Underwater acoustic target dynamic planning track-before-detect method based on azimuth potential function weighting

Abstract

The invention discloses an underwater acoustic target dynamic planning track-before-detect method based on azimuth potential function weighting, relates to an underwater acoustic target dynamic planning track-before-detect method based on azimuth potential function weighting, and belongs to the field of passive sonar underwater acoustic multi-target tracking. The objective of the invention is to solve the problem that a traditional track-before-detect algorithm based on dynamic programming is easy to cause tracking trajectory merging or tracking trajectory mismatching. The method comprises the following steps: establishing an augmented target state model containing a target azimuth and an azimuth potential coefficient, establishing a sliding window during iteration of a value function, calculating the azimuth potential coefficient of a trajectory in the window, and establishing an azimuth potential function by considering the azimuth potential coefficient defining an azimuth tracking trajectory change trend. The azimuth potential function is added into the value function to reflect the change trend of the azimuth trajectory, so that the probability of tracking performance reduction caused by spatial spectrum response overlapping when the trajectory is adjacent and crossed can be reduced.

Description

Technical Field

[0001] The invention relates to a method for underwater acoustic target dynamic programming before detection based on azimuth potential function weighting, and belongs to the field of passive sonar underwater acoustic multi-target tracking. Background Art

[0002] Passive sonar system is an underwater acoustic device that detects, locates and tracks targets by receiving target sound wave information. For the problem of passive sonar underwater acoustic multi-target tracking, one of the key research directions is to effectively detect and accurately track weak underwater acoustic targets with low signal-to-noise ratio.

[0003] In recent years, the tracking-before-detection method has gradually attracted people's attention. The tracking-before-detection method has good performance in the field of underwater acoustic weak target detection and tracking. Different from the traditional detection-before-tracking method, the tracking-before-detection method stores the data received by the sonar array that exceeds or falls below the threshold. It does not need threshold processing, but directly uses the sonar data for preprocessing, and tracks the target by using the correlation of the target azimuth estimation between adjacent frames and the non-correlation of noise.

[0004] The pre-detection tracking algorithms applied to underwater acoustic multi-targets are mainly divided into pre-detection tracking based on particle filtering and pre-detection tracking based on dynamic programming. Pre-detection tracking based on particle filtering relies on Monte Carlo integral approximation, and uses particles with weights to characterize the posterior probability density of the target state to establish and maintain the target trajectory. The essence of pre-detection tracking based on dynamic programming is to pre-calculate all possible trajectories of the target, store these trajectories, accumulate observation data along the trajectory, and finally take the trajectory corresponding to the accumulation amount exceeding the threshold as the true trajectory of the target. Compared with the pre-detection tracking method based on particle filtering, the pre-detection tracking method based on dynamic programming has the advantages of low computational complexity and strong determinism, and is more suitable for systems with high real-time requirements.

[0005] The traditional tracking-before-detection algorithm based on dynamic programming has good tracking performance when the tracking separation is good and there is no measurement ambiguity. However, when the targets are adjacent to each other, the strong target masks the weak target, the value function is aliased, and the tracking track merging or tracking track mismatching problems may occur easily.

[0006] In summary, the tracking performance of the existing passive sonar underwater acoustic target dynamic planning pre-detection tracking method is still relatively limited in the adjacent target scenario. It is very necessary to propose a new underwater acoustic adjacent target dynamic planning pre-detection tracking method. Summary of the invention

[0007] The purpose of the present invention is to solve the problem that when the traditional pre-detection tracking algorithm based on dynamic programming in the passive tracking scenario of underwater acoustic targets has adjacent intersections, the strong target covers the weak target, the value function is aliased, and the tracking trajectory is easily merged or mismatched. A pre-detection tracking method for underwater acoustic targets based on dynamic programming weighting of azimuth potential function is proposed.

[0008] A method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting is as follows:

[0009] Step 1, beamforming the passive sonar detection signal to obtain spatial spectrum data;

[0010] The spatial spectrum data is used as the measurement data, and the azimuth observation range of each frame in the measurement data is divided into B resolution units;

[0011] Step 2, initialize the state transition number q;

[0012] When the number of state transitions is q, the kth frame in the measurement data is transferred to the k+1th frame, and the target state information x of the k+1th frame is obtained. k+1 ;

[0013] Step 3, initializing the frame number k in the measurement data to 1;

[0014] Step 4, initialize the target count m=1;

[0015] Step 5: Define the azimuth potential coefficient, and define the augmented state information including the azimuth potential coefficient and the target azimuth information based on the azimuth potential coefficient

[0016] Step 6: Based on the measurement data from step 1 and the target state information x from step 2 k+1 , the augmented state information containing the azimuth potential coefficient and target azimuth information in step 5 Initialize the augmented state information of the first n frames with the same length as the window length n of the azimuth potential coefficient Get the value function and track backtracking function;

[0017] Step 7: Determine whether k>n. If yes, execute step 8; if no, set k=k+1 and repeat steps 6 to 7.

