A multi-model maneuvering target tracking method and device based on partial state interaction

Abstract

This invention discloses a multi-model maneuvering target tracking method and apparatus based on partial state interaction, belonging to the field of maneuvering target tracking. The method includes: firstly, using maneuvering models with the same structure to jointly describe a maneuvering mode, modeling maneuvering parameter jumps as switching between isomorphic maneuvering models, and initializing each maneuvering model's dedicated maneuvering parameter subspace using a partitioned Gaussian probability density function. Secondly, an innovative partial state interaction mechanism is designed, limiting the model's maneuvering parameter input to a weighted combination of its own historical estimates and initial parameter values, eliminating cross-interference in parameter estimates between models. This technical solution can achieve matching coverage and consistent initialization of the parameter space during maneuvering parameter jumps; and protect the optimal initialization filter from erroneous parameter estimation interference during maneuvering mode dwell periods. This invention can effectively address the maneuvering parameter jump problem and improve the tracking accuracy of maneuvering targets.

Description

Technical Field

[0001] This invention relates to the field of maneuvering target tracking technology, and in particular to a multi-model maneuvering target tracking method and apparatus based on partial state interaction. Background Technology

[0002] Maneuvering target tracking refers to the technique of estimating the state of a target whose motion characteristics change over time based on sensor observation data. Maneuvering parameters are physical quantities used to characterize the target's maneuvering motion, specifically including kinematic parameters such as target acceleration and turning rate. Single-model-based tracking methods often fail to meet high-precision tracking requirements due to a mismatch between the preset model and the actual motion pattern. In contrast, multi-model methods improve adaptability to target maneuvers by constructing a set of models of the target's potential motion patterns.

[0003] In related technologies, existing multi-model methods often employ parametric design methods, which discretize the maneuver parameter space into a finite number of quantization points, with each quantization point corresponding to a fixed model parameter of a model; or structured design methods, which model the maneuver parameters as a stochastic process and augment them into state components for real-time estimation. However, neither of these design methods can effectively cope with abrupt changes in maneuver parameters.

[0004] Therefore, there is an urgent need for a multi-model maneuvering target tracking method and device based on partial state interaction to solve the above-mentioned technical problems. Summary of the Invention

[0005] This invention provides a multi-model maneuvering target tracking method and apparatus based on partial state interaction, which can solve the problem that related technologies cannot effectively cope with abrupt changes in maneuvering parameters. The technical solution is as follows:

[0006] On the one hand, a multi-model maneuvering target tracking method based on partial state interaction is provided, the method comprising:

[0007] Step 100: Establish a set of motion models based on the motion patterns of the maneuvering target to be tracked, and divide the maneuvering parameter space of the maneuvering target into regions, determine the partition parameters of each region and the state equation of each motion model; wherein, the motion model includes a non-maneuvering model and a maneuvering model, and the same structure maneuvering model is used to describe one maneuvering pattern of the maneuvering target;

[0008] Step 102: Initialize the state and model probability for all the motion models, and set the model transition probability;

[0009] Step 104: At each filtering time, calculate the mixing probability of the motion model based on the model probability and the preset model transition probability;

[0010] Step 106: Perform partial state interaction on all motion models according to the mixture probability and the partitioning parameters, input the interaction results into the filter corresponding to the motion model, and filter to obtain the conditional state estimate and likelihood function of each motion model based on the observation data obtained at the current filtering time.

[0011] Step 108: Update the model probability of each motion model according to the likelihood function, and perform estimation fusion on the model conditional state estimate according to the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time.

[0012] On the other hand, a multi-model maneuvering target tracking device based on partial state interaction is provided, the device comprising:

[0013] The partitioning module is used to establish a set of motion models based on the motion patterns of the maneuvering target to be tracked, and to partition the maneuvering parameter space of the maneuvering target into regions, determining the partition parameters of each region and the state equation of each motion model; wherein, the motion model includes a non-maneuvering model and a maneuvering model, and the same structure maneuvering model is used to describe one maneuvering pattern of the maneuvering target;

[0014] An initialization module is used to initialize the state and model probabilities of all the motion models and set the model transition probabilities.

