A trajectory poisson multi-bernoulli mixture filter based coherent matched field tracking method and system thereof

Abstract

The application discloses a kind of based on trajectory poisson multi bernoulli hybrid filtering coherent matching field tracking method and system, belong to underwater acoustic technical field.Extract local peak value from the multiple frame ambiguity surface obtained from coherent matching field processing and carry out clustering, construct three-dimensional measurement set including target distance, depth and radial velocity estimation information;On this basis, establish matching field trajectory state space model, utilize trajectory poisson multi bernoulli hybrid filter in trajectory space to carry out state prediction, and update predicted result in combination with three-dimensional measurement set;Realize data association based on the likelihood consistency of distance, depth and radial velocity, to complete target trajectory estimation.The application is extracted to the peak value of three-dimensional ambiguity surface by clustering processing, effectively reduces the large number of highly similar redundant trajectories near real trajectory, reduces the calculation burden of algorithm and the uncertainty of data association.The matching field tracking problem under low signal-to-noise ratio environment is solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of underwater acoustic, and particularly relates to a coherent matched field tracking method based on trajectory Poisson multiple Bernoulli mixture filter and a system thereof. BACKGROUND

[0002] For moving targets, matched field processing (MFP) usually processes long-time data frame by frame and inverses target position frame by frame, and the series of independent optimal positioning results lack spatial and temporal correlation and are difficult to directly form the optimal motion trajectory of the target. Matched field tracking (MFT) is an extension of matched field processing, which effectively correlates multi-frame positioning results by introducing target motion state to realize continuous tracking of the target trajectory.

[0003] Early MFT presets all possible trajectories based on simplified target motion assumption or MFP ambiguity surface peak value, and determines the optimal trajectory by summing or averaging the ambiguity values corresponding to each trajectory. Method one proposes a matched field tracking method based on ambiguity surface averaging, which converts the position ambiguity function into a trajectory ambiguity function by setting the start and end points of the trajectory, and then determines the optimal trajectory by means of an optimization algorithm. However, this method relies on the assumption of uniform linear motion of the target, and has poor applicability in complex motion scenarios. Method two combines the motion model of the sound source with the prior information of the uncertain ocean environment, and proposes an optimal uncertainty field tracking algorithm based on the motion model of the target. The algorithm models the motion of the sound source as a Markov process, and regards each possible position of the sound source as a state of the process, and takes the maximum posterior probability path as the optimal trajectory. The multi-target Bayesian filtering method based on random finite set also provides a new research idea for matched field tracking. Method three uses trajectory Poisson multiple Bernoulli mixture filter (TPMBM) to combine the distance likelihood of the peak position of the ambiguity function with the peak amplitude likelihood to realize continuous matched field tracking. However, the above methods are currently mainly applicable to high signal-to-noise ratio conditions. The matched field tracking method based on Bayesian filtering relies on the construction of the measurement set by the MFP ambiguity surface, and in a low signal-to-noise ratio environment, the performance of the traditional MFP decreases significantly, resulting in insufficient quality of the measurement set and difficulty in realizing effective tracking. So far, there is still a lack of a robust method that can realize matched field tracking in a low signal-to-noise ratio environment. SUMMARY

[0004] The application provides a coherent matched field tracking method based on trajectory Poisson multiple Bernoulli mixture filter and a system thereof, which aims at the problem of target matched field tracking in a low signal-to-noise ratio environment.

[0005] The application is implemented by the following technical solutions:

[0006] A coherent matched field tracking method based on trajectory Poisson multiple Bernoulli mixture filter, the tracking method comprising the following steps:

[0007] Step 1: divide the long time series signal collected by the hydrophone receiving array into multiple frames and convert to frequency domain by Fourier transform;

[0008] Step 2: for each frame of frequency domain signal processed in step 1, calculate the three-dimensional ambiguity surface about distance, depth and radial velocity by coherent matched field algorithm;

[0009] Step 3: extract local peaks from each frame of ambiguity surface obtained in step 2, and the number of peaks obeys Poisson distribution;

