Target motion analysis method and device for suppressing abnormal value, equipment and storage medium

Abstract

The invention provides a target motion analysis method and device for suppressing an abnormal value, equipment and a storage medium. The method comprises the following steps: initializing model parameters and a covariance matrix of a target motion model; calculating according to the newest input vector and the target motion model to obtain a newest output vector, and calculating according to the newest output vector and the newest observation vector to obtain a newest residual vector; according to the norm of the newest residual error vector and a preset penalty function, a newest adjustment coefficient is obtained through calculation, and the preset penalty function is used for restraining the increasing amplitude of the newest adjustment coefficient along with the norm of the newest residual error vector when the norm of the newest residual error vector is larger than a preset threshold value; the model parameters and the covariance matrix are updated based on the improved RLS algorithm, and when the improved RLS algorithm calculates the newest gain vector, the covariance matrix before updating is replaced with a number multiplication matrix of the covariance matrix and the newest adjustment coefficient. According to the method and the device, the accuracy and the stability of model parameter updating in target tracking of the recursive least square method can be improved.

Description

Technical Field

[0001] This application relates to the technical field of target tracking, and specifically relates to a target motion analysis method, device, equipment, and storage medium for suppressing outliers. Background Art

[0002] Target tracking has a wide range of applications in many fields such as intelligent monitoring, unmanned driving, aerospace, etc. The Recursive Least Squares (RLS) method, as a commonly used parameter estimation method, has received much attention and application in the field of target tracking because it can update model parameters online in real time. It continuously uses new observation vectors to recursively update the model parameters to adapt to changes in the target motion state.

[0003] However, in actual target tracking scenarios, outliers often inevitably exist in the observation vectors. These outliers may stem from various factors such as sensor failures, external interferences, and atypical motions of the target. In the prior art, the Recursive Least Squares method is relatively sensitive to outliers in the observation vectors when updating model parameters. Due to its principle of minimizing the sum of squared residuals, outliers will produce a large residual contribution in the calculation, which will then cause a large interference to the model parameter update process. This seriously affects the accuracy and stability of the model parameter update, resulting in a decrease in target tracking accuracy and even possible target loss. Therefore, there is an urgent need for a method that can effectively suppress the influence of outliers in the observation vectors to improve the accuracy and stability of model parameter update in target tracking using the Recursive Least Squares method. Summary of the Invention

[0004] This application provides a target motion analysis method, device, equipment, and storage medium for suppressing outliers, which can solve the technical problem that outliers in the prior art seriously affect the accuracy and stability of model parameter update.

[0005] In a first aspect, an embodiment of this application provides a target motion analysis method for suppressing outliers, and the target motion analysis method includes: Initialize the model parameters and covariance matrix of the target motion model; Calculate the latest output vector based on the latest input vector and the target motion model, and calculate the latest residual vector based on the latest output vector and the latest observation vector; Calculate the latest adjustment coefficient based on the norm of the latest residual vector and a preset penalty function, where the preset penalty function is used to suppress the growth rate of the latest adjustment coefficient with the norm of the latest residual vector when the norm of the latest residual vector is greater than a preset threshold; Based on the improved RLS algorithm, update the model parameters and covariance matrix according to the latest adjustment coefficient, the latest input vector, and the latest residual vector. Among them, when calculating the latest gain vector in the improved RLS algorithm, the covariance matrix before update is replaced by the adjusted covariance matrix, and the adjusted covariance matrix is the multiplication matrix of the covariance matrix before update and the latest adjustment coefficient.

[0006] Further, in one embodiment, the preset penalty function is:

[0007] Wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold.

[0008] Further, in one embodiment, the preset penalty function is:

[0009] Wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold.

[0010] Further, in one embodiment, the preset penalty function is:

[0011] Wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is the adjustment parameter, , is the shape parameter, .

[0012] Further, in one embodiment, the preset penalty function is:

[0013]

[0014]

[0015] Wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is the adjustment parameter, , is the shape parameter, .

[0016] Further, in one embodiment, the preset penalty function is:

[0017]

[0018]

[0019] Among them, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is the adjustment parameter, , is the shape parameter, .

