An asynchronous multi-sensor target tracking adaptive fusion method and system

Abstract

The application relates to the field of target tracking and provides an asynchronous multi-sensor target tracking adaptive fusion method and system. The method comprises the following steps: constructing a target state equation and a sensor measurement equation; according to the target state equation and the sensor measurement equation, adopting an unscented Kalman filtering algorithm to calculate a target local optimal estimation; according to a dynamic communication topology structure built by sensor nodes, transmitting the target local optimal estimation of each sensor to a fusion center; and according to the target local optimal estimation of each sensor, the fusion center carries out adaptive fusion to obtain a fusion estimation result. The application considers the sampling rate, transmission rate and communication delay of sensors in the system, discusses the multi-sensor asynchronous measurement information fusion problem when the target is maneuvering, designs a layered adaptive fusion structure, and realizes layered sequential fusion of sensor measurement.

Description

Technical Field

[0001] This invention relates to the field of target tracking, and in particular to an asynchronous multi-sensor target tracking adaptive fusion method and system. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] The application of multiple sensors increases the perception capability and coverage of the tracking system. By combining the measurement information of multiple sensors, more accurate target state estimation can be provided, increasing the robustness and environmental adaptability of the tracking system. Therefore, how to effectively utilize the measurements of multiple sensors in the system has become the focus of discussion.

[0004] Currently, multi-sensor fusion methods are categorized into three types: centralized fusion, distributed fusion, and hybrid fusion. Centralized fusion estimates by concentrating information from all sensors into a single observation equation, essentially expanding the dimensions of the observed state, and finally outputting a fused state estimate. This method has the advantage of minimal information loss but involves high computational cost, and sensor anomalies can lead to filter divergence. Distributed fusion preprocesses the measurement information from each sensor, filtering out irrelevant clutter and interference to obtain local estimates, before sending them to the fusion center for fusion. This type of fusion has low computational cost and is unaffected by anomalies from a single sensor, making it a widely used method. Hybrid fusion combines the above two methods. The fusion center receives both the original measurement information and the preprocessed information, resulting in better fusion performance, but with higher structural complexity.

[0005] Current research on target tracking largely relies on specific experimental conditions. These conditions often assume that sensors in the system operate synchronously at a single rate, the target maintains a constant speed during tracking (i.e., it has no maneuverability), and the time for measurement information transmission and arrival at the fusion center is consistent. However, in real-world target tracking scenarios, factors such as target maneuverability, sensor type, sampling rate, transmission rate, and communication latency can significantly impact the accuracy of target tracking. Applying multi-sensor fusion based on only a single, specific condition may lead to filtering divergence due to data anomalies. Summary of the Invention

[0006] To address the technical problems existing in the background art, the present invention provides an asynchronous multi-sensor target tracking adaptive fusion method and system. It considers the sampling rate, transmission rate and communication delay of the sensors in the system, discusses the problem of multi-sensor asynchronous measurement information fusion during target maneuvering, and designs a hierarchical adaptive fusion structure to realize hierarchical sequential fusion of sensor measurements.

[0007] The first aspect of the present invention provides an asynchronous multi-sensor target tracking adaptive fusion method.

[0008] An asynchronous multi-sensor target tracking adaptive fusion method includes:

[0009] Construct the target state equation and sensor measurement equation;

[0010] Based on the target state equation and the sensor measurement equation, the unscented Kalman filter algorithm is used to calculate the local optimal estimate of the target;

[0011] A dynamic communication topology is built based on sensor nodes, and the local optimal estimates of the target from each sensor are transmitted to the fusion center.

[0012] The fusion center performs adaptive fusion based on the local optimal estimates of the target from each sensor to obtain the fusion estimation result.

[0013] Furthermore, the process of employing the unscented Kalman filter algorithm includes:

[0014] Initialize the target's state vector and corresponding covariance matrix;

[0015] Based on the target state equation, a set of sampling points is generated using unscented transformation. The propagation sampling points of the target are obtained according to the interactive multi-model algorithm. In this way, the prior estimate of the target state mean and the target state covariance matrix at the next time step is calculated.