[0010] Step 4: based on the local peaks of each frame of ambiguity surface obtained in step 3, perform clustering by non-maximum suppression algorithm, and only keep multiple peaks of the cluster center;

[0011] Step 5: based on the distance, depth and radial velocity estimated values corresponding to the multiple peaks of the cluster center obtained in step 4, construct a three-dimensional measurement set;

[0012] Step 6: based on the trajectory Poisson multi-Bernoulli hybrid filter modeling method and the state variables of the coherent matched field measurement set, establish the matched field trajectory state space model and measurement model;

[0013] Step 7: combine the matched field state space model and measurement model obtained in step 6 with the trajectory Poisson multi-Bernoulli hybrid filter to realize state prediction, update and data association in trajectory space, and complete trajectory estimation.

[0014] Further, the step 2 is specifically based on the frame of frequency domain signal processed in step 1, and a series of snapshots generated when the sound source is moved to different positions are coherently integrated; assuming that a frequency K is collected f , the distance and depth of the sound source at each time L are and , respectively, then the snapshot at this time is expressed as:

[0015]

[0016] wherein, represents noise from the sensor or the interference source, is the theoretical sound field, is the frequency f spectrum of the sound source;

[0017] For an ideal narrowband sound source, ; the coherent matched field removes the sound source phase from each snapshot, thereby obtaining a new measurement sequence : ​

[0018]

[0019] For candidate sound source trajectories The model will generate sequences The measurement dataset will then be... With model dataset Each is combined into a hypervector, that is

[0020]

[0021] in, Represents the hypervector measurement field. This represents the transpose of the new measurement sequence in the first snapshot. Indicates the first L A quick snapshot of the transpose of a new measurement sequence Represents a supervector copy field. This indicates the transpose of the copy field at the first snapshot position. Indicates the first L The transpose of the copy field at the snapshot location. This indicates the copy field distance and depth corresponding to the first snapshot. Indicates the first L The copy field distance and depth corresponding to each snapshot;

[0022] cross-spectral density matrix Represented as:

[0023]

[0024] in, Indicates the number of snapshots. Indicates the first s The conjugate transpose of a snapshot of a hypervector measurement field Indicates the first s A snapshot of the hypervector measurement field;

[0025] Output power of the coherent matched field at time 1 Represented as:

[0026]

[0027] in, Indicates the inversion parameters. These represent distance, depth, and radial velocity, respectively. Indicates frequency as f The hypervector copy field at time, This represents the conjugate transpose of a matrix. express Norm, Represents the trace of a matrix;

[0028] For broadband signals, the output power of the coherent matched field is expressed as the incoherent average of the output power of the narrowband signal, i.e.

[0029]

[0030] in, Indicates the number of frequency points. Indicates the first k The 3D fuzzy surface at a given time. Indicates the frequency point number. Represents frequency The corresponding three-dimensional fuzzy surface.

[0031] Furthermore, step 3 specifically involves extracting information from each three-dimensional ambiguity function. Extract Local maxima The number of them follows a Poisson distribution. ; for The parameter vectors corresponding to the local maxima, where .

[0032] Furthermore, step 4 specifically involves using a nonmaximum suppression algorithm to... Clustering is performed on the peaks, assuming the initial neighborhood radius of the cluster is... ,in, Represents the radius of the neighborhood in the distance dimension. Represents the radius of the depth-dimensional neighborhood. Representing the radial velocity radius, the peak points are sorted in descending order of amplitude. For the current peak point... ,in, Representing the k Time of the first c Distance inversion values ​​corresponding to each cluster center Representing the k Time of the first c The depth inversion value corresponding to each cluster center Representing the k Time of the first c The radial velocity inversion value corresponding to each cluster center is used to calculate the velocity relative to other points. Weighted Euclidean distance:

[0033]

[0034] If and only if hour, Belongs to the current cluster, for those containing Clustering of points and their corresponding amplitudes , cluster center is:

[0035]

[0036] wherein, is a cluster center serial number, represents the number of cluster centers, represents the parameter vector corresponding to the k-th cluster center, represents the amplitude of the k-th cluster center.

