Missile-borne sensor target moving characteristic acquisition method resistant to transmission delay
By employing finite-time control and buffer technology, the problem of error accumulation caused by transmission delay in sensor networks is solved, enabling efficient and accurate fusion of target movement characteristics in sensor networks and improving the stability and accuracy of the system.
Patent Information
- Application Number
- CN202310360359.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-04-06
- Publication Date
- 2025-11-21
- Estimated Expiration
- 2043-04-06
AI Technical Summary
Existing technologies in sensor networks have failed to effectively address the impact of transmission delay on information fusion accuracy and system stability. In particular, distributed filtering algorithms suffer from convergence issues and error accumulation due to hardware differences, which affect target tracking accuracy and system stability.
By employing finite-time control and buffer technology, and designing a buffer structure in the sensor network, the optimal motion characteristic fusion is achieved through a finite number of iterations using communication weights and network graph diameter, thereby reducing the impact of transmission delay on estimation errors.
It achieves convergence in a finite number of iterations, exhibits good robustness and minimum linear variance, and is suitable for weak hardware devices, thus improving the resilience of sensor networks against transmission delays.
Smart Images
Figure CN116389941B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of information processing, and particularly relates to a missile-borne sensor target moving characteristic acquisition method resisting transmission time delay. BACKGROUND
[0002] With the further development of science and technology, in order to realize the accurate acquisition of the surrounding environment and the measured target information, the control system puts forward higher requirements for the accuracy of the measurement information. The basic purpose of the multi-source data fusion technology is to fuse the multi-sensor data and the system estimation data through the corresponding fusion algorithm to obtain more accurate data information than a single sensor. Generally, it can be understood as intelligently comprehensively processing the original information from the multi-sensor, realizing effective multi-source data processing, so as to obtain new and meaningful information. The value of the new obtained information is higher than that of the information obtained by a single sensor, and it is more conducive to the system decision and state perception.
[0003] According to the existing literature, there are two methods of data fusion filtering technology. The first method is centralized filtering, which transmits the observation data of all sensors to the fusion center to generate the highest accuracy state estimation. However, since the data processing center needs to process a large amount of data, it will cause a serious computing burden. At the same time, if part of the sensors fail, it will directly affect the final fusion result, resulting in poor fusion accuracy and stability. In addition, the second method is distributed filtering, which first obtains local estimation at the sensor end, and then obtains the global optimal or suboptimal estimation result according to a certain information fusion algorithm. Distributed filtering has lower complexity and computational amount, and is more robust to sensor failure than centralized filtering.
[0004] In the past few decades, Kalman Filtering (KF) has been widely studied in the fields of target detection and tracking, industrial monitoring, signal processing, etc. Especially the Distributed Kalman Filtering (DKF) method, due to its excellent performance in dynamic target tracking, has been developed for many years. In view of the convenience of multi-sensor network control, it is required that the sensor network can realize consistency, and the consistency control has been widely applied in the fields of unmanned aerial vehicle formation, cluster satellite and mobile robot coordination, etc. Therefore, the consistency control method of the sensor network will be researched in the process of this paper.
[0005] In the reference (Yang H, Li H, Xia Y, et al. Distributed Kalman Filtering Over Sensor Networks With Transmission Delays [J]. IEEE Trans Cybern, 2020), a KF filtering algorithm with a finite length buffer is designed using the timestamp technology, which can realize effective fusion of multi-sensor measurement data under the condition of existing transmission delay, but the algorithm uses the average consensus control strategy. This strategy has the following shortcomings: First, in theory, it needs to be iterated infinitely to reach convergence. Secondly, in practical applications, the consensus iteration stopping strategy is not clear.
[0006] In the reference (Wu Z, Fu M, Xu Y, et al. A distributed Kalman filtering algorithm with fast finite-time convergence for sensor networks [J]. Automatica, 2018, 95:63-72.), a distributed finite-time KF filtering algorithm is proposed, which can realize dynamic monitoring of linear discrete dynamic systems using partially active and partially idle sensor networks, but it does not consider the influence of transmission delay on information fusion accuracy in sensor networks. In practical applications, due to hardware differences and algorithm execution delay, information transmission delay on the communication network is inevitable. Due to the existence of transmission delay, the filtering accuracy of the multi-sensor system may be reduced, resulting in too large state estimation deviation. When the error accumulates to a certain extent, it will even affect the stability of the system, making the filtering algorithm unusable.