[0015] The calculation module is used to calculate the mixing probability of the motion model at each filtering time based on the model probability and the preset model transition probability;

[0016] The filtering module is used to perform partial state interaction on all the motion models according to the mixing probability and the partitioning parameters, input the interaction result into the filter corresponding to the motion model, and filter to obtain the conditional state estimate and likelihood function of each motion model based on the observation data obtained at the current filtering time.

[0017] The fusion module is used to update the model probability of each motion model according to the likelihood function, and to perform estimation fusion on the model conditional state estimate according to the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time.

[0018] On the other hand, a computer device is provided, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for executing the computer program stored in the memory to implement the steps of the multi-model maneuvering target tracking method based on partial state interaction described above.

[0019] On the other hand, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when the computer program is executed by a processor, it implements the steps of the multi-model maneuvering target tracking method based on partial state interaction described above.

[0020] On the other hand, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the multi-model maneuvering target tracking method based on partial state interaction described above.

[0021] This invention provides a multi-model maneuvering target tracking method and apparatus based on partial state interaction, belonging to the field of maneuvering target tracking. Addressing the model parameter bias problem in existing multi-model methods under maneuver parameter jump scenarios, an innovative solution is proposed: First, a maneuvering mode is described using maneuvering models with the same structure, and the maneuvering parameter jump is modeled as a switch between isomorphic maneuvering models. Each maneuvering model initializes its dedicated maneuvering parameter subspace through a partitioned Gaussian probability density function. Second, an innovative partial state interaction mechanism is designed, limiting the model's maneuvering parameter input to a weighted combination of its own historical estimates and initial parameter values, eliminating cross-interference in parameter estimates between models. This technical solution can achieve matching coverage and consistent initialization of the parameter space during maneuvering parameter jumps; and protect the optimal initialization filter from erroneous parameter estimation interference during maneuvering mode dwell periods. This invention effectively addresses the maneuvering parameter jump problem and improves the tracking accuracy of maneuvering targets. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a multi-model maneuvering target tracking method based on partial state interaction provided by an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of position estimation error provided in an embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of velocity estimation error provided in an embodiment of the present invention;

[0026] Figure 4 This is a structural diagram of a multi-model maneuvering target tracking device based on partial state interaction provided in an embodiment of the present invention;

[0027] Figure 5This is a hardware architecture diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0029] As mentioned earlier, existing multi-model tracking methods typically employ parametric or structured design, neither of which can effectively address the maneuver parameter jumps of the target being tracked.

[0030] Based on this, the concept of the present invention is to describe the change of maneuver parameters as a switching between the same structural models through partial state interaction, and to process the maneuver parameter components in the state separately, so as to achieve robust tracking under the gradual and abrupt changes of maneuver parameters.

[0031] The specific implementation of the above concept is described below.

[0032] Please refer to Figure 1 This invention provides a multi-model maneuvering target tracking method based on partial state interaction, the method comprising:

[0033] Step 100: Establish a set of motion models based on the motion patterns of the maneuvering target to be tracked, and divide the maneuvering parameter space of the maneuvering target into regions, determine the partition parameters of each region and the state equation of each motion model; wherein, the motion model includes a non-maneuvering model and a maneuvering model, and the same structure maneuvering model is used to describe one maneuvering pattern of the maneuvering target;

[0034] Step 102: Initialize the state and model probability for all the motion models, and set the model transition probability;

[0035] Step 104: At each filtering time, calculate the mixing probability of the motion model based on the model probability and the preset model transition probability;

[0036] Step 106: Perform partial state interaction on all motion models according to the mixture probability and the partitioning parameters, input the interaction results into the filter corresponding to the motion model, and filter to obtain the conditional state estimate and likelihood function of each motion model based on the observation data obtained at the current filtering time.