[0037] Further, the step 5 is specifically to only keep the distance, depth and radial velocity estimation values corresponding to the peak position of each frame cluster center as the three-dimensional measurement values of coherent matching field tracking.

[0038] Further, the step 6 is specifically that the trajectory state of each underwater moving sound source is represented by a random finite set; wherein the target state is composed of the sound source distance , distance dimension velocity , sound source depth and depth dimension velocity , the dimension , the transition relationship of the target state between adjacent steps is given by the state equation:

[0039]

[0040] wherein, represents a time interval, represents a state noise variance, represents a state transition matrix, represents a noise matrix, represents the distance at time k -1, represents the depth at time k -1, represents the distance dimension velocity at time k -1, represents the depth dimension velocity at time k -1, the state equation is used to predict k the state at time k from the state at time -1;

[0041] The mapping relationship of is given by the measurement equation:

[0042] ​

[0043] wherein, represents the three-dimensional measurement value in the actual application, i.e. , represents a mapping matrix of the measurement vector and the state vector, represents measurement noise, represents distance dimension measurement noise, represents depth dimension measurement noise, represents radial velocity dimension measurement noise.

[0044] Further, the step 7 is specifically, k trajectory state estimation at time t is expressed as a probability of existence a trajectory state set with a threshold value and a trajectory state set that has ended before time t:

[0045]

[0046] wherein, represents a trajectory state mean with a probability of existence a threshold value, and the number of inequalities that are established is trajectory quantity estimation value that still exists at time t, represents a trajectory state mean set that has ended before time t, represents a maximum value of a posterior global association hypothesis probability, represents a union set, represents a trajectory serial number.

[0047] A coherent matched field tracking system based on a trajectory Poisson multi-Bernoulli hybrid filter, the tracking system uses a coherent matched field tracking method based on a trajectory Poisson multi-Bernoulli hybrid filter as described above, and the tracking system comprises the following steps:

[0048] A data acquisition processing module: dividing a long-time sequence signal acquired by a hydrophone receiving array into multiple frames and converting to a frequency domain through Fourier transform;

[0049] A calculation module: for each frame of frequency domain signal processed by the data acquisition processing module, a three-dimensional ambiguity surface about distance, depth and radial velocity is calculated through a coherent matched field algorithm;

[0050] A local peak extraction module: extracting local peaks from each frame of ambiguity surface obtained by the calculation module, and the number of peaks obeys a Poisson distribution;

[0051] ​The clustering module: based on the local peak value of each frame of the ambiguity surface obtained by the local peak value extraction module 3, the non-maximum suppression algorithm is used for clustering, and only the multiple peak values of the cluster center are reserved;

[0052] The three-dimensional measurement set construction module: based on the distance, depth and radial velocity estimation value corresponding to the multiple peak values of the cluster center obtained by the clustering module, a three-dimensional measurement set is constructed;

[0053] The trajectory tracking module: based on the trajectory Poisson multi-Bernoulli mixture filter modeling method and the coherent matching field measurement set state variable, a matching field trajectory state space model and a measurement model are established;

[0054] The display module: the matching field state space model and the measurement model obtained by the trajectory tracking module are combined with the trajectory Poisson multi-Bernoulli mixture filter to realize state prediction, update and data association in the trajectory space, and trajectory estimation is completed.

[0055] A computer device, comprising 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 realize the method as described above.

[0056] A computer readable storage medium, the computer readable storage medium stores a computer program, the computer program is executed by a processor to realize the method as described above.

[0057] The beneficial effects of the present application are:

[0058] The present application effectively reduces a large number of highly similar redundant trajectories near the real trajectory by clustering the extracted peak values of the three-dimensional ambiguity surface, and reduces the computational burden of the algorithm and the uncertainty of data association.

[0059] The present application can significantly alleviate the trajectory breakage and false association phenomenon in target tracking by introducing radial velocity information as additional measurement.

[0060] The present application utilizes the high space-time gain advantage of coherent matching field processing to construct a better measurement set in a low signal-to-noise ratio environment.

[0061] The present application utilizes the radial velocity estimation value of the coherent matching field as additional measurement to reduce the trajectory breakage and false association in trajectory tracking.