[0018] Step 8, adding the azimuthal potential function weight to the value function obtained in step 6 to obtain a new value function;

[0019] Based on the new value function and azimuth potential function, the track tracing function of the mth target in the kth frame is obtained;

[0020] Step 9: Determine if k=K. If yes, execute step 10. If no, set k=k+1 and repeat steps 8 to 9.

[0021] K is the total number of frames in the measurement data;

[0022] Step 10: Compare the obtained new value function with the decision threshold, remove the targets of the frames corresponding to the value function less than the decision threshold, and retain the targets of the frames corresponding to the value function greater than or equal to the decision threshold;

[0023] Step 11, based on the target of the frame corresponding to the value function retained in step 10 and the track backtracking function of step 8, the track backtracking is performed to obtain the mth target tracking track;

[0024] Step 12: determine if m=M, and if so, obtain M target tracking trajectories;

[0025] If not, set m=m+1 and repeat steps 8 to 12 until m=M, and obtain M target tracking trajectories.

[0026] The beneficial effects of the present invention are:

[0027] Aiming at the problem of target proximity tracking in passive sonar weak target scenarios, the present invention proposes an azimuth potential function-weighted dynamic programming tracking before detection (APWDP-TBD), establishes an augmented target state model including target azimuth and azimuth potential coefficient, establishes a sliding window when the value function is iterated, calculates the azimuth potential coefficient of the trajectory within the window, establishes the azimuth potential function by considering the azimuth potential coefficient that defines the changing trend of the azimuth tracking trajectory, and adds the azimuth potential function to the value function to reflect the changing trend of the azimuth trajectory, which can reduce the probability of tracking performance degradation caused by the overlap of spatial spectrum response when the trajectories are adjacent to each other.

[0028] When the target is adjacent to each other, the tracking-before-detection algorithm based on dynamic programming will have a broadening phenomenon of the value function due to mutual interference between different targets, which is easy to cause the accumulation of value function errors. The present invention proposes a value function with added azimuth potential function weighting to replace the conventional value function. When constructing the state variable, the influence of the two quantities of azimuth and azimuth potential coefficient is considered. The probability that the predicted state belongs to the current target is judged by recording the azimuth potential coefficient, so that the value functions of different targets can be correctly separated in the target proximity scene; it solves the problem that when the target is adjacent to each other, the traditional tracking-before-detection algorithm based on dynamic programming in the existing passive tracking scenario of underwater acoustic targets is prone to tracking trajectory merging or tracking trajectory mismatching due to strong targets covering weak targets when the target is adjacent to each other. The simulation experimental results show that the dynamic programming tracking-before-detection method based on azimuth potential function weighting can achieve correct tracking when the target is adjacent to each other in a low signal-to-noise ratio environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is a flow chart of the method of the present invention;

[0030] Figure 2 It is the time-azimuth history diagram after beamforming;

[0031] Figure 3(a) is a schematic diagram of the tracking results of the DP-TBD algorithm in a neighboring scene. DP-TBD is a tracking-before-detection algorithm based on dynamic programming.

[0032] Figure 3(b) is a schematic diagram of the tracking results of the APWDP-TBD algorithm in a neighboring scene. APWDP-TBD is a dynamic programming tracking before detection based on the weighted azimuth potential function.