[0037] Step 108: Update the model probability of each motion model according to the likelihood function, and perform estimation fusion on the model conditional state estimate according to the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time.

[0038] In this embodiment of the invention, firstly, based on the possible motion patterns of the target, a structured maneuver model is used to describe the target's maneuvering. One type of target maneuver is described by multiple maneuver models with the same structure. Next, the maneuver parameter space is divided according to the set number of maneuver models with the same structure, with each partition corresponding to one maneuver model, thus constructing a filter model set. Then, the filtering algorithm is initialized, setting the model transition probabilities and initializing the filtering state and model probabilities. Recursive filtering then begins. The recursive filtering process sequentially performs model mixture probability calculation, partial state interaction, model matching filtering, model probability update, model condition estimation fusion, and recursive filtering until tracking ends. This method achieves significantly better performance when the target's acceleration changes abruptly, thereby improving the tracking accuracy of maneuvering targets.

[0039] The following description Figure 1 The execution method of each step is shown.

[0040] First, for step 100, a set of motion models is established based on the motion pattern of the maneuvering target to be tracked, and the space of the maneuvering parameters of the maneuvering target is divided into regions to determine the partition parameters of each region and the state equation of each motion model.

[0041] In this embodiment of the invention, considering that the maneuvering parameters of a maneuvering target are not fixed, a structured model is used to describe the target's maneuvering to account for minute changes in the maneuvering parameters, and a set of motion models is established based on the motion pattern. The motion model includes non-maneuvering models and maneuvering models, with a structured maneuvering model used to describe one maneuvering pattern of the maneuvering target. For example, multiple cooperative turning models (NCT) jointly describe the near-constant turning maneuvering of the target.

[0042] Specifically, we assume that the target's possible motion modes include near-uniform motion and near-uniform acceleration / deceleration in a plane. In this case, the motion model set consists of a two-dimensional near-uniform motion model (NCV) and a two-dimensional near-uniform acceleration motion model (NCA). The NCV is a non-maneuvering model, and there is one and only one non-maneuvering model in the model set; the NCA is a maneuvering model, and there can be many NCAs, all of which have the same model structure, i.e., identical maneuvering models.

[0043] It is worth noting that if the maneuvering target has other motion modes such as turning, the maneuvering model also includes a turning NCT model. There can also be many turning models. Similarly, each turning model has the same structure and is specifically designed to describe the turning motion of the target.

[0044] In this embodiment of the invention, the maneuver parameter space of the maneuvering target is divided into regions, and the partition parameters of each region and the state equation of each motion model are determined. This includes the following steps: determining the maneuver parameter space based on the maneuver range of the maneuvering target to be tracked; dividing the corresponding maneuver parameter space according to the number of maneuvering models with the same structure, and calculating the partition parameters of each region; establishing the state equation of the non-maneuvering model based on the position and velocity of the maneuvering target; and establishing the state equation of the maneuvering model with the same structure based on the position, velocity, and maneuver parameters of the maneuvering target.

[0045] Specifically, based on the number of maneuver models with the same structure, the maneuver parameter space of this type of maneuver model is divided. For NCA, the possible acceleration range of the maneuvering target is the maneuver parameter space of this maneuver model, denoted as .

[0046]

[0047] Among them l x and h x These represent the acceleration components a in the x-direction. x The minimum and maximum values, l y and h y These represent the acceleration components a in the y-direction. y The minimum and maximum values.

[0048] Let the number of identical NCA models be n. Then the maneuver parameter space is divided into n regions. For example, when n = 9, the acceleration is divided into three equal parts in both the x and y directions, using n x and n y Let n represent the number of segments for dividing the maneuver parameters in the x and y directions, respectively. x =n y =3. To give another example, when n=8, x and y can be divided into 2×4. The specific division method can be adjusted according to actual needs, which will not be elaborated here.

[0049] Furthermore, assuming the target prior acceleration follows a uniform distribution, then for the divided partitions, the mean of the β-th acceleration partition (β∈{1,2,...,n}) is... for:

[0050]

[0051] in, and Let be the x and y components of the mean of the β-th partition; i represents the index of the segment in the x direction, and the number of segments in the x direction is n.x i represents numbers from 1 to n. x .