[0016] Continue propagating the sampling points, and calculate the sensor measurement mean and sensor measurement covariance matrix according to the sensor measurement equation;

[0017] Based on the prior estimates of the target state mean and target state covariance matrix, the sensor measurement mean and sensor measurement covariance matrix, intermediate variables are determined, and the local estimate of the target state and the posterior estimate of the target state covariance matrix are calculated.

[0018] Furthermore, the process of calculating the local optimum estimate of the target includes:

[0019] Based on the local estimate of the target state and the posterior estimate of the target state covariance matrix, the number of sensor sampling points during the tracking process is adjusted, and the measured values ​​are calculated so that the local estimate reaching the fusion center is the local optimal estimate of the target, and the local optimal estimate of the target is continuously updated.

[0020] Furthermore, the process of building a dynamic communication topology based on sensor nodes includes:

[0021] The least squares method is used to perform polygon localization on each sensor node and the target to determine the position of the sensor node;

[0022] Calculate the neighboring nodes of each sensor node, calculate the weights based on the coverage uniformity of the sensor nodes, and divide the sensors into multiple clusters to determine the cluster head node and the backup cluster head node in each cluster.

[0023] During target tracking, when a new sensor node is added or a node in the original structure loses tracking of the target, the process of determining the location of the sensor node, determining the cluster head node in each cluster, and determining the backup cluster head node is repeated.

[0024] Furthermore, the process of transmitting the local optimal estimates of the target from each sensor to the fusion center also includes:

[0025] Specify the load balancing strategy based on the relevant parameters of the data transmitted by the sensor nodes;

[0026] Weights are assigned based on the processing capacity of each cluster;

[0027] The weighted least connections algorithm is used to determine the next data transmission order of the cluster head node based on the number of connections and weights of active nodes within the cluster.

[0028] When a data transmission anomaly is detected, the load balancing mechanism is activated.

[0029] Furthermore, the process of transmitting the local optimal estimates of the target from each sensor to the fusion center also includes:

[0030] When nodes within a cluster are tracking the same target, they switch to a backup node if the primary node fails.

[0031] The fusion center sorts the received local estimates based on the generalized sequential algorithm. During fusion, after each cycle of fusion is completed, the timestamp is updated and the estimated values ​​that are lagging behind from the previous time are discarded.

[0032] A non-iterative fast covariance cross-fusion algorithm is used to calculate the trace of the local estimated covariance matrix to obtain the weight coefficients of the local estimate.

[0033] Based on the weighting coefficients, the fusion is performed in the order to be fused to obtain the target state after fusion at the current time and the corresponding covariance matrix.

[0034] Furthermore, after obtaining the fusion estimation results, the following is also included:

[0035] The fused target state and corresponding covariance matrix are compared with ground real-world data to evaluate the accuracy of the fusion estimation results. Feedback terms are designed using the target local optimum estimation transmitted by sensors and the fusion estimation results.

[0036] Adjust the weights of the sensor nodes based on the feedback items;

[0037] The adjusted weights of the sensor nodes are used for fusion in the next time step, and the fusion process is continuously iterated and improved.

[0038] A second aspect of the present invention provides an asynchronous multi-sensor target tracking adaptive fusion system.

[0039] An asynchronous multi-sensor target tracking adaptive fusion system includes:

[0040] The initialization module is configured to: construct the target state equation and the sensor measurement equation;

[0041] The local fusion module is configured to calculate the local optimal estimate of the target based on the target state equation and the sensor measurement equation using an unscented Kalman filter algorithm.

[0042] The transmission module is configured to: build a dynamic communication topology based on the sensor nodes, and transmit the local optimal estimates of the target of each sensor to the fusion center;

[0043] The global fusion module is configured such that the fusion center performs adaptive fusion based on the local optimal estimates of the target from each sensor to obtain the fusion estimation result.