[0020] Furthermore, in one embodiment, the steps of updating the model parameters and covariance matrix according to the latest adjustment coefficient, the latest input vector, and the latest residual vector based on the improved RLS algorithm include: Calculating the latest gain vector according to the first formula, and the first formula is:

[0021] Among them, is the latest gain vector, is the adjusted covariance matrix, is the latest input vector, is the forgetting factor, ; Calculating the new model parameters according to the second formula, and the second formula is:

[0022] Among them, is the updated model parameter, is the model parameter before update, is the latest observation vector, is the latest output vector, is the latest residual vector; Calculating the new model parameters according to the third formula, and the third formula is:

[0023] Among them, is the updated covariance matrix, is the covariance matrix before update, and I is the identity matrix.

[0024] In a second aspect, an embodiment of the present application further provides a target motion analysis device for suppressing outliers, and the target motion analysis device includes: An initialization module, configured to initialize the model parameters and covariance matrix of the target motion model; A prediction update module, configured to calculate a latest output vector according to a latest input vector and a target motion model, and calculate a latest residual vector according to the latest output vector and the latest observation vector; An adjustment and suppression module, configured to calculate a latest adjustment coefficient according to the norm of the latest residual vector and a preset penalty function, where the preset penalty function is used to suppress the growth rate of the latest adjustment coefficient with respect to the norm of the latest residual vector when the norm of the latest residual vector is greater than a preset threshold; A recursive update module, configured to update model parameters and a covariance matrix based on an improved RLS algorithm according to the latest adjustment coefficient, the latest input vector, and the latest residual vector. When calculating the latest gain vector in the improved RLS algorithm, the covariance matrix before update is replaced with an adjusted covariance matrix, and the adjusted covariance matrix is a product matrix of the covariance matrix before update and the latest adjustment coefficient.

[0025] In a third aspect, an embodiment of the present application further provides a target motion analysis device for suppressing outliers. The target motion analysis device includes a processor, a memory, and a target motion analysis program stored on the memory and executable by the processor. When the target motion analysis program is executed by the processor, the steps of the above target motion analysis method are implemented.

[0026] In a fourth aspect, an embodiment of the present application further provides a storage medium, on which a target motion analysis program is stored. When the target motion analysis program is executed by a processor, the steps of the above target motion analysis method are implemented.

[0027] In the present application, when the norm of the latest residual vector is greater than a preset threshold, the latest observation vector is considered an outlier. When the norm of the latest residual vector is greater than the preset threshold, the growth rate of the latest adjustment coefficient with respect to the norm of the latest residual vector is suppressed. When calculating the latest gain vector, the covariance matrix before update is replaced with a product matrix of it and the latest adjustment coefficient, so as to reduce the sensitivity of the latest gain vector to outliers and reduce the interference of outliers on the update of model parameters. Through the present application, the influence of outliers in the observation vector can be effectively suppressed, and the accuracy and stability of model parameter update in target tracking by the recursive least squares method are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is a schematic flowchart of a target motion analysis method in an embodiment of the present application; Figure 2 is a schematic diagram of functional modules of a target motion analysis device in an embodiment of the present application; Figure 3 is a schematic hardware structure diagram of a target motion analysis device involved in the solution of an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.

[0030] First, some technical terms in this application are explained to facilitate the understanding of this application by those skilled in the art.

[0031] In the target tracking scenario, an outlier is an observed value in a set of observed values that deviates too much from the average value (for example, exceeding twice the standard deviation range). The reasons for the appearance of outliers include sensor failures, external interferences, non - typical movements of the target, etc., resulting in the noise probability distribution in the sensor observations deviating from the Gaussian distribution and showing heavy - tailed characteristics.

[0032] To make the purpose, technical solution, and advantages of this application clearer, the embodiments of this application will be further described in detail below in conjunction with the accompanying drawings.

[0033] In a first aspect, an embodiment of this application provides a target motion analysis method for suppressing outliers.

[0034] Figure 1 The flowchart of the target motion analysis method in an embodiment of this application is shown.

[0035] Referring to Figure 1 , in one embodiment, the target motion analysis method includes the following steps: S1. Initialize the model parameters and covariance matrix of the target motion model.

[0036] Optionally, the target motion model can be a linear model or a non - linear model.

[0037] Optionally, the model parameters can be initialized as a zero vector or a random vector, and the covariance matrix can be initialized as a relatively large diagonal matrix to indicate a relatively large uncertainty in the initial estimate.

[0038] S2. Calculate the latest output vector according to the latest input vector and the target motion model, and calculate the latest residual vector according to the latest output vector and the latest observation vector.