[0052] The variance of the β-th partition is:

[0053]

[0054] The mean acceleration and the corresponding variance of the above partitions are the partition parameters.

[0055] Furthermore, the multi-motion model set constructed in this embodiment includes one NCV model and nine NCA models. Each NCA model is assigned different initial maneuvering parameters, and these NCA models are effective in different acceleration zones. The NCV model is denoted as m... 1 The nine NCA models are labeled as {m} 2 ,…,m 10 The total number of models in the model set is denoted as r, and there are a total of r = n + 1 models.

[0056] The state vector of the NCV model is The state vector of the NCA model is Where x k ,y k These represent the target's positions in the x and y directions, respectively. These are the velocities in the x and y directions, respectively. These are the accelerations in the x and y directions, respectively.

[0057] Therefore, the NCV model m is established. 1 The state equation is:

[0058]

[0059] in Here is the state transition matrix of the NCV model. The process noise gain matrix of the NCV model. To obey zero mean and covariance Gaussian white noise.

[0060] State transition matrix and process noise gain matrix As shown below:

[0061]

[0062] Where T is the data sampling interval.

[0063] NCA model m j The state equation for (j∈{2,...,r}) is:

[0064]

[0065] in Here is the state transition matrix of the NCA model. The process noise gain matrix of the NCA model. To obey zero mean and covariance Gaussian white noise.

[0066] State transition matrix and process noise gain matrix As shown below:

[0067]

[0068] For step 102, state initialization and model probability initialization are performed on all the motion models, and model transition probabilities are set.

[0069] In this embodiment of the invention, the algorithm initialization includes the following steps: calculating the initial state estimate of the motion model at the second moment based on the observation values ​​at the first and second moments; initializing the model probability of the motion model, where different model structures have the same probability and models with the same structure have equal probabilities; and determining the model transition probability of the motion model at two adjacent moments based on the switching relationship between the motion models and the switching relationship within the same structure maneuver model.

[0070] Specifically, this embodiment uses a two-point initialization method to initialize the state, but other initialization methods can also be used. The state estimate for the second time step is initialized using the observations z1 and z2 from the first two time steps.

[0071] NCV model m 1 The state estimate and corresponding covariance at time k=2 are:

[0072]

[0073]

[0074] In the formula, z k (1) represents the position observation in the x-direction at time k, z k (2) represents the position observation in the y direction at time k, and T represents the data sampling interval. R is the state estimation covariance of the NCV model at time k=2, where R is the observation noise covariance.

[0075] NCA model m j The state estimate and corresponding covariance of (j∈{2,…,r}) at k=2 are:

[0076]

[0077] In the formula, β∈{1,2,…,n}, 0 m×n This represents an m-row, n-column matrix consisting of 0s.

[0078] Furthermore, the model probability at the second time step is initialized; the model probability is the probability that the model is the true motion model at the current time step. Under initial conditions, different model structures have the same probability, and models with the same structure have equal probabilities. In this embodiment, the initial model probability of the NCV model is:

[0079]

[0080] The initial model probability of NCA is:

[0081]

[0082] Furthermore, modeling the jumps in motion parameters as switching between models with the same structure reveals that when setting the model transition probabilities for the motion model, we must consider not only the switching between different structural models but also the switching between models with the same structure. That is, given model m at time k-1... i Effective, model m at time k j The probability of it taking effect is expressed as:

[0083]

[0084] in, Representation model m i It takes effect at time k-1. Representation model m j It takes effect at time k.

[0085] Model transition probability π j|i The settings are as follows:

[0086]

[0087] In the first formula, j = i indicates that the model remains unchanged, i.e., P max The first formula represents the probability that the model remains unchanged. The second formula represents the switch from non-maneuvering mode to maneuvering mode. The third formula represents the switch from maneuvering mode to non-maneuvering mode. The last formula represents the switch between isomorphic maneuvering models and is used to describe the jump in maneuvering parameters.