[0044] A third aspect of the present invention provides a computer-readable storage medium.

[0045] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps in the asynchronous multi-sensor target tracking adaptive fusion method as described in the first aspect above.

[0046] A fourth aspect of the present invention provides a computer device.

[0047] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of the asynchronous multi-sensor target tracking adaptive fusion method as described in the first aspect above.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] In a distributed multi-sensor target tracking system, this invention considers the target's mobility and the actual tracking performance of the sensors. It effectively integrates information from different sensors through a hierarchical fusion approach. Based on the sampling rate and quality requirements of the sensors, the task load is flexibly allocated to ensure a relatively balanced load on each sensor, thereby ensuring the continuity and stability of target tracking.

[0050] This invention enhances the stability and accuracy of target positioning by adding a feedback mechanism to continuously provide feedback and make corrections, thereby improving the system's adaptability to different scenarios and environments. Attached Figure Description

[0051] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0052] Figure 1 This is a flowchart illustrating the asynchronous multi-sensor target tracking adaptive fusion method of the present invention;

[0053] Figure 2 This is a diagram of the multi-sensor communication topology shown in this invention. Detailed Implementation

[0054] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0055] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0056] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0057] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of this disclosure. It should be noted that each block in a flowchart or block diagram may represent a module, segment, or portion of code, which may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, may be implemented using a dedicated hardware-based system that performs the specified functions or operations, or using a combination of dedicated hardware and computer instructions.

[0058] Example 1

[0059] like Figure 1 As shown, this embodiment provides an asynchronous multi-sensor target tracking adaptive fusion method. This embodiment uses the application of this method to a server as an example for illustration. It is understood that this method can also be applied to terminals, and can also be applied to systems including terminals, servers, and other components, and implemented through interaction between the terminal and the server. The server can be an independent physical server, a server cluster composed of multiple physical servers, or a distributed system. It can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network servers, cloud communication, middleware services, domain name services, CDN security services, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited herein. In this embodiment, the method includes:

[0060] Step 1: Initialize the target to be measured and construct the general system equations for the target state and sensor measurements;

[0061] Step 2: Perform filtering operations based on the improved filtering algorithm to calculate the local optimal estimate for each sensor;

[0062] Step 3: The sensors dynamically build a communication topology and transmit the local optimal estimate to the fusion center;

[0063] Step 4: Adaptive fusion with local estimation of the fusion center;

[0064] Step 5: Let k = k + 1, repeat the above steps to form a closed loop operation and perform iterative calculations.

[0065] Specifically, in step 1, the target state is nonlinear. To avoid loss of generality, the target state equation and the sensor measurement equation can be constructed as follows:

[0066] t s k =f s (t s k-1 ,ω s k-1 )

[0067]

[0068] Where s = (1, 2...n) represents the set of sensors in the system, f(·) and h(·) represent the state transition function and measurement function, respectively, and the process noise ω k Observation noise ν k All noises are Gaussian white noise and are independent of each other, with corresponding noise covariance matrices Q and Q, respectively. k and R k .

[0069] In step 2, calculations are performed within each sensor based on the unscented Kalman filter (UKF) algorithm to process the nonlinear measurement information of the sensor about the target, calculate the local estimate of the target, and perform sensor-level fusion.

[0070] In real-world tracking scenarios, treating the target's motion model as prior information and assuming the target doesn't maneuver, maintaining only a single, unchanging motion model, can interfere with accuracy and even cause system filter divergence. Therefore, the Interactive Multiple Model (IMM) algorithm is combined with the Unscented Kalman Filter (UKF) algorithm. A linear transformation is performed, followed by a standard Kalman filter operation. Using the local estimate at time k-1, the covariance matrix, and the measurement information at time k, a local estimate at time k is obtained. Then, it is determined whether the target's motion model has changed, the model probability is updated, and the filtering operation for the next time step is executed.