[0033] Figure 4(a) shows the optimal sub-pattern assignment error of the DP-TBD algorithm, and OSPA is the optimal sub-pattern assignment;

[0034] Figure 4(b) shows the optimal sub-mode allocation error of the APWDP-TBD algorithm. DETAILED DESCRIPTION

[0035] Specific implementation method 1. Combination Figure 1 The present embodiment is described. The specific process of the underwater acoustic target dynamic programming pre-detection tracking method based on azimuth potential function weighting is as follows:

[0036] Step 1, beamforming the passive sonar detection signal to obtain spatial spectrum data;

[0037] The spatial spectrum data is used as the measurement data, and the azimuth observation range of each frame in the measurement data is divided into B resolution units;

[0038] Step 2, initialize the state transition number q;

[0039] When the number of state transitions is q, the kth frame in the measurement data is transferred to the k+1th frame, and the target state information x of the k+1th frame is obtained. k+1 ;

[0040] Step 3, initializing the frame number k in the measurement data to 1;

[0041] Step 4, initialize the target count m=1;

[0042] Step 5: Define the azimuth potential coefficient, and define the augmented state information including the azimuth potential coefficient and the target azimuth information based on the azimuth potential coefficient

[0043] Step 6: Based on the measurement data from step 1 and the target state information x from step 2 k+1 , the augmented state information containing the azimuth potential coefficient and target azimuth information in step 5 Initialize the augmented state information of the first n frames with the same length as the window length n of the azimuth potential coefficient Get the value function and track backtracking function;

[0044] Step 7: Determine whether k>n. If yes, execute step 8; if no, set k=k+1 and repeat steps 6 to 7.

[0045] Step 8, adding the azimuthal potential function weight to the value function obtained in step 6 to obtain a new value function;

[0046] Based on the new value function and azimuth potential function, the track tracing function of the mth target in the kth frame is obtained;

[0047] Step 9: Determine if k=K. If yes, execute step 10. If no, set k=k+1 and repeat steps 8 to 9.

[0048] K is the total number of frames in the measurement data;

[0049] Step 10: Compare the obtained new value function with the decision threshold, remove the targets of the frames corresponding to the value function less than the decision threshold, and retain the targets of the frames corresponding to the value function greater than or equal to the decision threshold;

[0050] Step 11, based on the target of the frame corresponding to the value function retained in step 10 and the track backtracking function of step 8, the track backtracking is performed to obtain the mth target tracking track;

[0051] Step 12: determine if m=M, and if so, obtain M target tracking trajectories;

[0052] If not, set m=m+1 and repeat steps 8 to 12 until m=M, and obtain M target tracking trajectories.

[0053] Specific implementation method 2: This implementation method is different from specific implementation method 1 in that: in step 1, beamforming is performed on the passive sonar detection signal to obtain spatial spectrum data;

[0054] The spatial spectrum data is used as the measurement data, and the azimuth observation range of each frame in the measurement data is divided into B resolution units;

[0055] The specific process is:

[0056] Step 11, initialize the number of linear array elements, element spacing, sampling frequency, sound velocity, and frame number;

[0057] Step 12: beamforming the passive sonar detection signal to obtain spatial spectrum data;

[0058] Step 13: Use the spatial spectrum data as measurement data, denoted as Z 1:K ={z 1 ,z 2 ,...,z k ,...,z K};

[0059] Among them, Z 1:K is the measurement data set, z 1 is the first frame of the measurement data, z 2 is the second frame of the measurement data, z k is the kth frame in the measurement data, z K is the Kth frame in the measurement data, K is the total number of frames in the measurement data;

[0060] Step 14, dividing the azimuth observation range of each frame in the measurement data into B resolution units, and the size of each resolution unit is Δθ; the observation range of each frame is to be divided into the same number of resolution units;

[0061] Use z k (b) represents the amplitude of the measured data at the b-th resolution unit in the k-th frame, and the amplitudes of the measured data in all resolution units are independent and identically distributed, b = 1, 2, ..., B;

[0062] z k (b) is expressed as:

[0063]

[0064] Among them, ω k (b) represents the noise amplitude of the b-th resolution unit in the k-th frame, A k (b) represents the target signal amplitude of the b-th resolution unit in the k-th frame.

[0065] The other steps and parameters are the same as those in the first embodiment.

[0066] Specific implementation method three: This implementation method is different from specific implementation method one or two in that: in step 2, the initialization state transition number q (the initialization state transition number q has a value range of 0-41);

[0067] When the number of state transitions is q, the kth frame in the measurement data is transferred to the k+1th frame, and the target state information x of the k+1th frame is obtained. k+1 ;

[0068] The specific process is:

[0069] Initialize the state transition number q, which is used to represent the number of possible state transitions of the target in the next observation phase in the dynamic programming algorithm;

[0070] When the number of state transitions is q, the kth frame in the measurement data transitions to the k+1th frame, and the target state information x of the k+1th frame is k+1 for:

[0071]

[0072] Among them, θk is the orientation information of the kth frame, Δθ is the size of the resolution unit, δ q is the state transition range when the state transition number is q; x k+1 Indicates the target state information of the k+1th frame.