[0007] Therefore, in view of the above situation, there is currently no literature that considers the finite-time consensus of sensor networks, communication weights and anti-transmission delay at the same time. It is worth noting that if the sensor system has inconsistency and transmission delay, it will have a negative impact on the detection accuracy of the sensor network, and will also complicate the dynamic target tracking problem. Therefore, it is of great practical significance to study the filtering method of the sensor network. According to the above description, the present application uses finite-time control technology and buffer technology to propose an anti-transmission delay missile-borne sensor target moving characteristic acquisition method. SUMMARY
[0008] The purpose of the present application is to provide an anti-transmission delay missile-borne sensor target moving characteristic acquisition method, so as to realize the existence of.
[0009] The technical solution of the application is a missile-borne sensor target movement characteristic acquisition method against transmission delay, and the steps are as follows:
[0010] Step 1: estimating the initial value of the target movement characteristic by using a mobile target system model, and calculating the movement characteristic fusion result at the current time k=1 and the movement characteristic covariance Step 2 is entered.
[0011] Step 2: performing +1 on the current time k, and judging whether the current time k satisfies 2≤k≤d t +1, d t represents the information transmission delay boundary value, if it is satisfied, the movement characteristic fusion result in the range of [2, d t +1] of the current time k is calculated and the movement characteristic covariance Step 3 is entered; if it is not satisfied, Step 3 is entered.
[0012] Step 3: performing +1 on the current time k, and judging whether the current time k satisfies k≥d t +2; if it is satisfied, the movement characteristic fusion result in the range of [d t +2, +∞] of the current time k is calculated and the movement characteristic covariance Step 4 is entered; if it is not satisfied, Step 4 is entered.
[0013] Step 4: fusing the movement characteristics at different times to obtain the optimal fusion result of the movement characteristic in the sense of the minimum linear variance that is, the accurate target movement characteristic is obtained.
[0014] Compared with the prior art, the application has the following advantages:
[0015] (1) The estimation error obtained by the application can be converged through a limited number of iterations.
[0016] (2) The application has good robustness to time-varying transmission delay.
[0017] (3) The fusion result of the application has the optimal estimation in the sense of the minimum linear variance.
[0018] (4) The application has simple implementation structure and low complexity, and can be run on a weak hardware device. BRIEF DESCRIPTION OF DRAWINGS
[0019] Figure 1 is a data buffer structure schematic diagram of the application.
[0020] Figure 2 is a network schematic diagram between sensors of the application.
[0021] Figure 3 This is a flowchart of a method for acquiring target movement characteristics using a missile-borne sensor that is resistant to transmission delay.
[0022] Figure 4 This is a flowchart of the FtDKF method of the present invention. Detailed Implementation
[0023] This invention proposes a method for acquiring target movement characteristics of airborne sensors that is resistant to transmission delay, utilizing finite-time control technology and buffer technology, which can achieve optimal fusion in the sense of minimum variance.
[0024] Combination Figure 3 and Figure 4 The present invention discloses a method for acquiring target movement characteristics of a missile-borne sensor with resistance to transmission delay, comprising the following steps:
[0025] Step 1: To fuse the moving target detection information from n onboard sensors, assume the expected state of the i-th onboard sensor is μ0 and the covariance is P0; estimate the initial values of the motion characteristics using the moving target system model, and calculate the motion characteristic fusion result at the current time k=1. Motion property covariance The specific calculation method is as follows:
[0026] When the transmission delay between the missile-borne sensors is t=1, according to the measurement matrix of the i-th missile-borne sensor... Measuring noise covariance and the measurement value of the i-th missile-borne sensor on the moving target Calculate the intermediate measurement values from the i-th missile-borne sensor to the j-th missile-borne sensor respectively. and conversion intermediate values
[0027]
[0028]
[0029] Where s represents the information transmission time, s = 1 in this step, T denotes matrix transpose, and i and j represent the i-th and j-th onboard sensors, respectively. The information transmitted at the current time k is temporarily stored... Figure 1 In the cache shown.
[0030] Will and use Figure 2 The communication network between the missile-borne sensors shown transmits data to the j-th missile-borne sensor. Then, the fused value Θ of the measurement motion characteristics from the j-th missile-borne sensor to the i-th missile-borne sensor is calculated respectively. i(t) and the conversion moving characteristic fusion value Ω i (t) is:
[0031]
[0032]
[0033] wherein, is the communication weight between the ith missile-borne sensor and the jth missile-borne sensor, N i represents the set of other sensors having a network connection relationship with the ith missile-borne sensor.