[0039] Exemplarily, is the input vector at time k, is the observation vector at time k, is the output vector at time k, is the residual vector at time k, , is the target motion model, calculate The model parameters used are the model parameters updated at time k-1 .

[0040] S3. Calculate the latest adjustment coefficient according to the norm of the latest residual vector and the preset penalty function, where the preset penalty function is used to suppress the growth rate of the latest adjustment coefficient with the norm of the latest residual vector when the norm of the latest residual vector is greater than the preset threshold.

[0041] Specifically, the norm is a way to measure a vector or a matrix. It can map a vector or a matrix to a non-negative real number and is used to measure the size of the vector or the matrix.

[0042] Optionally, in this embodiment, the norm can be the L1 norm or the L2 norm.

[0043] In this embodiment, it is judged whether the latest observation vector is an outlier according to the size relationship between the norm of the latest residual vector and the preset threshold. The preset penalty function is segmented at the preset threshold. When the norm of the latest residual vector is less than or equal to the preset threshold, the latest adjustment coefficient increases with the increase of the norm of the latest residual vector. When the norm of the latest residual vector is greater than the preset threshold, the latest adjustment coefficient decreases with the increase of the norm of the latest residual vector, or remains unchanged, or increases with the increase of the norm of the latest residual vector, but the growth rate is less than the growth rate when the norm of the latest residual vector is less than or equal to the preset threshold.

[0044] S4. Based on the improved RLS algorithm, update the model parameters and the covariance matrix according to the latest adjustment coefficient, the latest input vector and the latest residual vector, where when calculating the latest gain vector in the improved RLS algorithm, the covariance matrix before update is replaced by the adjusted covariance matrix, and the adjusted covariance matrix is the product matrix of the covariance matrix before update and the latest adjustment coefficient.

[0045] Specifically, in the existing RLS algorithm, the update formula is as follows:

[0046]

[0047]

[0048] Among them, is the gain vector at time k, is the input vector at time k, is the forgetting factor, , is the covariance matrix updated at time k, is the model parameter updated at time k, is the observation vector at time k, is the output vector at time k, is the residual vector at time k.

[0049] For the sake of simplicity of expression, for time k, is the latest input vector, is the latest observation vector, is the latest output vector, is the latest residual vector, is the latest gain vector, is the covariance matrix before update, is the covariance matrix after update, is the model parameter before update, is the model parameter after update.

[0050] The gain vector is used to determine the influence degree of the observation vector at time k on the parameter update. In the calculation formula, the numerator is used to combine historical experience with current data to preliminarily determine the direction and amplitude of parameter update. In the denominator, is used to control the influence of old data on the current estimate. When the weights of all data are the same. When the weights of old data will gradually decrease with time.

[0051] In the denominator, represents the degree of correlation between current data and historical data. When is large, it indicates that there is a strong correlation between current data and historical data, or that the new information contained in current data is relatively small. At this time, the denominator is large, and the value of the gain vector will be relatively small, meaning that when updating the parameter, the dependence on current data will be reduced, and more reliance will be placed on historical experience. On the contrary, when is small, it indicates that current data brings more new information, the denominator is small, and the value of the gain vector will be relatively large, so that current data will be more fully utilized when updating parameters.

[0052] It can be seen that will affect the degree of dependence on historical experience when controlling parameter update. In the improved RLS algorithm proposed in this embodiment, when calculating the latest gain vector, the covariance matrix before update is replaced by , is the latest adjustment coefficient, so as to flexibly control the dependence on historical experience when updating parameters according to the latest observation data. When the latest observation data may be an outlier, compared with the existing algorithm, the dependence on historical experience is increased, and the interference of the outlier on parameter update is reduced.

[0053] Specifically, step S4 specifically includes: Calculate the latest gain vector according to the first formula, and the first formula is:

[0054] Wherein, is the latest gain vector, is the adjusted covariance matrix, is the latest input vector, is the forgetting factor, ; Calculate the new model parameters according to the second formula, and the second formula is:

[0055] Wherein, is the updated model parameter, is the model parameter before update, is the latest observation vector, is the latest output vector, is the latest residual vector; Calculate the new model parameters according to the third formula, and the third formula is:

[0056] Wherein, is the updated covariance matrix, is the covariance matrix before update, and I is the identity matrix.