[0073] The other steps and parameters are the same as those in the first or second embodiment.

[0074] Specific implementation method 4: This implementation method is different from any one of the specific implementation methods 1 to 3 in that: in step 5, the azimuth potential coefficient is defined, and the augmented state information including the azimuth potential coefficient and the target azimuth information is defined based on the azimuth potential coefficient. The specific process is:

[0075] Step 51: Definition Represents the azimuth potential coefficient of the mth target in the kth frame; it is expressed as:

[0076]

[0077] Where n is the window length of the azimuth potential coefficient, m is the target count, ω i is the weight of the i-th frame, θ i is the orientation information of the i-th frame, θ i-1 is the orientation information of the i-1th frame;

[0078] The weight of points whose frames are closer to the current frame should be larger to reflect the change in potential coefficient;

[0079] Step 52: Based on the azimuth potential coefficient of the mth target in the kth frame Define the augmented state information of the mth target in the kth frame, which includes the azimuth potential coefficient and the target azimuth information; it is expressed as:

[0080]

[0081] in, Indicates the target position information of the mth target in the kth frame; Represents the augmented state information of the mth target in the kth frame.

[0082] The other steps and parameters are the same as those in Specific Embodiments 1 to 3.

[0083] Specific implementation mode 5: This implementation mode is different from any one of the specific implementation modes 1 to 4 in that: in step 6, the measured data of step 1 and the target state information x of step 2 are used as the basis for the measurement data of step 1 and the target state information x of step 2. k+ 1. Augmented state information in step 5 including azimuth potential coefficient and target azimuth information Initialize the augmented state information of the first n frames with the same length as the window length n of the azimuth potential coefficient Get the value function and track backtracking function;

[0084] The specific process is:

[0085] Step 61: Express the state transfer equation and measurement equation of the target as follows:

[0086]

[0087] Among them, f k-1 (·) represents the state transfer function, h k (·) represents the measurement function, x k represents the augmented state information of the kth frame, x k-1 represents the augmented state information of the k-1th frame, w k represents the state transition noise, v k represents the measurement noise, which follows Gaussian distribution, z k represents the measurement data of the kth frame;

[0088] Step 62: The target sequence with the largest posterior probability Right now:

[0089]

[0090] in, is the augmented state information of the target with the largest posterior probability in the first frame, is the augmented state information of the target with the largest posterior probability density in the second frame, is the augmented state information of the target with the largest posterior probability in the kth frame, is the target state set with the largest posterior probability in the previous k frames, is the maximum a posteriori probability density function, p(X 1:k |Z 1:k ) is the posterior probability density function;

[0091] The posterior probability density function p(X 1:k |Z 1:k ) is obtained from the prior information of the target and the noise;

[0092] The posterior probability density function of the target represents the measurement data set Z 1:k The target state set X 1:k The probability of is determined;

[0093] Step 63: Perform Bayesian transformation on equation (6) in step 62 to obtain:

[0094]

[0095] Among them, p(X 1:k |Z 1:k) is the posterior probability density function of the first k frames, X 1:k is the target state set of the previous k frames, Z 1:k is the measurement data set, p(z k |x k ) is the likelihood function of the kth frame, z k represents the measurement data of the kth frame, x k is the target state information of the kth frame, p(x k |x k-1 ) is the state transition density function, p(X 1:k-1 |Z 1:k-1 ) is the posterior probability density function of the previous k-1 frames, p(z k |Z 1:k-1 ) is the state transition density function of the k-th frame measurement data;

[0096] Value Function Value function I(X 1:k ) in the k-th frame value function I(x k ) is obtained as follows:

[0097]

[0098] Among them, X 1:K-1 is the target state set of the previous k-1 frames, I(x k ) is the value function of the kth frame, I(x k-1 ) is the value function of the k-1th frame;

[0099] Take the logarithm of both sides of formula (8) and convert p(z k |Z 1:k-1 ) as the coefficient normalization, and the value function of the dynamic programming algorithm is obtained; the expression is:

[0100]

[0101] Formula (9) describes the transfer of the value function from the k-1th frame to the kth frame. It is divided into two parts to implement the iterative update of the value function. One part is the log-likelihood function logp(z k |x k ), which is used to characterize the probability that the measurement area is from the target; the other part is the transfer cost function logp(x k |x k-1 ), used to describe the probability of the target moving in each direction between the k-1th frame and the kth frame;

[0102] Step 64: based on the augmented state information x of the kth frame k , initialize the target state, value function and track backtracking function of the previous n frames.