[0034] The moving characteristic fusion result at the current time k = 1 is calculated respectively and the moving characteristic covariance
[0035]
[0036]
[0037] wherein, d g is the network graph diameter between missile-borne sensors, represents the maximum hop count of two missile-borne sensor nodes, Θ i (d g ) represents the measurement moving characteristic fusion value after d g times of information transmission, Ω i (d g ) represents the conversion moving characteristic fusion value after d g times of information transmission, is the prior moving characteristic fusion result, is the prior moving characteristic covariance, and the calculation formula is:
[0038]
[0039]
[0040] wherein, Φ s-1 is the known moving target system matrix, Q s-1 is the missile-borne sensor detection process noise covariance; through information transmission, the moving characteristic fusion result and the moving characteristic covariance
[0041] Step 2: Perform +1 for the current time k, and judge whether the current time k satisfies 2 ≤ k ≤ d t +1, if it satisfies, calculate the moving characteristic fusion result of the current time k in the range of [2, d t +1] moving characteristic covariance If not, go to step 3, as follows:
[0042] Judge whether the current time k satisfies 2≤k≤d t +1, where d t is the information transmission delay boundary value. Let the following equation hold:
[0043]
[0044]
[0045] Step 2.1: For the current time k, the initial value of the transmission time s in the current step is 2, and the initial value of t is 1. Substitute the results of equations (9) and (10) into equations (5) and (6) in step 1, and collect information in the sensor network. Repeat iterations (1) to (4) until t is equal to d g , output Θ i (d g ) and Ω i (d g ). Substitute Θ i (d g ) and Ω i (d g ) into equations (5) and (6) to obtain the current transmission time s and the corresponding mobile characteristic fusion result Mobile characteristic covariance
[0046] Step 2.2: Again, perform +1 on the sensor information transmission time s, and set the initial value of t to 1. Collect information in the sensor network and return to step 2.1; until the transmission time s is equal to the current time k; obtain the mobile characteristic fusion result in the range of [2, d t +1] for the current time k Mobile characteristic covariance
[0047] Step 2.3: Update the current time, perform +1 on the current time k, and return to step 2.1 until the updated k>d t +1; if 2≤k≤d t +1 is not satisfied, go to step 3.
[0048] Step 3: Perform +1 on the current time k, and judge whether the current time k satisfies k≥d t +2. If satisfied, calculate the mobile characteristic fusion result in the range of [d t +2, +∞] for the current time k Mobile characteristic covariance If not, go to step 4 for optimal fusion in the linear variance sense. Details are as follows:
[0049] determine whether the current time k satisfies k≥d t +2, the following equation is established:
[0050]
[0051]
[0052] Step 3.1: In the current step, the initial value of s is k-d t +1, t is initially 1, the results of equations (11) and (12) are substituted into equations (5) and (6) in Step 1, and information in the sensor network is collected, and equations (1) to (4) are repeatedly executed until t is equal to d g , Θ i (d g ) and Ω i (d g ) are output. Θ i (d g ) and Ω i (d g ) are substituted into equations (5) and (6) to obtain the current transmission time s and the corresponding mobile characteristic fusion result mobile characteristic covariance
[0053] Step 3.2: The transmission time s is incremented by 1 again, t is initially 1, information in the sensor network is collected, and Step 3.1 is returned. Until the transmission time s is equal to the current time k, that is, s=k. The mobile characteristic fusion result in the range [d t +2,+∞] of the current time k can be obtained mobile characteristic covariance
[0054] Step 3.3: The current time k is incremented by 1, Step 3.1 is returned, and the process is repeated until the current time k exceeds the detection time constraint boundary. Step 4 is entered to perform optimal fusion in the sense of linear variance.
[0055] If k≥d t +2 is not satisfied, optimal fusion in the sense of linear variance is performed.
[0056] Step 4: Mobile characteristics at different times are fused to obtain an optimal mobile characteristic fusion result in the sense of minimum linear variance as follows:
[0057] Mobile characteristics at different times are fused using the following equation to obtain the final mobile characteristic fusion result
[0058]
[0059] wherein the weight coefficient matrix Γ = Ξ -1 e(e T Ξ -1 e) -1 , e = [I n ,..., I n ] T is an all-one matrix, I n is an identity matrix, and Ξ is a cross-covariance.