[0071] The Unscented Kalman Filter (UKF) algorithm transforms a nonlinear system into a linear system through symmetric sampling. The number of sampling points is determined by the dimension of the target state t. To achieve accurate positioning during target maneuvers, the sampling point density is dynamically adjusted according to the target's maneuverability. The filtering operation steps are as follows:

[0072] Step (2-1) Initialize the state vector representing the target's position, velocity, and other parameters, as well as the corresponding covariance matrix;

[0073] Step (2-2) uses unscented transformation to generate a set of sampling points, propagates the sampling points of the target model calculated by the IMM algorithm, and calculates the prior estimates of the target state mean and covariance matrix at the next time step.

[0074] Step (2-3) continues to propagate the sampling points, estimates the sensor measurement mean and its covariance matrix, calculates the cross-covariance matrix between the target state and the measured values, Kalman gain and other intermediate values, and obtains the local estimate of the target state and the posterior estimate of the covariance matrix.

[0075] During the filtering process, the uncertainty of the target state estimation is assessed by analyzing the covariance matrix each time sampling is performed. Adaptive sampling is achieved by adjusting the number of sampling points to ensure the effectiveness of filtering. When the system uncertainty is low, i.e., the target maneuverability is low and it is moving at a constant or near-constant speed, the number of sampling points is reduced; while when the system uncertainty is high or the tracking performance deteriorates, i.e., the target maneuverability is high and it is changing speed or turning, the number of sampling points needs to be increased.

[0076] Multiple measurements are continuously generated during sensor tracking. Therefore, in step 2, calculations are needed to process these measurements to ensure that the local estimate reaching the fusion center is the optimal estimate. Inside sensor s, in The local estimate at time t is expressed as The corresponding covariance matrix is ​​represented as P k’ i’ , in (t k-1 ,t k Based on measurement information within the time interval The updated local estimate is calculated using the following formula:

[0077]

[0078]

[0079] in, and Let these represent the updated local estimate and the corresponding covariance matrix, respectively. They represent from arrive Through the state transition matrix The intermediate parameters obtained from the calculation.

[0080] parameter The calculation formula is:

[0081]

[0082]

[0083] In step 3, the local estimates of the target processed by each sensor are transmitted through a dynamically constructed communication topology, providing a stable and effective data transmission method that ensures the orderly transmission of the local estimation information from step 2. For example... Figure 2 As shown, a directed graph G = (V, E) is used to represent the network structure, where V = s represents the set of nodes in the graph, i.e., the set of sensor nodes, and E represents the edges of the graph, i.e., the communication links between two sensor nodes. The steps for constructing the communication topology are as follows:

[0084] Step (3-1-1) uses the least squares method to perform polygon localization on the sensor nodes and the target in the system, determining the position of the sensor nodes to minimize the sum of the squared errors between the measured distance and the distance calculated based on the estimated target position. The calculation formula is as follows:

[0085]

[0086] Where (x,y) represents the target's position estimate at time k-1, (x s ,y s ) represents the measurement distance of sensor s, d s This represents the distance between sensor s and the target.

[0087] Step (3-1-2) Calculate the neighboring nodes of each node, calculate the weights based on the coverage uniformity of the sensor nodes, divide the sensors into multiple clusters, and determine the cluster head node and backup cluster head node in each cluster of the system.

[0088] In step (3-1-3) during target tracking, when a new sensor node is added or a node in the original structure loses tracking of the target, repeat steps (3-1-1) and (3-1-2) above to dynamically update the topology.

[0089] To mitigate the impact of sudden data anomalies, instability, or unbalanced transmission on stability during multi-sensor data transmission, a combination of dynamic load balancing and failover is employed to ensure transmission stability. The measures are as follows:

[0090] Step (3-2-1) Develop a load balancing strategy based on the data transmission rate and latency of the sensor nodes in the system;

[0091] Step (3-2-2) assigns appropriate weights based on the processing capacity of each cluster. Clusters with a large number of sensor nodes should have their cluster head nodes assigned higher weights.