[0062] The present application reduces a large number of highly similar redundant trajectories near the real trajectory by clustering the extracted peak values of the three-dimensional ambiguity surface, and reduces the computational burden of the algorithm and the uncertainty of data association.

[0063] The application combines the coherent matching field processing and the trajectory Poisson multi-Bernoulli hybrid filtering algorithm, and improves the target tracking accuracy and robustness in a low signal-to-noise ratio environment. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 is a method flowchart of the application.

[0065] Figure 2 is a result graph of local peak clustering achieved by the non-maximum suppression algorithm of the application.

[0066] Figure 3 is a three-dimensional measurement set result constructed by the application in a low signal-to-noise ratio environment, which is a distance, depth and radial velocity measurement history graph, respectively.

[0067] Figure 4 is a tracking trajectory result graph of a target distance trajectory and a tracking trajectory result graph of a depth trajectory of the application in a low signal-to-noise ratio environment.

[0068] Figure 5 is a target distance trajectory tracking history graph of the application in a low signal-to-noise ratio environment. DETAILED DESCRIPTION

[0069] In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular architectures, technologies, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

[0070] It should be understood that the term "comprises" when used in this specification and the appended claims, specifies the presence of stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0071] It should also be understood that the terms used in the specification and the appended claims are intended to describe particular embodiments and do not intend to limit the present application. As used in the specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0072] The following will be described in conjunction with the accompanying drawings of the present application Figures 1-5The technical solutions in the embodiments of the present application are clearly and completely described. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0073] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be practiced in other ways different from those described herein without departing from the scope of the present application, and those skilled in the art can make similar extensions without departing from the connotation of the present application, so the present application is not limited to the specific embodiments disclosed below.

[0074] Embodiment one

[0075] The present embodiment provides a coherent matched field tracking method based on trajectory Poisson multi-Bernoulli hybrid filter, as shown in Figure 1 The tracking method comprises the following steps:

[0076] Step 1: Divide the long-time sequence signal collected by the hydrophone receiving array into multiple frames and convert to the frequency domain through Fourier transform;

[0077] Step 2: For each frame of frequency domain signal processed in step 1, calculate a three-dimensional ambiguity surface about distance, depth and radial velocity through the coherent matched field algorithm;

[0078] Step 3: Extract local peaks from each frame of ambiguity surface obtained in step 2, and the number of peaks obeys Poisson distribution;

[0079] Step 4: Based on the local peaks of each frame of ambiguity surface obtained in step 3, cluster through the non-maximum suppression algorithm, and only keep multiple peaks of the cluster center;

[0080] Step 5: Based on the distance, depth and radial velocity estimation values corresponding to the multiple peaks of the cluster center obtained in step 4, construct a three-dimensional measurement set;

[0081] Step 6: Based on the modeling method of trajectory Poisson multi-Bernoulli hybrid filter and the state variables of coherent matched field measurement set, establish a matched field trajectory state space model and a measurement model;

[0082] Step 7: Combine the matched field state space model and the measurement model obtained in step 6 with the trajectory Poisson multi-Bernoulli hybrid filter to realize state prediction, update and data association in the trajectory space, and complete trajectory estimation.

[0083] Further, the step 2 is specifically based on the frequency domain signal processed in step 1 Kframe frequency domain signals, a series of snapshots produced when moving the sound source to different positions are coherently integrated; assume that the sound source is moved to different positions at frequency f A set of L snapshots are collected, at each time , the distance and depth of the sound source are and respectively, then the snapshot at this time is expressed as:

[0084]

[0085] where represents the noise from the sensor or the interference source, etc. is the theoretical sound field, is the sound source spectrum at frequency f ;

[0086] For an ideal narrowband sound source, ; the coherent matching field removes the sound source phase from each snapshot, thus obtaining a new measurement sequence :

[0087]

[0088] For a candidate sound source trajectory , the model will generate a sequence ; then the measurement dataset and the model dataset are combined into a hyper vector respectively, that is

[0089]