[0088] For step 104, at each filtering time, the mixing probability of the motion model is calculated based on the model probability and the preset model transition probability.

[0089] In this embodiment of the invention, the prediction probability and mixture probability of the model can be calculated based on the model probability and the pre-set model transition probability.

[0090] The mixture probability is calculated as follows:

[0091]

[0092] in, For the model Predicted probability:

[0093]

[0094] For step 106, partial state interaction is performed on all the motion models according to the mixture probability and the partition parameters, and the interaction results are input into the filter corresponding to the motion model. Based on the observation data obtained at the current filtering time, the conditional state estimate and likelihood function of each motion model are obtained by filtering.

[0095] State interaction refers to the weighted combination of state estimates from each model at time k-1 as input to the matched filter of the model at time k. Unlike traditional full-state interaction strategies, this method proposes a partial state interaction strategy, interacting only with common state components in the model set. In this embodiment, the NCV model state includes target position and velocity, and the NCA model state includes target position, velocity, and acceleration. Interaction is only performed on the target position and velocity state components in the model set. The acceleration input to the filter corresponding to each NCA model is a weighted combination of the initial acceleration and the model's own estimate from the previous time step.

[0096] The interaction between the position and velocity state components yields the following results:

[0097]

[0098] Where j∈{1,2,…,r}, The model at time k-1 State estimation Position and velocity components in The model at time k-1 State estimation covariance The position and velocity covariance matrices.

[0099] For NCV model m 1 In terms of state interaction results and It is directly used as the input of its corresponding filter, denoted as:

[0100]

[0101] For NCA model m j For (j∈{2,...,r}), the acceleration components in the input also need to be calculated:

[0102]

[0103] Where j∈{2,...,r}, The model at time k-1 State estimation The acceleration component in The model at time k-1 State estimation covariance The acceleration covariance matrix in η j-1 Let be the mean of the acceleration partition corresponding to the β=j-1th NCA model, i.e.

[0104] Σ j-1 The covariance matrix of the acceleration partition corresponding to the β = j-1th NCA model is:

[0105]

[0106] Therefore, the input to the matched filter of the NCA model is:

[0107]

[0108] In this embodiment of the invention, the interaction result is input into the filter corresponding to the motion model, and the conditional state estimate and likelihood function of each motion model are obtained by filtering based on the observation data acquired at the current filtering time. The steps include: establishing an observation model for filtering; inputting the first interaction result into the filter corresponding to the non-maneuvering model, and inputting the second interaction result into the filter corresponding to each maneuvering model respectively; updating the filtering result according to the acquired observation value at the current filtering time and the observation model to obtain the target's state estimate, estimated covariance, and likelihood function at the current filtering time.

[0109] Specifically, the first step is to establish an observation model. This implementation plan considers a linear observation model:

[0110]

[0111] In the formula, z k This represents the observation received at time k. For model m j The observation matrix, w k The observation noise is subject to zero mean and covariance R.

[0112] NCV model m 1 The observation matrix is:

[0113]

[0114] NCA model m j The observation matrix for (j∈{2,...,r}) is:

[0115]

[0116] Next, based on the observations obtained at the current filtering time, Kalman filtering is performed to update the state estimate. and corresponding covariance

[0117]

[0118]

[0119]

[0120] In the formula, For the new information covariance; This is the Kalman gain.

[0121] And calculate the likelihood function.

[0122]

[0123] For step 108, the model probability of each motion model is updated according to the likelihood function, and the model conditional state estimate is estimated and fused according to the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time.