[0103] The other steps and parameters are the same as those in Specific Embodiments 1 to 4-1.

[0104] Specific implementation method 6: This implementation method is different from any one of the specific implementation methods 1 to 5 in that: in step 64, based on the augmented state information x of the kth frame k , initialize the target state, value function and track backtracking function of the previous n frames; the expression is:

[0105]

[0106] Where k = 1, 2, ..., n, θ k is the azimuth of the kth frame, p(z k |x k ) is the likelihood function; is the track tracing function.

[0107] The other steps and parameters are the same as those in Specific Implementation Methods 1 to 5-1.

[0108] Specific implementation method 7: This implementation method is different from any one of specific implementation methods 1 to 6 in that: in step 8, the value function obtained in step 6 is weighted by adding the azimuthal potential function to obtain a new value function;

[0109] Based on the new value function and azimuth potential function, the track tracing function of the mth target in the kth frame is obtained;

[0110] The specific process is:

[0111] Step 81: Based on the azimuthal potential coefficient of step 51, define the azimuthal potential function μ(X k-n:k-1 |X k-n+1:k );

[0112] Azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ) reflects the state set X k-n+1:k ={x k ,x k-1 ,x k-n+1}From the previous state set X k-n:k-1 ={x k-1 ,x k-2 ,x k-n}Probability size;

[0113] Azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ) is expressed as:

[0114]

[0115] Among them, μ(X k-n:k-1 |X k-n+1:k ) is the azimuthal potential function, Xk-n:k-1 is the state set of the knth to k-1th frames, X k-n+1:k is the state set from the k-n+1th to the kth frame, λ k is the azimuth potential coefficient of the kth frame, λ k-1 is the azimuth potential coefficient of the k-1th frame;

[0116] γ represents the variance of the Lorentz function. We can select an appropriate γ to construct the azimuthal potential function and perform value function weighting according to the situation.

[0117] For the value function I(x k+1 ) and I(x k ), their state variables should have similar azimuth potential coefficients, which will be reflected in the azimuth potential function to obtain a larger confidence value; while for different targets, their azimuth potential coefficients will vary greatly, which will be reflected in the azimuth potential function to obtain a smaller confidence value;

[0118] Step 82: For dynamic programming pre-detection tracking considering neighboring targets, based on the likelihood function p(z k |x k ), state transition density function p(x k |x k-1 ) and the azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), add the azimuthal potential function weight to the value function to obtain a new value function;

[0119] Based on the new value function and azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), and obtain the track tracing function of the mth target in the kth frame.

[0120] The other steps and parameters are the same as those in Specific Implementations 1 to 6-1.

[0121] Specific implementation eight: This implementation differs from any one of the specific implementations one to seven in that: in step 82, for the dynamic planning pre-detection tracking considering the neighboring targets, based on the likelihood function p(z k |x k ), state transition density function p(x k |x k-1 ) and the azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), add the azimuthal potential function weight to the value function to obtain a new value function;

[0122] Based on the new value function and azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), obtain the track tracing function of the mth target in the kth frame;

[0123] The specific process is:

[0124] The new value function is expressed as:

[0125]

[0126] Among them, τ(x k ) is the state that the target state can transfer to within one frame, p(x k |x k-1 ) is the state transition density function of the kth frame, I(x k-1 ) is the value function of the k-1 frame, p(z k |x k ) is the likelihood function;

[0127] Based on the new value function and azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), and obtain the track tracing function of the mth target in the kth frame. The track tracing function is expressed as:

[0128]

[0129] in, is the track tracing function of the mth target in the kth frame.

[0130] By weighting the value function and the track backtracking function through the azimuth potential function, the value function belonging to the same target will have a higher confidence, thereby enhancing the accumulated value of the value function of the target here.

[0131] The other steps and parameters are the same as those in Specific Implementation 1 to 7-1.

[0132] Specific implementation method 9: This implementation method is different from any one of specific implementation methods 1 to 8 in that: in step 10, the new value function obtained is compared with the decision threshold, the target of the frame corresponding to the value function less than the decision threshold is eliminated, and the target of the frame corresponding to the value function greater than or equal to the decision threshold is retained; the specific process is:

[0133]

[0134] in, represents the target state of the kth frame corresponding to the value function exceeding the threshold, Λ T Indicates the decision threshold.