[0090] where represents the hyper vector measurement field, represents the transpose of the new measurement sequence of the first snapshot, represents the transpose of the new measurement sequence of the L th snapshot, represents the hyper vector copy field, represents the transpose of the copy field at the position of the first snapshot, represents the transpose of the copy field at the position of the L th snapshot, represents the distance and depth of the copy field corresponding to the first snapshot, represents the distance and depth of the copy field corresponding to the L th snapshot;

[0091] The cross-spectral density matrix is expressed as:

[0092]

[0093] wherein, denotes the number of snapshots, denotes the conjugate transpose of the hyper vector measurement field of the s th snapshot, denotes the hyper vector measurement field of the s th snapshot;

[0094] the output power of the coherent matched field at time is denoted as:

[0095]

[0096] wherein, denotes the inversion parameters, denote the range, depth and radial velocity, respectively, denotes the hyper vector copy field at frequency f denotes the conjugate transpose of a matrix, denotes norm, denotes the trace of a matrix;

[0097] For wideband signals, the output power of the coherent matched field is denoted as the incoherent average of the output power of narrowband signals, i.e.

[0098]

[0099] wherein, denotes the number of frequency bins, denotes the three-dimensional ambiguity surface at the k th time instant, denotes the frequency bin index, denotes the frequency corresponding three-dimensional ambiguity surface.

[0100] Further, the step 3 is specifically extracting local maximum points from each three-dimensional ambiguity function , the number of which obeys the Poisson distribution ; is the parameter vector corresponding to the local maximum peak points, wherein .

[0101] Further, the step 4 is specifically clustering the peaks by a non-maximum suppression algorithm, assuming that the initial neighborhood radius of clustering is , wherein, represents the range dimension neighborhood radius, represents the depth dimension neighborhood radius, ​representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point wherein, representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point k representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point c representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point k representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point c representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point k representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point c representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point representing the radial velocity dimension radius, arranging the peak points in descending order of amplitude, for the current peak point

[0102]

[0103] iff , belongs to the current cluster, for a cluster containing points and its corresponding amplitude , the cluster center is:

[0104]

[0105] wherein, is the cluster center serial number, represents the number of cluster centers, represents the parameter vector corresponding to the th cluster center at time k, represents the amplitude of the th cluster center.

[0106] Further, the step 5 is specifically to only keep the distance, depth and radial velocity estimation values corresponding to the peak position of each frame cluster center as the three-dimensional measurement values of coherent matching field tracking.

[0107] Further, the step 6 is specifically that each underwater moving sound source track state is represented by a random finite set; wherein, the target state is composed of sound source distance , distance dimension velocity , sound source depth and depth dimension velocity , the dimension , the transition relationship of the target state between adjacent steps is given by the state equation:

[0108] ​

[0109] in, Represents a time interval. Represents the variance of state noise. Represents the state transition matrix. Represents the noise matrix. represent k The distance at time -1 represent k The depth at time -1 represent k Velocity in the distance dimension at time -1 represent k The depth-dimensional velocity at time -1, the state equation is used according to k State prediction at time -1 k The state at any given moment.

[0110] and The mapping relationship is given by the measurement equation:

[0111]

[0112] in, Represents the three-dimensional measurement values ​​used in practical applications, i.e. = , The mapping matrix representing the measurement vector and the state vector, Represents measurement noise. Represents distance dimension measurement noise. Represents depth dimension measurement noise. This represents the noise in the radial velocity dimension measurement.

[0113] Furthermore, step 7 specifically involves using the trajectory trajectory state space model, measurement model, and three-dimensional measurement set constructed in step 6 to perform prediction, update, and data association in the trajectory space through a trajectory Poisson-Dobernoli hybrid filtering algorithm.

[0114] k Time-Trajectory State Estimation Represented as the probability of existence The set of trajectory states of the threshold and The set of trajectory states that have ended before the specified time:

[0115]

[0116] in, Represents the probability of existence The mean of the trajectory state of the threshold, and the number of inequalities that hold true. for The number of track estimates that are still alive at the moment, represent The mean set of track states that have ended before the moment, represent the maximum value of the posterior global association hypothesis probability, represent the union set, represent the track serial number.