[0124] In this embodiment of the invention, after obtaining the likelihood function of the model, the model probability can be updated for all models in the model set:

[0125]

[0126] Where c k Normalization constant

[0127] Finally, the final state estimate and corresponding covariance at time k are calculated. Since the state estimates from the NCV and NCA models have unequal dimensions, the estimate from the NCV model is filled with zero mean and zero variance. State Estimation Covariance After dimension alignment, they are respectively denoted as and Using the model probability-weighted conditional estimation, the final estimate is:

[0128]

[0129] In summary, the above steps can achieve tracking of a maneuvering target at a filtering time; after initialization in steps 100 and 102, repeating steps 104-108 can achieve continuous tracking of a maneuvering target.

[0130] The following simulation example demonstrates the feasibility of the above method.

[0131] A maneuvering target starts at an initial speed of x0 = [8000m, 8000m, 25m / s, 200m / s]. T Motion begins. Actual acceleration a = [a] x ,a y ] T The sequences are shown in Table 1, where a x and a y These are the acceleration components along the x and y directions. A sensor located at the Cartesian coordinate origin reports the target's Cartesian position, and the observation is subject to a covariance of R. k =1250I2m 2 The interference from zero-mean observation noise. The sampling period is T = 1s, and the lower boundaries of the acceleration components in the x and y directions are assumed to be -40m / s². 2 The upper boundary is 40m / s 2 Perform 1000 Monte Carlo simulations, setting the probability that the model remains unchanged to P. max =0.98.

[0132] Table 1

[0133]

[0134] The average Euclidean errors of the PSIMM and traditional IMM algorithms in 1000 Monte Carlo simulations are as follows: Figure 2 As shown, the speed-averaged Euclidean error of PSIMM and the traditional IMM algorithm under 1000 Monte Carlo simulations is as follows: Figure 3 As shown in the figure, the PSIMM algorithm proposed in this invention achieves significantly better performance when the target acceleration changes abruptly, thereby improving the tracking accuracy of maneuvering targets.

[0135] Please refer to Figure 4 This invention provides a multi-model maneuvering target tracking device based on partial state interaction, the device comprising:

[0136] The partitioning module 400 is used to establish a set of motion models based on the motion pattern of the maneuvering target to be tracked, and to partition the maneuvering parameter space of the maneuvering target into regions, and to determine the partition parameters of each region and the state equation of each motion model; wherein, the motion model includes a non-maneuvering model and a maneuvering model, and the same structure maneuvering model is used to describe one maneuvering pattern of the maneuvering target;

[0137] The initialization module 402 is used to initialize the state and model probability of all the motion models and set the model transition probability.

[0138] The calculation module 404 is used to calculate the mixing probability of the motion model at each filtering time based on the model probability and the preset model transition probability;

[0139] The filtering module 406 is used to perform partial state interaction on all the motion models according to the mixing probability and the partitioning parameters, input the interaction result into the filter corresponding to the motion model, and filter to obtain the conditional state estimate and likelihood function of each motion model based on the observation data obtained at the current filtering time.

[0140] The fusion module 408 is used to update the model probability of each motion model according to the likelihood function, and to perform estimation fusion on the model conditional state estimate based on the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time. In this embodiment of the invention, the motion model includes a non-maneuvering model and a maneuvering model, and the same-structure maneuvering model is used to describe a maneuvering mode of the maneuvering target.

[0141] In this embodiment of the invention, when the partitioning module 400 performs the following operations to divide the maneuver parameter space of the maneuvering target into regions and determine the partition parameters of each partition and the state equation of each motion model: determining the maneuver parameter space based on the maneuver range of the maneuvering target to be tracked; dividing the corresponding maneuver parameter space according to the number of maneuvering models with the same structure and calculating the partition parameters of each partition; establishing the state equation of the non-maneuvering model based on the position and velocity of the maneuvering target; and establishing the state equation of the maneuvering model with the same structure based on the position, velocity, and maneuver parameters of the maneuvering target.

[0142] In this embodiment of the invention, when the initialization module 402 performs state initialization and model probability initialization on all the motion models and sets the model transition probabilities, it specifically performs the following operations: calculating the initial state estimate of the motion model at the second moment based on the observation values ​​at the first and second moments; initializing the model probabilities of the motion models, where different model structures have the same probability and models with the same structure have equal probabilities; and determining the model transition probabilities of the motion models at two adjacent moments based on the switching relationships between the motion models and the switching relationships within the same structure maneuver models.

