiring multi-physics field multi-channel background noise raw signals based on a sound / flow / electric composite detection unit; according to the set sliding time window and step size, extracting the background noise raw signals of each channel of the sound field, flow field, and electric field into multiple frames, the length of each frame signal being the length of the sliding time window of the corresponding channel; for the sound field background noise signal, calculating the time-domain energy detection quantity and EMD energy entropy detection quantity of each frame signal under each channel; for the flow field background noise signal, calculating the Hilbert energy detection quantity of each frame signal under each channel; for the electric field background noise signal, performing a fast Fourier transform on each frame signal under each channel to extract the characteristic frequency amplitude detection quantity of the electric field detection target; statistically analyzing all the acquired detection quantities. The system analyzes and fits a physical field and the probability distribution of the corresponding detection quantity under the corresponding channel of that physical field. Based on this probability distribution, a threshold for the false alarm rate corresponding to the detection quantity under the corresponding channel is set. The multi-channel signals of the real-time acquired sound field, flow field, and electric field are intercepted according to the set sliding time window and step size to obtain the signal of the current frame. The detection quantity under each physical field corresponding to the channel is calculated based on the obtained signal of the current frame. The obtained detection quantity under each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results under the sound field, flow field, and electric field. Based on the obtained sound field and flow field... The multi-channel detection results under the electric field are calculated based on the optimal threshold fusion criterion, and the multi-channel fusion detection results under the sound field, flow field, and electric field are calculated respectively. Based on the multi-channel fusion detection results under the sound field, flow field, and electric field, the basic probability allocation functions under the sound field, flow field, and electric field are calculated respectively. Based on the basic probability allocation functions under the sound field, flow field, and electric field, the multi-channel fusion detection results of the flow field and the electric field are fused according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection results of the sound field to obtain the final target detection result.

[0141] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory). This computer-readable storage medium is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device.

[0142] One or more instructions stored in a computer-readable storage medium can be loaded and executed by the processor to implement the corresponding steps of the real-time underwater target detection method in the above embodiments; based on the acoustic / flow / electric composite detection unit, the background noise raw signals of multiple physical fields and multiple channels are collected, and according to the set sliding time window and step size, the background noise raw signals of each channel of the acoustic field, flow field and electric field are respectively extracted into multiple frames, and the length of each frame signal is the length of the sliding time window of the corresponding channel;

[0143] For the sound field background noise signal, the time-domain energy detection quantity and EMD energy entropy detection quantity of each frame signal under each channel are calculated respectively; for the flow field background noise signal, the Hilbert energy detection quantity of each frame signal under each channel is calculated respectively; for the electric field background noise signal, a fast Fourier transform is performed on each frame signal under each channel to extract the characteristic frequency amplitude detection quantity of the electric field detection target.

[0144] Statistical analysis was performed on all the detection quantities obtained above, and a probability distribution of a physical field and the corresponding detection quantities under the corresponding channel of the physical field was obtained. Based on the probability distribution, the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel was set.

[0145] According to the set sliding time window and step size, the multi-channel signals of the sound field, flow field and electric field collected in real time are intercepted to obtain the signal of the current frame. Based on the obtained signal of the current frame, the detection quantity of each physical field corresponding to the channel is calculated. The detection quantity of each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results of the sound field, flow field and electric field.

[0146] Based on the acquired multi-channel detection results under sound field, flow field, and electric field, the multi-channel fusion detection results under sound field, flow field, and electric field are calculated respectively based on the optimal threshold fusion criterion. Based on the multi-channel fusion detection results under sound field, flow field, and electric field, the basic probability allocation function under sound field, flow field, and electric field is calculated respectively.

[0147] Based on the basic probability allocation functions under the sound field, flow field, and electric field, the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field are fused according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection results of the sound field to obtain the final target detection result.