[0117] As Figure 1 shown, first, preset the distance, depth and radial velocity search interval, based on the known sound source emission frequency information, calculate the three-dimensional copy field through the sound propagation model, and construct the synthetic copy field through the synthetic aperture idea. For the long time sequence signal received by the vertical array, divide it into multiple frames with a certain overlap degree according to the covariance integration time, and perform Fourier transform to the frequency domain, and construct the synthetic measurement field after phase compensation.

[0118] Then, the spatial correlation of the synthetic copy field and the synthetic measurement field is performed, and the three-dimensional ambiguity surface of each frame is calculated. Extract several local peaks from each frame ambiguity surface, and the number of peaks obeys the Poisson distribution with a set mean. Through the non-maximum suppression algorithm, the local peak values are clustered, as Figure 2 shown, only the peak information of the cluster center is retained. Based on the sound source distance, depth and radial velocity estimates corresponding to the cluster center, a three-dimensional measurement set is constructed, as Figure 3 shown, respectively, the distance, depth and radial velocity measurement history diagram in a low signal-to-noise ratio environment.

[0119] Finally, the track space state model and the measurement model are constructed, and the target state estimation, update and data association are realized based on the constructed three-dimensional measurement set and the track Poisson multi-Bernoulli hybrid filtering algorithm, so as to realize the continuous tracking of the target. Figure 4 For the tracking result of the target track, the upper and lower figures are the tracking trajectories of the target distance track and the depth track respectively, and it can be seen that the method proposed in the application can realize continuous and accurate tracking of the target track in a low signal-to-noise ratio environment. Figure 5 For the history diagram of the target distance track tracking.

[0120] Embodiment two

[0121] The embodiment provides a coherent matched field tracking system based on track Poisson multi-Bernoulli hybrid filtering, which uses a coherent matched field tracking method based on track Poisson multi-Bernoulli hybrid filtering as described in embodiment one, and the tracking system comprises:

[0122] A data preprocessing module: the long time sequence signal collected by the hydrophone receiving array is divided into multiple frames and converted to the frequency domain through Fourier transform.

[0123] Coherent matched field processing module: input sound source distance, depth and radial velocity parameter search interval, sound source emission frequency and matching type (linear matching, synthetic adaptive matching), calculate the copy field based on the synthetic aperture technology to form a synthetic copy field. Using the coherence of multi-frame signals, the vertical array receiving signal phase compensation processing and stacking are used to form a synthetic measurement field. Further based on the synthetic copy field and the synthetic measurement field, the spatial correlation between the two is calculated, and finally the three-dimensional ambiguity surface corresponding to each frame is output.

[0124] Measurement set generation module: input the mean value of the Poisson distribution, extract a number of local peaks subject to the Poisson distribution from each frame ambiguity surface, and cluster these peaks by a non-maximum suppression algorithm. The distance, depth and radial velocity estimates corresponding to the cluster center peak values are constructed into a three-dimensional measurement set.

[0125] Trajectory tracking module: input trajectory Poisson multi-Bernoulli filter parameters (state noise mean and covariance, measurement noise mean and covariance, detection probability, existence probability threshold, etc.), establish a trajectory space state model and a measurement model, and realize trajectory tracking based on the trajectory Poisson multi-Bernoulli filter.

[0126] Display module: output peak clustering results, measurement set ambiguity sequence and trajectory tracking history diagram.

[0127] Embodiment three

[0128] The electronic device provided in the embodiments of the present application includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the memory is configured to store software programs and modules, and the processor is configured to execute various functional applications and data processing by running the software programs and modules stored in the memory. The memory and the processor are connected through a bus. Specifically, the processor realizes any step in the above embodiment one by running the above computer program stored in the memory.

[0129] It should be understood that, in the embodiments of the present application, the processor can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0130] The memory can include read-only memory, flash memory, and random access memory, and provide the processor with instructions and data. Part or all of the memory can also include non-volatile random access memory.