[0143] In this embodiment of the invention, when the filtering module 406 performs partial state interaction on all the motion models based on the mixing probability and the partitioning parameters, it specifically performs the following operations:

[0144] The non-maneuvering model is interactively processed based on the common state component estimates of the non-maneuvering model and the maneuvering model to obtain the first interactive result of the non-maneuvering model. and

[0145]

[0146] in, The model at time k-1 State estimation Position and velocity components in the data; The model at time k-1 State estimation covariance The position and velocity covariance matrices;

[0147] The maneuver models are interactively processed based on the common state component estimates and the specific state component estimates of the maneuver models to obtain a second interactive result for each maneuver model. and

[0148]

[0149] in, The model at time k-1 State estimation The acceleration component in The model at time k-1 State estimation covariance The acceleration covariance matrix in [the matrix].

[0150] In this embodiment of the invention, the filtering module 406, when executing the process of inputting the interaction result into the filter corresponding to the motion model and filtering based on the observation data acquired at the current filtering time, obtains the conditional state estimate and likelihood function of each motion model, including: establishing an observation model for filtering; inputting the first interaction result into the filter corresponding to the non-maneuvering model, and inputting the second interaction result into the filter corresponding to each maneuvering model respectively; updating the filtering result according to the acquired observation value at the current filtering time and the observation model to obtain the target's state estimate, estimated covariance, and likelihood function at the current filtering time.

[0151] In this embodiment of the invention, when the fusion module 408 performs estimation fusion on the model conditional state estimate based on the updated model probability to obtain the final state estimate of the tracking maneuvering target at the current time, it specifically performs the following operations: fills and aligns the model conditional state estimate and the estimated covariance at the current filtering time according to zero mean and zero variance, respectively; and performs weighted estimation on the processed model conditional state estimate and the estimated covariance according to the updated model probability to calculate the final state estimate.

[0152] It should be noted that the multi-model maneuvering target tracking device based on partial state interaction provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the multi-model maneuvering target tracking device based on partial state interaction provided in the above embodiments and the multi-model maneuvering target tracking method embodiments based on partial state interaction belong to the same concept. The specific implementation process is detailed in the method embodiments and will not be repeated here.

[0153] Embodiments of this application also provide a computer device, please refer to... Figure 5 The computer device includes a processor and a memory, the memory storing at least one instruction, at least one program, code set, or instruction set, wherein at least one instruction, at least one program, code set, or instruction set is loaded and executed by the processor to implement the multi-model maneuvering target tracking method based on partial state interaction provided in the above-described method embodiments.

[0154] Embodiments of this application also provide a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the multi-model maneuvering target tracking method based on partial state interaction provided in the above-described method embodiments.

[0155] Embodiments of this application also provide a computer program product, which includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium and executes the computer program, causing the computer device to perform any of the multi-model maneuvering target tracking methods based on partial state interaction described in the above embodiments.

[0156] For ease of description, the above systems or devices are described separately as various modules or units based on their functions. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware components.