[0057] Thus, in this embodiment, when the norm of the latest residual vector is greater than the preset threshold, the latest observation vector is considered an outlier. When the norm of the latest residual vector is greater than the preset threshold, the growth rate of the latest adjustment coefficient with respect to the norm of the latest residual vector is suppressed. When calculating the latest gain vector, the covariance matrix before update is replaced with its product matrix with the latest adjustment coefficient, thereby reducing the sensitivity of the latest gain vector to outliers and reducing the interference of outliers on model parameter update. Through this embodiment, the influence of outliers in the observation vector can be effectively suppressed, and the accuracy and stability of model parameter update in recursive least squares method for target tracking are improved.

[0058] Furthermore, in one embodiment, the preset penalty function is:

[0059] Wherein, is the latest adjustment coefficient, and r is the norm of the latest residual vector. is the preset threshold.

[0060] In this embodiment, the preset penalty function adopts the Huber penalty function. When the error is large, the Huber penalty function grows linearly, and its sensitivity to outliers is lower than that of the traditional squared error loss function (such as the loss function in the least squares method). It can, to a certain extent, suppress the influence of outliers on the estimation result, making the estimation result more stable and reliable, and having good robustness to outliers.

[0061] Furthermore, in one embodiment, the preset penalty function is:

[0062] where is the latest adjustment coefficient, and r is the norm of the latest residual vector. is the preset threshold.

[0063] In this embodiment, the preset penalty function adopts the Tukey penalty function. The greatest advantage of the Tukey penalty function is its strong resistance to outliers. In the traditional least squares method, the square of the error will cause outliers to have a greater impact on the result. However, after the residual exceeds the preset threshold, the value of the Tukey penalty function no longer increases, effectively suppressing the interference of outliers and making the estimation result more robust.

[0064] Furthermore, in one embodiment, the preset penalty function is:

[0065] where is the latest adjustment coefficient, and r is the norm of the latest residual vector. is the preset threshold, is the adjustment parameter, , is the shape parameter, .

[0066] In this embodiment, the preset penalty function is set based on the Generalized Minimax Concave Penalty (GMCP for short), and the latest adjustment coefficient is obtained by adding 1 to the GMCP calculation result. GMCP adaptively adjusts the penalty intensity according to the error size in the data. For smaller errors, it will impose a relatively large penalty to prompt the model to better fit normal data points; for larger errors, that is, possible outliers, the penalty intensity increases relatively slowly, avoiding these outliers from having too large an impact on the model estimation.

[0067] In the simulation experiment, it is found that directly using GMCP as the preset penalty function has poor effects. After adding 1 to the GMCP calculation result, the effects are significantly improved. After adding 1, it can be regarded as further enhancing the adjustment strength of parameter estimation on the basis of the original penalty, and in this way, the gain calculation can be more significantly affected, making the model more inclined to select parameter values that meet the constraint conditions during the optimization process. In addition, after adding 1, its numerical characteristics can be changed, so that when multiplying with other terms, the result is in a more appropriate range, ensuring the stability and rationality of the calculation process.

[0068] Further, in one embodiment, the preset penalty function is:

[0069]

[0070]

[0071] Wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is the adjustment parameter, , is the shape parameter, .

[0072] In this embodiment, the preset penalty function is comprehensively set based on the Huber penalty function and GMCP. Multiply the calculation result of the Huber penalty function by the sum of the GMCP calculation result and 1, integrating the advantages of the two penalty functions to enhance the suppression effect on outliers.

[0073] Further, in one embodiment, the preset penalty function is:

[0074]

[0075]

[0076] Wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is the adjustment parameter, , is the shape parameter, .

[0077] In this embodiment, the preset penalty function is comprehensively set based on the Tukey penalty function and GMCP. Multiply the calculation result of the Tukey penalty function by the sum of the GMCP calculation result and 1, integrating the advantages of the two penalty functions to enhance the suppression effect on outliers.

[0078] In a second aspect, an embodiment of the present application further provides a target motion analysis device for suppressing outliers.

[0079] Figure 2 The schematic diagram of the functional modules of the target motion analysis device in an embodiment of the present application is shown.