[0060] Embodiment 1
[0061] Step 1: To complete the information fusion of 12 missile-borne sensors for detecting a moving target, it is assumed that the expected state of the i-th missile-borne sensor is μ0, and the covariance is P0; the initial value of the motion characteristic is estimated by using the model of the moving target system, and the fusion result of the motion characteristic at the current time k = 1 is calculated as follows: Motion characteristic covariance i = 1,..., 12; the specific calculation method is as follows:
[0062] When the transmission time delay value t = 1 between the missile-borne sensors, according to the measurement matrix of the missile-borne sensor measurement noise covariance and the measurement value of the i-th missile-borne sensor for the moving target the intermediate measurement value from the i-th missile-borne sensor to the j-th missile-borne sensor is calculated as and the conversion intermediate value is calculated as
[0063]
[0064]
[0065] wherein s is the information transmission time, s = 1 in this step, T represents the matrix transpose, i and j represent the i-th missile-borne sensor and the j-th missile-borne sensor respectively. The information transmitted at the current time k is temporarily stored in the buffer shown in Figure 1 .
[0066] The and are transmitted to the j-th missile-borne sensor by using the communication network between the missile-borne sensors shown in Figure 2 . Then, the measurement motion characteristic fusion value Θ i (t) from the j-th missile-borne sensor to the i-th missile-borne sensor and the conversion motion characteristic fusion value Ω i (t) are calculated as follows:
[0067]
[0068]
[0069] wherein, is the communication weight between the ith missile-borne sensor and the jth missile-borne sensor, N i represents a set of other sensors having a network connection relationship with the ith missile-borne sensor.
[0070] respectively calculate the movement characteristic fusion result at the current time k = 1 and the movement characteristic covariance
[0071]
[0072]
[0073] wherein, d g is the network graph diameter between missile-borne sensors, represents the maximum hop count of two missile-borne sensor nodes, and is assumed to be 5, Θ i (d g ) represents the measurement movement characteristic fusion value after 5 times of information transmission, Ω i (d g ) represents the conversion movement characteristic fusion value after 5 times of information transmission, is the prior movement characteristic fusion result, is the prior movement characteristic covariance, and the calculation formula is:
[0074]
[0075]
[0076] wherein, Φ s-1 is the known movement target system matrix, Q s-1 is the missile-borne sensor detection process noise covariance; through information mutual transmission, the movement characteristic fusion result of 12 missile-borne sensors at the time k = 1 is completed and the movement characteristic covariance
[0077] Step 2: perform +1 to the current time k, and judge whether the current time k satisfies 2 ≤ k ≤ d t +1, if it is satisfied, calculate the movement characteristic fusion result of the current time k in the range of [2, d t +1] movement characteristic covariance If it is not satisfied, turn to Step 3, which is as follows:
[0078] judge whether the current time k satisfies 2 ≤ k ≤ d t +1, wherein d tThe information transmission delay boundary value is assumed to be 4 at most. The following equations are established:
[0079]
[0080]
[0081] Step 2.1: For the current time k, the initial value of the transmission time s in the current step is 2, and the initial value of t is 1. The results of equations (9) and (10) are substituted into equations (5) and (6) in step 1, and the sensor network information is collected. Equations (1) to (4) are repeatedly executed until t is equal to d g , and Θ i (5) and Ω i (5) are output. Then, Θ i (5) and Ω i (5) are substituted into equations (5) and (6) to obtain the corresponding mobile characteristic fusion result of the current transmission time s.
[0082] Step 2.2: The transmission time s of the sensor information is increased by 1, and the initial value of t is set to 1. The sensor network information is collected, and the process returns to step 2.1. This process is repeated until the transmission time s is equal to the current time k. The mobile characteristic fusion result in the range of [2, 5] for the current time k is obtained.
[0083] Step 2.3: Update the current time. The current time k is increased by 1, and the process returns to step 2.1. This process is repeated until the updated k > 5. If 2 ≤ k ≤ 5 is not satisfied, the process proceeds to step 3.
[0084] Step 3: The current time k is increased by 1, and it is determined whether the current time k satisfies k ≥ 6. If it is satisfied, the mobile characteristic fusion result in the range of [6, +∞] for the current time k is calculated. If it is not satisfied, the process proceeds to step 4 for linear variance-based optimal fusion. The specific process is as follows:
[0085] When the current time k satisfies k ≥ 6, the following equations are established:
[0086]
[0087]
[0088] Step 3.1: In the current step, the initial value of s is k-3, and the initial value of t is 1. The results of formula (11) and formula (12) are substituted into formula (5) and formula (6) in step 1, and information in the sensor network is collected. Iterative execution of (1) to (4) is repeated until t is equal to 5, and Θ i (5) and Ω i (5) are output. i (5) and Ω i (5) are substituted into formula (5) and formula (6) to obtain the corresponding mobile characteristic fusion result Mobile characteristic covariance
[0089] Step 3.2: The transmission time s is incremented by 1, and the initial value of t is 1. Information in the sensor network is collected, and step 3.1 is returned. Until the transmission time s is equal to the current time k, that is, s=k. The mobile characteristic fusion result of the current time k in the range of [6,+∞] can be obtained Mobile characteristic covariance
[0090] Step 3.3: The current time k is incremented by 1, and step 3.1 is returned until the current time k exceeds the detection time constraint boundary. Step 4 is entered to perform optimal fusion in the sense of linear variance.