[0135] The other steps and parameters are the same as those in Specific Embodiments 1 to 8.

[0136] Specific implementation method ten: This implementation method is different from any one of specific implementation methods one to nine in that: in step 11, based on the target of the frame corresponding to the value function retained in step 10 and the track backtracking function in step 8, the track backtracking is performed to obtain the mth target tracking track; the specific process is:

[0137] If the value function I(x K )≥Λ T , then for k=K-1,…,1, we have

[0138]

[0139] Among them, ψ k (x k ) represents the result of track backtracking;

[0140] If the value function I(x K )<Λ T , then no track backtracking is performed;

[0141] Take the result of track backtracking as the target state estimation information tracking result to obtain the estimated target trajectory set

[0142]

[0143] in Represents the target state estimation information tracking results from the 1st to the Kth frame.

[0144] The other steps and parameters are the same as those in Specific Embodiments 1 to 9.

[0145] Experimental Section

[0146] To verify the tracking performance of the present invention in the scenario of weak underwater acoustic target proximity. The receiving sonar is a passive 40-element uniform linear array with a half-wavelength spacing between elements and an operating frequency band of f 0 =500Hz~2500Hz, signal-to-noise ratio -25dB, data length 120 frames, three targets are set in the simulation scene, of which target 1 and target 2 are adjacent, and their trajectories cross at the 40th frame, and target 3 is separated. The spatial spectrum after conventional beamforming is shown in the figure.

[0147] The APWDP-TBD (the present invention) and the traditional DP-TBD algorithm are used to track the simulation target respectively, with the state transfer value q=21, the resolution unit size 0.1 degree, and the DWPD-TBD azimuth potential coefficient window length set to 2. The tracking results are compared, and the optimal sub-mode allocation error (OSPA error) is selected as the evaluation index of the simulation.

[0148] The DP-TBD algorithm does not consider the proximity of targets, so when the tracks of target 1 and target 2 intersect, a track mismatch problem occurs, causing its OSPA error to increase after 40 frames, and the tracking performance of the algorithm decreases. The APWDP-TBD proposed in the present invention can obtain correct tracking results for the adjacent intersections of targets 1 and 2 and the tracks of target 3, and can significantly improve the tracking performance of the algorithm in the target proximity scenario, and can significantly reduce the OSPA error caused by track mismatch and other problems.

[0149] The calculation example given in the present invention is only an example of the calculation method and process of the present invention, and is not a limited implementation method of the algorithm and process of the present invention. For those skilled in the art, on the basis of the present invention, according to the present invention, some implementation schemes that cannot be listed in the present invention can be made, and these corresponding implementation scheme changes or deformations should still fall within the protection scope of the claims given in the present invention.

Claims

1. A method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting, characterized in that: The specific process of the method is: Step 1, beamforming the passive sonar detection signal to obtain spatial spectrum data; The spatial spectrum data is used as the measurement data, and the azimuth observation range of each frame in the measurement data is divided into B resolution units; Step 2, initialize the state transition number q; When the number of state transitions is q, the kth frame in the measurement data is transferred to the k+1th frame, and the target state information x of the k+1th frame is obtained. k+1 ; Step 3, initialize the frame number k in the measurement data to 1; Step 4, initialize the target count m=1; Step 5: Define the azimuth potential coefficient, and define the augmented state information including the azimuth potential coefficient and the target azimuth information based on the azimuth potential coefficient Step 6: Based on the measurement data from step 1 and the target state information x from step 2 k+1 , the augmented state information containing the azimuth potential coefficient and target azimuth information in step 5 Initialize the augmented state information of the first n frames with the same length as the window length n of the azimuth potential coefficient Get the value function and track backtracking function; Step 7: Determine whether k>n. If yes, execute step 8; if no, set k=k+1 and repeat steps 6 to 7. Step 8, adding the azimuthal potential function weight to the value function obtained in step 6 to obtain a new value function; Based on the new value function and azimuth potential function, the track tracing function of the mth target in the kth frame is obtained; Step 9: Determine if k=K. If yes, execute step 10. If no, set k=k+1 and repeat steps 8 to 9. K is the total number of frames in the measurement data; Step 10: Compare the obtained new value function with the decision threshold, remove the targets of the frames corresponding to the value function less than the decision threshold, and retain the targets of the frames corresponding to the value function greater than or equal to the decision threshold; Step 11, based on the target of the frame corresponding to the value function retained in step 10 and the track backtracking function of step 8, the track backtracking is performed to obtain the mth target tracking track; Step 12: determine if m=M, and if so, obtain M target tracking trajectories; If not, set m=m+1 and repeat steps 8 to 12 until m=M, and obtain M target tracking trajectories.