[0131] It should be understood that the above-mentioned integrated modules / units, if implemented in the form of software functional units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiments can also be completed by computer programs instructing related hardware, and the above-mentioned computer programs can be stored in a computer readable storage medium. When the processor executes the computer programs, the steps of the above-mentioned various method embodiments can be implemented. The above-mentioned computer programs include computer program codes, which can be in the form of source code, object code, executable files or some intermediate forms, etc. The above-mentioned computer readable medium can include any entity or device capable of carrying the above-mentioned computer program codes, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier wave signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the above-mentioned computer readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0132] The above description of disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0133] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the above-mentioned device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be realized in the form of hardware or software. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0134] It should be noted that the method and its details provided by the above-mentioned embodiments can be combined into the device and equipment provided by the embodiments, and mutual reference will not be described here.

[0135] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0136] In the embodiments provided by the present application, it should be understood that the disclosed device / terminal equipment and method can be implemented in other ways. For example, the above-mentioned device / device embodiment is only schematic, for example, the division of the above-mentioned modules or units is only a logical function division, and actual implementation can be in another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0137] The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A coherent matched field tracking method based on trajectory Poisson-Dobernoli hybrid filtering, characterized in that, The tracking method includes the following steps: Step 1: Divide the long-sequence signal collected by the hydrophone receiving array into multiple frames and convert it to the frequency domain using Fourier transform; Step 2: For each frame of frequency domain signal obtained in Step 1, a three-dimensional ambiguity surface with respect to distance, depth, and radial velocity is calculated using the coherent matched field algorithm; Step 3: Extract local peaks from the blur surface of each frame obtained in Step 2. The number of peaks follows a Poisson distribution. Step 4: Based on the local peak values ​​of the blur surface of each frame obtained in Step 3, clustering is performed using a non-maximum suppression algorithm, retaining only the peak values ​​at the cluster centers; Step 5: Based on the distance, depth, and radial velocity estimates corresponding to the multiple peaks of the cluster centers obtained in Step 4, construct a three-dimensional measurement set; Step 6: Based on the trajectory Poisson-Bernoulli hybrid filter modeling method and the state variables of the coherent matched field measurement set, establish the matched field trajectory state space model and measurement model; Step 7: Combine the matching field state space model and measurement model obtained in Step 6 with a trajectory Poisson-Dobernoli hybrid filter to achieve state prediction, update and data association in the trajectory space, and complete trajectory estimation; Step 2 specifically involves, Based on the processing in step 1 K The frame rate domain signal coherently integrates a series of snapshots generated when the sound source moves to different positions; assuming frequency... Collected L A series of quick shots, at each moment The distance and depth of the sound source are respectively and Then the snapshot at that moment Represented as: in, This indicates noise originating from a sensor or interference source. For the theoretical sound field, It is frequency The sound source spectrum at that location; For an ideal narrowband sound source The coherent matched field removes the source phase from each snapshot, thus obtaining a new measurement sequence. : For candidate sound source trajectories The model will generate sequences The measurement dataset will then be... With model dataset Each is combined into a hypervector, that is in, Represents the hypervector measurement field. This represents the transpose of the new measurement sequence in the first snapshot. Indicates the first L A quick snapshot of the transpose of a new measurement sequence Represents a supervector copy field. This indicates the transpose of the copy field at the first snapshot position. Indicates the first L The transpose of the copy field at the snapshot location. This indicates the copy field distance and depth corresponding to the first snapshot. Indicates the first L The copy field distance and depth corresponding to each snapshot; cross-spectral density matrix Represented as: in, Indicates the number of snapshots. Indicates the first s The conjugate transpose of a snapshot of a hypervector measurement field Indicates the first s A quick snapshot of the hypervector measurement field; Output power of the coherent matched field at time 1 Represented as: in, Indicates the inversion parameters. These represent distance, depth, and radial velocity, respectively. Indicates frequency as f The hypervector copy field at time, This represents the conjugate transpose of a matrix. express Norm, Represents the trace of a matrix; For broadband signals, the output power of the coherent matched field is expressed as the incoherent average of the output power of the narrowband signal, i.e. in, Indicates the number of frequency points. Indicates the first k The 3D fuzzy surface at a given time. Indicates the frequency point number. Represents frequency The corresponding three-dimensional fuzzy surface.