[0092] Step (3-2-3) uses the weighted least connections algorithm to determine the next data transmission order of the cluster head node based on the number of connections and weights of active nodes within the cluster.

[0093] Step (3-2-4): Upon detecting a data transmission anomaly, initiate a load balancing mechanism. Redistribute data traffic among available paths, nodes, or communication channels to ensure no single path or node is overloaded, maintaining overall network stability.

[0094] Step (3-2-5) of the failover mechanism specifically refers to the operation of switching to a standby node when the primary node fails in the system. In the system, when nodes within a cluster track the same target, the measurement information has a certain degree of redundancy. However, when the cluster head node fails, the communication link of that cluster will be interrupted, and the cluster head node will be unable to forward data. At this time, the system will switch to the standby cluster head node to retransmit the data, thereby minimizing the risk of data loss or interruption due to cluster head node failure.

[0095] The fusion center sorts the received local estimates based on a generalized sequential algorithm, in (t) k-1 , t k The order of fusion within the time frame can be represented as follows: in, This represents the local estimate to be fused from the s-th sensor. During the fusion operation, the timestamp is updated after each fusion cycle is completed, and the estimated values ​​that are lagging behind from the previous time step are discarded to ensure the timeliness of the local estimate fusion.

[0096] In step 4, the fusion center performs a higher-level fusion of the received local estimates from the sensors. Since the sensors transmit local estimates and their covariance matrices in the distributed fusion structure, the correlation between the various local estimates is unknown, and it is difficult to calculate the cross-covariance matrix between the estimates. Therefore, a non-iterative fast covariance cross-fusion algorithm is used to achieve effective fusion of local estimates with unknown correlation.

[0097] The fusion center obtains the weighting coefficients of the local estimate by calculating the trace of the covariance matrix of the local estimate, as shown in the following formula:

[0098]

[0099] Among them, tr(P) s The expression represents the trace operation of the covariance matrix corresponding to the local estimate of sensor s. The weight coefficients of each sensor in the system should satisfy ω1 + ω2 + ... + ω n =1 0≤ω s A linear constraint ≤ 1 is applied. Then, fusion is performed according to the order of fusion to obtain the fused target state and its corresponding covariance matrix at the current time step, expressed by the formula:

[0100]

[0101]

[0102] in, This represents the weighting coefficient of sensor s at time k. and The local optimal estimates and their covariance matrices to be fused are respectively.

[0103] To achieve dynamic response to changes in sensor reliability and system requirements, and to ensure optimal fusion performance, the fusion estimate is compared with ground real-time data, and the results are fed back down to the sensor nodes within each cluster through the cluster head node. The weights during sensor fusion are dynamically adjusted to obtain more reliable fusion results.

[0104] The steps for updating the weights are as follows:

[0105] Step (4-1) compares the fusion estimation results with the ground-based data to evaluate the accuracy of the fusion estimation. The feedback term F is then calculated using the local estimates from the sensors and the fusion estimation. s calculate;

[0106] Step (4-2) according to F s The analysis adjusts the sensor weights, increasing the weight of sensors with better performance and decreasing the weight of sensors with poorer performance. The weight update formula can be expressed as: Where β represents the parameter of the feedback information adjustment rate;

[0107] Step (4-3) normalizes the updated weights to ensure that the sum of the weights is 1, thus ensuring that the weights reflect the relative contribution of each sensor to the fusion process.

[0108] Step (4-4) involves fusing the data at the next time step based on the updated weights, continuously analyzing feedback indicators, updating the weights accordingly, and iteratively improving the fusion process.

[0109] By analyzing feedback and evaluating performance, the weights are updated, enabling the system to adapt to changing conditions, dynamically adjust the contribution of each sensor to the fusion estimation, improve accuracy, and achieve better fusion results over time.