2. The method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting according to claim 1 is characterized in that: In the step 1, beamforming is performed on the passive sonar detection signal to obtain spatial spectrum data; The spatial spectrum data is used as the measurement data, and the azimuth observation range of each frame in the measurement data is divided into B resolution units; The specific process is: Step 11, initialize the number of linear array elements, element spacing, sampling frequency, sound velocity, and frame number; Step 12: beamforming the passive sonar detection signal to obtain spatial spectrum data; Step 13: Use the spatial spectrum data as measurement data, denoted as Z 1:K ={z1,z2,...,z k ,...,z K }; Among them, Z 1:K is the measurement data set, z1 is the first frame in the measurement data, z2 is the second frame in the measurement data, and z k is the kth frame in the measurement data, z K is the Kth frame in the measurement data, K is the total number of frames in the measurement data; Step 14, dividing the azimuth observation range of each frame in the measurement data into B resolution units, and the size of each resolution unit is Δθ; Use z k (b) represents the amplitude of the measured data at the b-th resolution unit in the k-th frame, and the amplitudes of the measured data in all resolution units are independent and identically distributed, b = 1, 2, ..., B; z k (b) is expressed as: Among them, ω k (b) represents the noise amplitude of the b-th resolution unit in the k-th frame, A k (b) represents the target signal amplitude of the b-th resolution unit in the k-th frame.

3. The method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting according to claim 2 is characterized in that: Initialize the state transition number q in step 2; When the number of state transitions is q, the kth frame in the measurement data is transferred to the k+1th frame, and the target state information x of the k+1th frame is obtained. k+1 ; The specific process is: Initialize the state transition number q; When the number of state transitions is q, the kth frame in the measurement data transitions to the k+1th frame, and the target state information x of the k+1th frame is k+1 for: Among them, θ k is the orientation information of the kth frame, Δθ is the size of the resolution unit, δ q is the state transition range when the state transition number is q; x k+1 Indicates the target state information of the k+1th frame.

4. The method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting according to claim 3 is characterized in that: In step 5, the azimuth potential coefficient is defined, and augmented state information including the azimuth potential coefficient and target azimuth information is defined based on the azimuth potential coefficient. The specific process is: Step 51: Definition Represents the azimuth potential coefficient of the mth target in the kth frame; it is expressed as: Where n is the window length of the azimuth potential coefficient, m is the target count, ω i is the weight of the i-th frame, θ i is the orientation information of the i-th frame, θ i-1 is the orientation information of the i-1th frame; Step 52: Based on the azimuth potential coefficient of the mth target in the kth frame Define the augmented state information of the mth target in the kth frame, which includes the azimuth potential coefficient and the target azimuth information; it is expressed as: in, Indicates the target position information of the mth target in the kth frame; Represents the augmented state information of the mth target in the kth frame.

5. The method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting according to claim 4 is characterized in that: In step 6, the measured data in step 1 and the target state information x in step 2 are used as the basis. k+1 , the augmented state information containing the azimuth potential coefficient and target azimuth information in step 5 Initialize the augmented state information of the first n frames with the same length as the window length n of the azimuth potential coefficient Get the value function and track backtracking function; The specific process is: Step 61: Express the state transfer equation and measurement equation of the target as follows: Among them, f k-1 (·) represents the state transfer function, h k (·) represents the measurement function, x k represents the augmented state information of the kth frame, x k-1 represents the augmented state information of the k-1th frame, w k represents the state transition noise, v k represents the measurement noise, which follows Gaussian distribution, z k represents the measurement data of the kth frame; Step 62: The target sequence with the largest posterior probability Right now: in, is the augmented state information of the target with the largest posterior probability in the first frame, is the augmented state information of the target with the largest posterior probability density in the second frame, is the augmented state information of the target with the largest posterior probability in the kth frame, is the target state set with the largest posterior probability in the previous k frames, is the maximum a posteriori probability density function, p(X 1:k |Z 1:k ) is the posterior probability density function; Step 63: Perform Bayesian transformation on equation (6) in step 62 to obtain: Among them, p(X 1:k |Z 1:k ) is the posterior probability density function of the first k frames, X 1:k is the target state set of the previous k frames, Z 1:k is the measurement data set, p(z k |x k ) is the likelihood function of the kth frame, z k represents the measurement data of the kth frame, x k is the target state information of the kth frame, p(x k |x k-1 ) is the state transition density function, p(X 1:k-1 |Z 1:k-1 ) is the posterior probability density function of the previous k-1 frames, p(z k |Z 1:k-1 ) is the state transition density function of the k-th frame measurement data; Value Function Value function I(X 1:k ) in the k-th frame value function I(x k ) is obtained as follows: Among them, X 1:K-1 is the target state set of the previous k-1 frames, I(x k ) is the value function of the kth frame, I(x k-1 ) is the value function of the k-1th frame; Take the logarithm of both sides of formula (8) and convert p(z k |Z 1:k-1 ) as the coefficient normalization to obtain the value function; the expression is: Step 64: based on the augmented state information x of the kth frame k , initialize the target state, value function and track backtracking function of the previous n frames.