2. The coherent matched field tracking method according to claim 1, characterized in that, Step 3 specifically involves, from each three-dimensional ambiguity function... Extract Local maxima The number of them follows a Poisson distribution. ; for The parameter vectors corresponding to the local maxima, where .

3. The coherent matched field tracking method according to claim 2, characterized in that, Step 4 specifically involves using a nonmaximum suppression algorithm to... Clustering is performed on the peaks, assuming the initial neighborhood radius of the cluster is... ,in, Represents the radius of the neighborhood in the distance dimension. Represents the radius of the depth-dimensional neighborhood. Representing the radial velocity radius, the peak points are sorted in descending order of amplitude. For the current peak point... ,in, Representing the k Time of the first c Distance inversion values ​​corresponding to each cluster center Representing the k Time of the first c The depth inversion value corresponding to each cluster center Representing the k Time of the first c The radial velocity inversion value corresponding to each cluster center is used to calculate the velocity relative to other points. Weighted Euclidean distance: If and only if hour, Belongs to the current cluster, for those containing Clustering of points and their corresponding amplitudes Cluster center for: in, This is the cluster center number. Indicates the number of cluster centers. Represents the k-th time. The parameter vector corresponding to each cluster center Indicates the first The amplitude of each cluster center.

4. The coherent matched field tracking method according to claim 1, characterized in that, Step 5 specifically involves retaining only each frame. Estimates of distance, depth, and radial velocity corresponding to the peak positions of each cluster center As a three-dimensional measurement for coherent matched field tracking.

5. The coherent matched field tracking method according to claim 4, characterized in that, Step 6 specifically involves determining the trajectory status of each underwater moving sound source. Represented by a random finite set; where the target state Distance from the sound source Distance-dimensional velocity Sound source depth and depth dimension speed Composition, dimension Target state The transition relationship between adjacent step sizes is given by the state equation: in, Represents a time interval. Represents the variance of state noise. Represents the state transition matrix. Represents the noise matrix. represent k The distance at time -1 represent k The depth at time -1 represent k Velocity in the distance dimension at time -1 represent k The depth-dimensional velocity at time -1, the state equation is used according to k State prediction at time -1 k The state at any given moment; and The mapping relationship is given by the measurement equation: in, Represents the three-dimensional measurement values ​​used in practical applications, i.e. = , The mapping matrix representing the measurement vector and the state vector, Represents measurement noise. Represents distance dimension measurement noise. Represents depth dimension measurement noise. This represents the noise in the radial velocity dimension measurement.

6. A coherent matched field tracking system based on a trajectory Poisson-Dobernoli hybrid filter, characterized in that, The tracking system uses a coherent matched-field tracking method based on a trajectory Poisson-Dobernoli hybrid filter as described in any one of claims 1-5, and the tracking system includes the following steps: Data acquisition and processing module: Divides the long-sequence signal acquired by the hydrophone receiving array into multiple frames and converts it to the frequency domain through Fourier transform; Calculation module: For each frame of frequency domain signal processed by the data acquisition module, a three-dimensional ambiguity surface with respect to distance, depth and radial velocity is calculated using the coherent matched field algorithm; Local peak extraction module: Extracts local peaks from the blur surface of each frame obtained by the calculation module. The number of peaks follows a Poisson distribution. Clustering module: Based on the local peak values ​​of the blur surface of each frame obtained by the local peak extraction module 3, clustering is performed by the non-maximum suppression algorithm, and only multiple peak values ​​at the cluster center are retained; 3D Measurement Set Construction Module: Based on the distance, depth, and radial velocity estimates corresponding to multiple peaks of the cluster centers obtained by the clustering module, a 3D measurement set is constructed. Trajectory tracking module: Based on the trajectory Poisson-Dobernoli hybrid filter modeling method and the state variables of the coherent matched field measurement set, a matched field trajectory state space model and measurement model are established; Display module: Combines the matching field state space model and measurement model obtained from the trajectory tracking module with a trajectory Poisson-Dobernoli hybrid filter to achieve state prediction, update and data association in the trajectory space, thus completing trajectory estimation.

7. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1-5.

Images