[0157] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0158] Finally, it should be noted that in this document, relational terms such as first, second, third, and fourth are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0159] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A multi-model maneuvering target tracking method based on partial state interaction, characterized in that, The method includes: Step 100: Establish a set of motion models based on the motion patterns of the maneuvering target to be tracked, and divide the maneuvering parameter space of the maneuvering target into regions, determine the partition parameters of each region and the state equation of each motion model; wherein, the motion model includes a non-maneuvering model and a maneuvering model, and the same structure maneuvering model is used to describe one maneuvering pattern of the maneuvering target; Step 102: Initialize the state and model probabilities for all the motion models, and set the model transition probabilities, including: Based on the observations at the first and second moments, the initial state estimate of the motion model at the second moment is calculated; Initialize the model probability of the motion model. Different model structures have the same probability, and models with the same structure have equal probabilities. Based on the switching relationships between the motion models and the switching relationships within the same structure maneuver model, the model transition probability of the motion model at two adjacent time points is determined; Step 104: At each filtering time, calculate the mixing probability of the motion model based on the model probability and the preset model transition probability; Step 106, performing partial state interaction on all motion models according to the mixture probability and the partitioning parameters, including: The non-maneuvering model is interactively processed based on the common state component estimates of the non-maneuvering model and the maneuvering model to obtain the first interactive result of the non-maneuvering model. and : in, for Moments from the model State estimation Position and velocity components in the data; for Moments from the model State estimation covariance The position and velocity covariance matrices; The maneuver model is interactively processed based on the common state component estimation and the specific state component estimation of the maneuver model to obtain the second interactive result of the maneuver model. and : in, for Moments from the model State estimation The acceleration component in for Moments from the model State estimation covariance The acceleration covariance matrix in the figure; The interaction results are then input into the filter corresponding to the motion model, and based on the observation data obtained at the current filtering time, the conditional state estimate and likelihood function of each motion model are obtained through filtering. Step 108: Update the model probability of each motion model according to the likelihood function, and perform estimation fusion on the model conditional state estimate according to the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time.

2. The method as described in claim 1, characterized in that, The step of dividing the maneuvering target's maneuvering parameter space into regions, and determining the partition parameters of each region and the state equation of each motion model, includes: The maneuver parameter space is determined based on the maneuver range of the target to be tracked; The space of the corresponding maneuver parameters is divided according to the number of maneuver models with the same structure, and the partition parameters of each partition are calculated. Based on the position and velocity of the maneuvering target, establish the state equation of the non-maneuvering model; Based on the position, velocity, and maneuver parameters of the maneuvering target, establish the state equations of the same-structure maneuvering model.

3. The method as described in claim 1, characterized in that, The step of inputting the interaction results into the filter corresponding to the motion model, and filtering to obtain the conditional state estimate and likelihood function of each motion model based on the observation data acquired at the current filtering time, includes: Establish an observation model for filtering; The first interaction result is input into the filter corresponding to the non-maneuvering model, and the second interaction result is input into the filter corresponding to each maneuvering model respectively. The filtering results are updated based on the current filtering time observations and the observation model to obtain the target state estimate, estimated covariance, and likelihood function at the current filtering time.

4. The method as described in claim 1, characterized in that, The step of estimating and fusing the model conditional state estimate based on the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time includes: The estimated model conditional state and estimated covariance at the current filtering time are padded and aligned based on the zero mean and zero variance, respectively. The final state estimate is calculated by weighting the processed model conditional state estimate and the estimated covariance based on the updated model probability.

5. A multi-model maneuvering target tracking device based on partial state interaction, characterized in that, The apparatus, used in the method of any one of claims 1-4, comprises: The partitioning module is used to establish a set of motion models based on the motion patterns of the maneuvering target to be tracked, and to partition the maneuvering parameter space of the maneuvering target into regions, determining the partition parameters of each region and the state equation of each motion model; wherein, the motion model includes a non-maneuvering model and a maneuvering model, and the same structure maneuvering model is used to describe one maneuvering pattern of the maneuvering target; An initialization module is used to initialize the state and model probabilities of all the motion models and set the model transition probabilities. The calculation module is used to calculate the mixing probability of the motion model at each filtering time based on the model probability and the preset model transition probability; The filtering module is used to perform partial state interaction on all the motion models according to the mixing probability and the partitioning parameters, input the interaction result into the filter corresponding to the motion model, and filter to obtain the conditional state estimate and likelihood function of each motion model based on the observation data obtained at the current filtering time. The fusion module is used to update the model probability of each motion model according to the likelihood function, and to perform estimation fusion on the model conditional state estimate according to the updated model probability to obtain the final state estimate of tracking the maneuvering target at the current time.

6. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory to implement the steps of the method according to any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method described in any one of claims 1-4.

8. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1-4.

Images