6. The method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting according to claim 5 is characterized in that: In step 64, based on the augmented state information x of the kth frame k , initialize the target state, value function and track backtracking function of the previous n frames; the expression is: Where k = 1, 2, ..., n, θ k is the azimuth of the kth frame, p(z k |x k ) is the likelihood function; is the track tracing function.

7. The method for underwater acoustic target tracking before detection based on dynamic programming and weighted azimuth potential function according to claim 6 is characterized in that: In step 8, the value function obtained in step 6 is weighted by adding the azimuthal potential function to obtain a new value function; Based on the new value function and azimuth potential function, the track tracing function of the mth target in the kth frame is obtained; The specific process is: Step 81: Based on the azimuthal potential coefficient of step 51, define the azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ); Azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ) is expressed as: Among them, μ(X k-n:k-1 |X k-n+1:k ) is the azimuthal potential function, X k-n:k-1 is the state set of the knth to k-1th frames, X k-n+1:k is the state set from the k-n+1th to the kth frame, λ k is the azimuth potential coefficient of the kth frame, λ k-1 is the azimuth potential coefficient of the k-1th frame; γ represents the variance of the Lorentz function; Step 82: Based on the likelihood function p(z k |x k ), state transition density function p(x k |x k-1 ) and the azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), add the azimuthal potential function weight to the value function to obtain a new value function; Based on the new value function and azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), and obtain the track tracing function of the mth target in the kth frame.

8. The method for underwater acoustic target tracking before detection based on dynamic programming and weighted azimuth potential function according to claim 7 is characterized in that: In step 82, based on the likelihood function p(z k |x k ), state transition density function p(x k |x k-1 ) and the azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), add the azimuthal potential function weight to the value function to obtain a new value function; Based on the new value function and azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), obtain the track tracing function of the mth target in the kth frame; The specific process is: The new value function is expressed as: Among them, τ(x k ) is the state that the target state can transfer to within one frame, p(x k |x k-1 ) is the state transition density function of the kth frame, I(x k-1 ) is the value function of the k-1 frame, p(z k |x k ) is the likelihood function; Based on the new value function and azimuthal potential function μ(X k-n:k-1 |X k-n+1:k ), and obtain the track tracing function of the mth target in the kth frame. The track tracing function is expressed as: in, is the track tracing function of the mth target in the kth frame.

9. The method for underwater acoustic target tracking before detection based on dynamic programming of azimuth potential function weighting according to claim 8 is characterized in that: In step 10, the obtained new value function is compared with the decision threshold, the target of the frame corresponding to the value function less than the decision threshold is eliminated, and the target of the frame corresponding to the value function greater than or equal to the decision threshold is retained; the specific process is: in, represents the target state of the kth frame corresponding to the value function exceeding the threshold, Λ T Indicates the decision threshold.

10. The method for underwater acoustic target tracking before detection based on dynamic programming and weighted azimuth potential function according to claim 9, characterized in that: In step 11, the target of the frame corresponding to the value function retained in step 10 and the track backtracking function in step 8 are used to perform track backtracking to obtain the mth target tracking track; the specific process is: If the value function I(x K )≥Λ T , then for k=K-1,…,1, we have Among them, ψ k (x k ) represents the result of track backtracking; If the value function I(x K )<Λ T , then no track backtracking is performed; Take the result of track backtracking as the target state estimation information tracking result to obtain the estimated target trajectory set in Represents the target state estimation information tracking results from the 1st to the Kth frame.

Images