[0091] If k≥6 is not satisfied, optimal fusion in the sense of linear variance is performed.
[0092] Step 4: The mobile characteristics at different times are fused to obtain the optimal mobile characteristic fusion result in the sense of minimum linear variance Specifically as follows:
[0093] The mobile characteristics at different times are fused using the following formula to obtain the final mobile characteristic fusion result
[0094]
[0095] Wherein, the weight coefficient matrix Γ=Ξ -1 e(e T Ξ -1 e) -1 , e=[I 12 ,…,I 12 ] T is a full matrix, I 12 is a unit matrix, and Ξ is a cross-covariance.
Claims
1. A method for acquiring target movement characteristics using a missile-borne sensor with resistance to transmission delay, characterized in that, The steps are as follows: Step 1: Estimate the initial values of the target's movement characteristics using the moving target system model, and calculate the fusion result of the movement characteristics at the current time k=1. and movement characteristic covariance If the current time k is a positive integer, increment k by 1 as follows: To achieve the fusion of moving target detection information from n missile-borne sensors, assume that the expected state of the moving target characteristics measured by the i-th missile-borne sensor is μ0, and the covariance of the moving characteristics is P0; Initial values of mobility characteristics are estimated using a moving target system model, and the fusion result of mobility characteristics is calculated at the current time k=1. Mobility characteristic covariance The specific calculation method is as follows: When the transmission delay between the missile-borne sensors is t=1, according to the measurement matrix of the i-th missile-borne sensor... Measuring noise covariance and the measurement value of the i-th missile-borne sensor on the moving target. Calculate the values of the i-th airborne sensor to the j-th airborne sensor respectively. and Where s is the information transmission time, which is a positive integer, and in this step s = 1, T represents matrix transpose; Will and The data is transmitted to the j-th onboard sensor via the communication network between the onboard sensors; then the fused value Θ of the measurement movement characteristics from the j-th onboard sensor to the i-th onboard sensor is calculated respectively. i (t) and the fused value of the conversion movement characteristic Ω i (t): in, N represents the communication weight between the i-th and j-th onboard sensors. i This represents the set of other sensors that have a network connection with the i-th airborne sensor. This represents the intermediate measurement value between the j-th onboard sensor and the i-th onboard sensor. This represents the intermediate value of the conversion from the j-th onboard sensor to the i-th onboard sensor; Calculate the motion characteristic fusion results at the current time k=1 respectively. and movement characteristic covariance Where, d g Θ represents the diameter of the network graph between missile-borne sensors, and Θ represents the maximum number of hops between two missile-borne sensor nodes. i (d g ) indicates after d g The fusion value of the measurement movement characteristics of the secondary information transmission, Ω i (d g ) indicates after d g The fusion value of the transformation mobility characteristics of the secondary information transmission The result is the fusion of prior mobility characteristics. The formula for calculating the prior movement characteristic covariance is: Where, φ s-1 Given the system matrix of the moving target, Q s-1 The noise covariance during the detection process of the missile-borne sensors; through mutual information transmission, the motion characteristics fusion result of n missile-borne sensors at time k=1 is completed. With motion characteristic covariance Increment k by 1 at the current time. Proceed to step 2; Step 2: Determine whether the current time k satisfies 2≤k≤d t +1, d t This represents the boundary value of information transmission delay. If it is satisfied, calculate the current time k in [2, d]. t Motion characteristic fusion results within the range of +1] and movement characteristic covariance Increment k by 1 at the current time and proceed to step 2; if the condition is not met, proceed to step 3. Step 3: Determine whether k at the current time satisfies d. t +2≤k≤T all T all T represents the detection time constraint boundary. all >d t +2; if the calculation of the current time k in [d] is satisfied... t +2,T all Mobility characteristic fusion results within the range and movement characteristic covariance Increment k by 1 at the current time and proceed to step 3; if the condition is not met, proceed to step 4. Step 4: Analyze the movement characteristics at different times. The optimal fusion result of the mobility characteristics is obtained by performing fusion, which has the meaning of minimum linear variance. This means obtaining accurate target movement characteristics.