[0148] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0149] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0150] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0151] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0152] This invention proposes a real-time underwater target detection method based on acoustic / fluid / electrical decision fusion: The method processes the raw background noise signal from multiple physical fields and channels to obtain different detection quantities under different physical fields. Probability distributions are calculated for these different detection quantities, and a false alarm rate threshold is set based on the calculation results to obtain a verification standard for signal detection under different physical fields. Then, the optimal threshold fusion criterion is used to obtain the multi-channel fusion detection result and basic probability allocation function for a single physical field. Based on the DS fusion criterion, the {acoustic / fluid / electrical} fusion detection result is obtained as the final detection result and updated in real time. This method fully utilizes the multi-physical field and multi-channel information of acoustic / fluid / electrical signals, reduces false alarms and missed detections, and improves detection accuracy.

[0153] This invention uses a sliding window with the same time dimension to analyze small segments of signals under multiple physical fields and multiple channels in real time and synchronously, including energy, energy entropy, and characteristic frequency. It combines the advantages of different physical fields and different features at the same time, shortens the algorithm execution time, and improves the reliability and real-time performance of underwater target detection.

Claims

1. A method for real-time detection of underwater targets, characterized in that, Includes the following steps: Based on the acoustic / flow / electric composite detection unit, the background noise raw signals of multiple physical fields and multiple channels are collected. According to the set sliding time window and step size, the background noise raw signals of each channel of the acoustic field, flow field and electric field are extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel. For the sound field background noise signal, calculate the temporal energy detection quantity and EMD energy entropy detection quantity for each frame of signal in each channel; for the flow field background noise signal, calculate the Hilbert energy detection quantity for each frame of signal in each channel. For the electric field background noise signal, a fast Fourier transform is performed on each frame signal in each channel to extract the characteristic frequency amplitude detection quantity of the electric field detection target. Statistical analysis was performed on all the acquired detection quantities, and a probability distribution of a physical field and the corresponding detection quantities under the corresponding channel of the physical field was obtained. Based on the probability distribution, the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel was set. According to the set sliding time window and step size, the multi-channel signals of the sound field, flow field and electric field collected in real time are intercepted to obtain the signal of the current frame. Based on the obtained signal of the current frame, the detection quantity of each physical field corresponding to the channel is calculated. The detection quantity of each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results of the sound field, flow field and electric field. Based on the acquired multi-channel detection results under sound field, flow field, and electric field, the multi-channel fusion detection results under sound field, flow field, and electric field are calculated respectively based on the optimal threshold fusion criterion. Based on the multi-channel fusion detection results under sound field, flow field, and electric field, the basic probability allocation function under sound field, flow field, and electric field is calculated respectively. Based on the basic probability allocation functions under the sound field, flow field, and electric field, the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field are fused according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection results of the sound field to obtain the final target detection result.

2. The method for real-time detection of underwater targets according to claim 1, characterized in that, According to the set sliding time window and step size, the original background noise signal of each channel of the sound field, flow field, and electric field is extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel, including: Setting the time dimension of the sliding window K 1. The time dimension of the sliding step is K 2. The sampling frequencies of the sound field, flow field, and electric field signals are respectively... , , The corresponding window lengths are respectively , , The corresponding step sizes are respectively , , ; The original background noise signal is: Then, any two adjacent frame signals obtained after the sliding window is truncated are as follows: , .

3. The method for real-time detection of underwater targets according to claim 1, characterized in that, The calculation of the detection quantity under different physical field background noise signals includes the following steps: For the background noise signal in the sound field: the first i The formula for calculating the time-domain energy detection quantity of the frame signal is as follows: In the formula, W For the corresponding window length, The amplitude corresponding to discrete points in the time series of the frame signal; For the background noise signal in the sound field: the first i The formula for calculating the EMD energy entropy detection value of the frame signal is as follows: In the formula, The first one obtained after EMD decomposition k The relative energy of each eigenmode component is calculated using the following formula: In the formula, The first one obtained from EMD decomposition k The energy of each intrinsic mode component The total energy, i.e. ; For the background noise signal of the flow field: the first i The formula for calculating the Hilbert energy detection value of a frame signal is as follows: In the formula, w 1 and w 2 represents the lower and upper cutoff frequencies of the characteristic frequency band of the flow field target. The Hilbert energy spectrum is given by the following formula: In the formula, For the first i Hilbert time spectrum of the frame signal.