[0110] Step 5 is used to repeat steps 2 to 4 above until the fusion of all local estimates in the system is completed.

[0111] Example 2

[0112] This embodiment provides an asynchronous multi-sensor target tracking adaptive fusion system.

[0113] An asynchronous multi-sensor target tracking adaptive fusion system includes:

[0114] The initialization module is configured to: construct the target state equation and the sensor measurement equation;

[0115] The local fusion module is configured to calculate the local optimal estimate of the target based on the target state equation and the sensor measurement equation using an unscented Kalman filter algorithm.

[0116] The transmission module is configured to: build a dynamic communication topology based on the sensor nodes, and transmit the local optimal estimates of the target of each sensor to the fusion center;

[0117] The global fusion module is configured such that the fusion center performs adaptive fusion based on the local optimal estimates of the target from each sensor to obtain the fusion estimation result.

[0118] It should be noted that the initialization module, local fusion module, transmission module, and global fusion module described above are the same examples and application scenarios implemented in Embodiment 1, but are not limited to the content disclosed in Embodiment 1. It should also be noted that these modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0119] Example 3

[0120] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the asynchronous multi-sensor target tracking adaptive fusion method as described in Embodiment 1 above.

[0121] Example 4

[0122] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the asynchronous multi-sensor target tracking adaptive fusion method as described in Embodiment 1 above.

[0123] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0124] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0125] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0126] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0127] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0128] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An asynchronous multi-sensor target tracking adaptive fusion method, characterized in that, The application relates to a target tracking method based on a self-adaptive fusion algorithm. The method comprises the following steps: building a target state equation and a sensor measurement equation; calculating a target local optimal estimation by using an unscented Kalman filter algorithm according to the target state equation and the sensor measurement equation; the process of using the unscented Kalman filter algorithm comprises the following steps: initializing a state vector of the target and a corresponding covariance matrix; generating a group of sampling points by using an unscented transformation based on the target state equation, calculating a target model propagation sampling point according to an IMM algorithm, calculating a target state mean value and a prior estimation of the covariance matrix at the next moment, estimating a sensor measurement mean value and a covariance matrix, calculating a cross-covariance matrix between the target state and the measurement value, a Kalman gain, and obtaining a target state local estimation and a posterior estimation of the covariance matrix; in the filtering operation process, the uncertainty of the target state estimation is evaluated by analyzing the covariance matrix every time sampling is performed, and adaptive sampling is realized by adjusting the number of sampling points; when the system uncertainty is low, i.e. the target maneuverability is low, the target moves at a constant speed or an approximate constant speed, the number of sampling points is reduced; when the system uncertainty is high or the tracking performance is reduced, i.e. the target maneuverability is high, the target moves at a variable speed or turns, the number of sampling points needs to be increased; a dynamic communication topology structure is built according to the sensor nodes, and the target local optimal estimations of all the sensors are transmitted to a fusion center; 2. The asynchronous multi-sensor target tracking adaptive fusion method according to claim 1, characterized in that, the fusion center performs adaptive fusion according to the target local optimal estimations of all the sensors to obtain a fusion estimation result. the process of calculating the target local optimal estimation comprises the following steps:

3. The asynchronous multi-sensor target tracking adaptive fusion method according to claim 1, wherein, the number of sensor sampling points in the tracking process is adjusted according to the target state local estimation and the posterior estimation of the target state covariance matrix, and a measurement value is calculated, so that the local estimation reaching the fusion center is the target local optimal estimation, and the target local optimal estimation is continuously updated. the process of building the dynamic communication topology structure according to the sensor nodes comprises the following steps: a least square method is used to multi-laterate the target and all the sensor nodes to determine the positions of the sensor nodes; neighboring nodes of all the sensor nodes are calculated, weights are calculated according to the uniformity of the sensor node coverage, the sensors are divided into multiple clusters, cluster head nodes and backup cluster head nodes in each cluster are determined; 4. The asynchronous multi-sensor target tracking adaptive fusion method according to claim 1, characterized in that, in the target tracking process, when a new sensor node is added or a node in the original structure loses the tracking of the target, the process of determining the positions of the sensor nodes, determining the cluster head nodes and the backup cluster head nodes in each cluster is repeated. in the process of transmitting the target local optimal estimations of all the sensors to the fusion center, the following steps are further included: a load balancing strategy is specified according to relevant parameters of the sensor node data transmission; weights are allocated according to the processing capacity of each cluster; a weighted minimum connection algorithm is used to determine the next data transmission sequence of the cluster head nodes according to the connection number of the active nodes in the cluster and the weight level; 5. The asynchronous multi-sensor target tracking adaptive fusion method according to claim 1, wherein, when an abnormal data transmission is detected, a load balancing mechanism is started. in the process of transmitting the target local optimal estimations of all the sensors to the fusion center, the following steps are further included: when the main nodes in the cluster track the same target, the backup nodes are switched when the main nodes fail. The fusion center ranks the received local estimations based on a generalized sequential algorithm, and each time a cycle of fusion is completed, the time stamp is updated and the estimation value of the previous time lag is discarded; The trace of the covariance matrix of the local estimation is calculated by using a non-iterative fast covariance intersection fusion algorithm to obtain the weight coefficient of the local estimation; According to the weight coefficient, the fusion is performed in the order of the to-be-fused, and the target state after fusion and the corresponding covariance matrix at the current time are obtained.

6. The asynchronous multi-sensor target tracking adaptive fusion method according to claim 1, wherein, After obtaining the fusion estimation result, the following steps are further included: The target state after fusion and the corresponding covariance matrix are compared with the ground truth data to evaluate the accuracy of the fusion estimation result, a feedback term is designed by using the target local optimal estimation transmitted by the sensor and the fusion estimation result; The weight of the sensor node is adjusted according to the feedback term; The weight of the adjusted sensor node is used for the fusion at the next time, and the fusion process is continuously iterated and improved.

7. An asynchronous multi-sensor target tracking adaptive fusion system, characterized in that, It includes: An initialization module configured to construct a target state equation and a sensor measurement equation; A local fusion module configured to calculate a target local optimal estimation by using an unscented Kalman filter algorithm according to the target state equation and the sensor measurement equation; The process of using the unscented Kalman filter algorithm includes: Initializing the state vector of the target and the corresponding covariance matrix; Based on the target state equation, a set of sampling points are generated by using unscented transformation, the target model propagation sampling points are calculated according to the IMM algorithm, the prior estimation of the target state mean value and the covariance matrix at the next time are calculated, the sensor measurement mean value and its covariance matrix are estimated, the cross-covariance matrix between the target state and the measurement value, the Kalman gain are calculated, and the posterior estimation of the target state local estimation and the covariance matrix are obtained; During the filtering operation process, each time sampling is performed, the uncertainty of the target state estimation is evaluated by analyzing the covariance matrix, and adaptive sampling is realized by adjusting the number of sampling points; when the uncertainty of the system is low, i.e., the target maneuverability is low, the target moves at a constant speed or approximately at a constant speed, the number of sampling points is reduced; when the system uncertainty is high or the tracking performance is reduced, i.e., the target maneuverability is high, the target moves at a variable speed or turns, the number of sampling points needs to be increased; A transmission module configured to build a dynamic communication topology structure according to the sensor nodes, and transmit the target local optimal estimation of each sensor to the fusion center; A global fusion module configured to perform adaptive fusion by the fusion center according to the target local optimal estimation of each sensor to obtain the fusion estimation result.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to realize the steps in the adaptive fusion method of the asynchronous multi-sensor target tracking according to any one of claims 1-6.

9. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the steps in the adaptive fusion method of the asynchronous multi-sensor target tracking according to any one of claims 1-6.

Images