[0080] Referring to Figure 2 , in an embodiment, the target motion analysis device includes: An initialization module 10, configured to initialize the model parameters and covariance matrix of the target motion model; A prediction update module 20, configured to calculate a latest output vector according to the latest input vector and the target motion model, and calculate a latest residual vector according to the latest output vector and the latest observation vector; An adjustment suppression module 30, configured to calculate a latest adjustment coefficient according to the norm of the latest residual vector and a preset penalty function, where the preset penalty function is used to suppress the growth rate of the latest adjustment coefficient with the norm of the latest residual vector when the norm of the latest residual vector is greater than a preset threshold; A recursive update module 40, configured to update the model parameters and covariance matrix based on an improved RLS algorithm according to the latest adjustment coefficient, the latest input vector, and the latest residual vector, where, when calculating the latest gain vector, the improved RLS algorithm replaces the covariance matrix before update with an adjusted covariance matrix, and the adjusted covariance matrix is a multiplication matrix of the covariance matrix before update and the latest adjustment coefficient.

[0081] Further, in an embodiment, the preset penalty function is:

[0082] where is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold.

[0083] Further, in an embodiment, the preset penalty function is:

[0084] where is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold.

[0085] Further, in an embodiment, the preset penalty function is:

[0086] where is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is a regulation parameter, , is a shape parameter, .

[0087] Further, in one embodiment, the preset penalty function is:

[0088]

[0089]

[0090] wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is a regulation parameter, , is a shape parameter, .

[0091] Further, in one embodiment, the preset penalty function is:

[0092]

[0093]

[0094] wherein, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, is a regulation parameter, , is a shape parameter, .

[0095] Further, in one embodiment, the recursive update module 40 is used for: Calculating the latest gain vector according to the first formula, and the first formula is:

[0096] wherein, is the latest gain vector, is the adjusted covariance matrix, is the latest input vector, is the forgetting factor, ; Calculating the new model parameters according to the second formula, and the second formula is:

[0097] wherein, is the updated model parameter, is the model parameter before update, is the latest observation vector, is the latest output vector, is the latest residual vector; According to the third formula, a new model parameter is calculated. The third formula is:

[0098] wherein, is the updated covariance matrix, is the covariance matrix before update, and I is the identity matrix.

[0099] Among them, the function implementation of each module in the above target motion analysis device corresponds to each step in the above target motion analysis method embodiment, and its function and implementation process will not be elaborated here one by one.

[0100] In a third aspect, an embodiment of the present application provides a target motion analysis device for suppressing outliers. The target motion analysis device can be a device with data processing functions such as a personal computer (PC), a laptop computer, a server, etc.

[0101] Figure 3 shows a schematic hardware structure diagram of the target motion analysis device involved in the embodiment of the present application.

[0102] Referring to Figure 3 , in the embodiment of the present application, the target motion analysis device may include a processor, a memory, a communication interface, and a communication bus.

[0103] Among them, the communication bus can be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0104] The communication interface includes input / output (I / O) interfaces, physical interfaces, and logical interfaces, etc., which are used to implement the interconnection of internal components of the target motion analysis device, as well as interfaces for interconnecting the target motion analysis device with other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device can be a display (Display), a keyboard (Keyboard), etc.

[0105] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical memory, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0106] The processor can be a general-purpose processor, which can call the target motion analysis program stored in the memory and execute the target motion analysis method provided by the embodiments of the present application. For example, the general-purpose processor can be a central processing unit (CPU). Among them, the method executed when the target motion analysis program is called can refer to the various embodiments of the target motion analysis method of the present application, which will not be elaborated here.

[0107] Those skilled in the art can understand that Figure 3 the hardware structure shown in does not constitute a limitation to the present application, and may include more or fewer components than shown in the figure, or combine some components, or arrange different components.

[0108] In a fourth aspect, the embodiments of the present application further provide a storage medium.

[0109] The target motion analysis program is stored on the storage medium of the present application. When the target motion analysis program is executed by a processor, the steps of the target motion analysis method as described above are implemented.

[0110] Among them, the method implemented when the target motion analysis program is executed can refer to the various embodiments of the target motion analysis method of the present application, which will not be elaborated here.

[0111] It should be noted that the serial numbers of the above embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

[0112] In the description of the specification, claims and the above-mentioned drawings of this application, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices. Descriptions such as "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit that "first", "second" and "third" are different types.

[0113] In the description of the embodiments of this application, words such as "exemplary", "for example" or "for illustration" are used to indicate examples, illustrations or explanations. Any embodiment or design solution described as "exemplary", "for example" or "for illustration" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary", "for example" or "for illustration" is intended to present relevant concepts in a specific manner.