4. The method for real-time detection of underwater targets according to claim 3, characterized in that, No. k Under the channel, the detection volume Statistical analysis was performed on different values ​​across multiple frames, and the corresponding probability density function was obtained based on histogram statistical fitting. ; for , , In the k Under the channel, based on the preset false alarm rate Determine the decision threshold for the detection quantity , , They respectively satisfy: , , ; for In the k Under the channel, based on the preset false alarm rate Determine the decision threshold for the detection quantity ,satisfy: .

5. The method for real-time detection of underwater targets according to claim 1, characterized in that, Real-time reading of the current number i Frame sound field signal, in the first k Under the channel, calculate the first i Frame signal detection quantity and , respectively with threshold , The comparison yielded the following corresponding detection results: , ; Real-time reading of the current number i Frame flow field signal, in the first k Under the channel, calculate the first i Frame signal detection quantity , with threshold The comparison yields the following detection result for this channel: Real-time reading of the current number i Frame electric field signal, in the first k Under the channel, calculate the first i Frame signal detection quantity , with threshold The comparison yields the following detection result for this channel: 。 6. The method for real-time detection of underwater targets according to claim 1, characterized in that, The number of sound field channels is According to the optimal threshold fusion criterion, the detection result based on the sound field is: : In the formula, and It is based on the detection volume and The weights assigned to the performance.

7. The method for real-time detection of underwater targets according to claim 6, characterized in that, The basic probability allocation function based on sound field detection is: 。 8. A real-time underwater target detection system, characterized in that, include: The preprocessing module is used to collect the raw background noise signals of multiple physical fields and multiple channels based on the acoustic / flow / electric composite detection unit. According to the set sliding time window and step size, the raw background noise signals of each channel of the acoustic field, flow field and electric field are extracted into multiple frames. The length of each frame signal is the length of the sliding time window of the corresponding channel. The detection module is used to calculate the temporal energy detection quantity and EMD energy entropy detection quantity of each frame signal in each channel for the sound field background noise signal; and to calculate the Hilbert energy detection quantity of each frame signal in each channel for the flow field background noise signal. For the electric field background noise signal, a fast Fourier transform is performed on each frame signal in each channel to extract the characteristic frequency amplitude detection quantity of the electric field detection target. The threshold module is used to perform statistical analysis on all acquired detection quantities and fit a physical field and the probability distribution of the corresponding detection quantities under the corresponding channel of the physical field. Based on the probability distribution, the threshold of the false alarm rate corresponding to the corresponding detection quantity under the corresponding channel is set. The multi-channel detection module extracts the real-time acquired multi-channel signals of sound field, flow field and electric field according to the set sliding time window and step size to obtain the signal of the current frame. Based on the obtained signal of the current frame, it calculates the detection quantity of each physical field corresponding to the channel. The obtained detection quantity of each physical field corresponding to the channel is compared with the set false alarm rate threshold to obtain the multi-channel detection results of sound field, flow field and electric field. The probability allocation function calculation module calculates the multi-channel fusion detection results under the sound field, flow field, and electric field based on the obtained multi-channel detection results under the sound field, flow field, and electric field, respectively, and calculates the basic probability allocation function under the sound field, flow field, and electric field based on the multi-channel fusion detection results under the sound field, flow field, and electric field, respectively. The target detection module, based on the basic probability allocation functions under the sound field, flow field, and electric field, fuses the multi-channel fusion detection results of the flow field and the multi-channel fusion detection results of the electric field according to the DS evidence theory to obtain the flow field / electric field fusion result. The flow field / electric field fusion result is then fused with the multi-channel fusion detection result of the sound field to obtain the final target detection result.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the real-time underwater target detection method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the real-time underwater target detection method as described in any one of claims 1 to 7.

Images