2. The method for acquiring target movement characteristics of a missile-borne sensor with resistance to transmission delay according to claim 1, characterized in that, In step 2, determine whether the current time k satisfies 2 ≤ k ≤ d. t +1, if the calculation of the current time k is in [2, d t Motion characteristic fusion results within the range of +1] Mobility characteristic covariance If the conditions are not met, proceed to step 3, as follows: Let the initial value of the sending time s be 2, and determine whether the current time k satisfies 2 ≤ k ≤ d. t When +1, where d t Let the information transmission delay boundary value be denoted by ; then let the following equation hold: Step 2.1: For the current time k, the initial value of the transmission delay t is 1. Run equations (1) to (4) to collect information in the sensor network, increment the transmission delay t by 1, and determine whether t satisfies 1≤t≤d. g If satisfied, continue running equations (1) to (4); if not satisfied, output Θ. i (d g ) and Ω i (d g Resolve equations (5) and (6) to proceed to step 2.2; Step 2.2: Fuse the mobility characteristics results from the previous transmission time s-1. Mobility characteristic covariance Substituting into equations (7) and (8), Θ i (d g ) and Ω i (d g Substituting into equations (5) and (6), we obtain the motion characteristic fusion result corresponding to the current transmission time s. Mobility characteristic covariance Proceed to step 2.3; For the first run of step 2, when s=2, the motion characteristic fusion result... Mobility characteristic covariance The results correspond to equations (9) and (10) respectively; Step 2.3: Determine whether the information sending time s satisfies 2 ≤ s < k. If it is satisfied, increment the information sending time s by 1, set the initial value of t to 1, collect the information in the sensor network, and return to Step 2.1; if it is not satisfied, it means that the transmission time s is equal to the current time k, and obtain the fusion result of the movement characteristics of the current time k within the range of [2, d t +1] Movement characteristic covariance Proceed to Step 2.4; Step 2.4: Update the current time. Increment the current time k by 1, and check if k satisfies 2 ≤ k ≤ d. t +1. If satisfied, return to step 2.1; otherwise, proceed to step 3 to solve for d. t +2≤k≤T all Mobility feature fusion results within the range Mobility characteristic covariance 3. The method for acquiring target movement characteristics of a missile-borne sensor with resistance to transmission delay according to claim 2, characterized in that, In step 3, determine whether the current time k satisfies d. t +2≤k≤T all If the calculation of the current time k in [d] is satisfied... t +2,T all Mobility characteristic fusion results within the range Mobility characteristic covariance If the conditions are not met, proceed to step 4 to perform optimal fusion in the sense of linear variance, as follows: In the current step, the initial value of s is kd. t +1, determine if the current time k satisfies d. t +2≤k≤T all When, then let the following expression hold true: Step 3.1: For the current time k, the initial value of the transmission delay t is 1. Run equations (1) to (4) to collect information in the sensor network, increment the transmission delay t by 1, and determine whether t satisfies 1≤t≤d. g If satisfied, continue running equations (1) to (4); if not satisfied, output Θ. i (d g ) and Ω i (d g ) to equations (5) and (6), proceed to step 3.2; Step 3.2: Fuse the mobility characteristics results from the previous transmission time s-1. Mobility characteristic covariance Substituting equations (7) and (8) in claim 2, Θ i (d g ) and Ω i (d g Substituting into equations (5) and (6), we obtain the motion characteristic fusion result corresponding to the current transmission time s. Mobility characteristic covariance Proceed to step 3.3; For the first run of step 3, when s = kd t +1, Mobility Feature Fusion Result Mobility characteristic covariance The results correspond to equations (11) and (12) respectively; Step 3.3: Determine whether the information sending time s satisfies k - d t + 1 ≤ s < k. If it is satisfied, increment the information sending time s by 1, set the initial value of t to 1, collect the information in the sensor network, and return to Step 3.1; if it is not satisfied, it means that the transmission time s is equal to the current time k, and obtain the fusion result of the movement characteristics within the range of [d t + 2, T all for the current time k Movement characteristic covariance Enter Step 3.4; Step 3.4: Update the current time, increment the current time k by 1, and check if k satisfies d. t +2≤k≤T all If satisfied, return to step 3.1; if not satisfied, proceed to step 4 to perform optimal fusion in the sense of linear variance.