[0114] In the description of the embodiments of this application, unless otherwise specified, " / " means "or". For example, A / B can mean A or B; "and / or" in the text is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the description of the embodiments of this application, "a plurality of" means two or more than two.

[0115] In some processes described in the embodiments of this application, there are a plurality of operations or steps that appear in a specific order. However, it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of this application or may be executed in parallel. The serial numbers of the operations are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0116] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. 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 is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above and includes several instructions to enable a terminal device to execute the methods described in various embodiments of this application.

[0117] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present application.

Claims

1. A target motion analysis method for suppressing outliers, characterized in that: The target motion analysis method comprises: Initialize the model parameters and covariance matrix of the target motion model; The latest output vector is calculated based on the latest input vector and the target motion model, and the latest residual vector is calculated based on the latest output vector and the latest observation vector; The latest adjustment coefficient is calculated according to the norm of the latest residual vector and a preset penalty function, wherein the preset penalty function is used to suppress the growth rate of the latest adjustment coefficient along with the norm of the latest residual vector when the norm of the latest residual vector is greater than a preset threshold; Based on the improved RLS algorithm, the model parameters and covariance matrix are updated according to the latest adjustment coefficient, the latest input vector and the latest residual vector. When calculating the latest gain vector, the improved RLS algorithm replaces the covariance matrix before the update with the adjusted covariance matrix, and the adjusted covariance matrix is ​​the multiplication matrix of the covariance matrix before the update and the latest adjustment coefficient.

2. The target motion analysis method according to claim 1, characterized in that: The default penalty function is: in, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold.

3. The target motion analysis method according to claim 1, characterized in that: The default penalty function is: in, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold.

4. The target motion analysis method according to claim 1, characterized in that: The default penalty function is: in, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, To adjust the parameters, , is the shape parameter, .

5. The target motion analysis method according to claim 1, characterized in that: The default penalty function is: in, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, To adjust the parameters, , is the shape parameter, .

6. The target motion analysis method according to claim 1, characterized in that: The default penalty function is: in, is the latest adjustment coefficient, r is the norm of the latest residual vector, is the preset threshold, To adjust the parameters, , is the shape parameter, .

7. The target motion analysis method according to any one of claims 1 to 6, characterized in that: The step of updating the model parameters and the covariance matrix based on the improved RLS algorithm according to the latest adjustment coefficient, the latest input vector and the latest residual vector comprises: The latest gain vector is calculated according to the first formula, which is: in, is the latest gain vector, To adjust the covariance matrix, is the latest input vector, For the forgetting factor, ; The new model parameters are calculated according to the second formula, which is: in, are the updated model parameters, are the model parameters before updating, is the latest observation vector, is the latest output vector, is the latest residual vector; The new model parameters are calculated according to the third formula, which is: in, is the updated covariance matrix, is the covariance matrix before updating, and I is the identity matrix.

8. A target motion analysis device for suppressing abnormal values, characterized in that: The target motion analysis device comprises: An initialization module, used to initialize the model parameters and covariance matrix of the target motion model; A prediction update module is used to calculate the latest output vector based on the latest input vector and the target motion model, and calculate the latest residual vector based on the latest output vector and the latest observation vector; An adjustment suppression module, used to calculate a latest adjustment coefficient according to the norm of the latest residual vector and a preset penalty function, wherein the preset penalty function is used to suppress the growth rate of the latest adjustment coefficient with the norm of the latest residual vector when the norm of the latest residual vector is greater than a preset threshold; The recursive update module is used to update the model parameters and covariance matrix based on the improved RLS algorithm according to the latest adjustment coefficient, the latest input vector and the latest residual vector. When calculating the latest gain vector, the improved RLS algorithm replaces the covariance matrix before the update with the adjusted covariance matrix, and the adjusted covariance matrix is ​​the multiplication matrix of the covariance matrix before the update and the latest adjustment coefficient.

9. A target motion analysis device for suppressing outliers, characterized in that: The target motion analysis device includes a processor, a memory, and a target motion analysis program stored in the memory and executable by the processor, wherein when the target motion analysis program is executed by the processor, the steps of the target motion analysis method as described in any one of claims 1 to 7 are implemented.

10. A storage medium, characterized in that: The storage medium stores a target motion analysis program, wherein when the target motion analysis program is executed by a processor, the steps of the target motion analysis method according to any one of claims 1 to 7 are implemented.

Images