An efficient trajectory fusion method based on local matching

Abstract

Aiming to solve the technical problem that the time overhead of mobile entity trajectory data fusion processing increases significantly under the condition of a sharp increase in data traffic, which may cause system crash, the present invention provides an efficient trajectory fusion method based on local matching, which includes the following steps: initial parameter setting, trajectory data rasterization cache processing, cache data life detection, trajectory data sequence matching calculation; matching success condition determination. The present invention can significantly reduce the number of trajectory similarity calculations by pre-grouping mobile entity trajectory data, thereby improving the efficiency of fusion processing. It is suitable for scenarios with diverse trajectory data sources, large data traffic, and wide geographical distribution of trajectories.

Description

Technical Field

[0001] The invention belongs to the field of computer technology, and in particular relates to an efficient trajectory fusion method based on local matching. Background Art

[0002] With the development of electronic information technology, the methods of obtaining the trajectory of mobile entities have gradually become more abundant, such as radar signals, laser positioning equipment, AIS (academic professional organization in the field of information systems, ship positioning technology), ADS-B (automatic dependent surveillance-broadcast, civil aviation positioning technology), etc. Each method can independently obtain the trajectory of mobile entities within a certain time and space range.

[0003] However, for the integrated information system, if the trajectory data from different sources within the same time and space range are accessed, data duplication will occur; in addition, the trajectory data from different sources have different characteristics, and the positioning accuracy and attribute integrity of data from different sources are also different. In order to merge the duplicate trajectories and make the trajectory data of mobile entities unique, complete and accurate, it is necessary to fuse the multi-channel trajectory data accessed in real time.

[0004] The basis of trajectory fusion processing is trajectory similarity calculation. By comparing the trajectory similarity of data from different sources, we can find a trajectory group with high enough similarity, and determine these trajectories as the trajectory of the same mobile entity, and further optimize the attribute completion and positioning accuracy. In the actual system, trajectory data from different sources continuously enter the fusion processing program in the form of data streams. The fusion processing program needs to cache the data for a period of time, and then regularly calculate the similarity of all data with different numbers and sources.

[0005] Assuming that the time cost of the fusion processing program for similarity calculation is c, and the number of cached original mobile entities is n, then the time cost required for one processing process is n×(n-1)×c / 2. It can be seen that with the increase of data traffic, the time cost of each fusion processing process increases significantly, which may cause the system to crash. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides an efficient trajectory fusion method based on local matching. The method greatly reduces the number of trajectory similarity calculations by pre-grouping the mobile entity trajectory data, thereby improving the efficiency of the fusion process.

[0007] The present invention provides an efficient trajectory fusion method based on local matching, which comprises the following steps:

[0008] Step 1: Initial parameter setting, including trajectory data cache life and matching success condition judgment parameters; trajectory data cache life T lifeDefined as: the survival time of trajectory data in each grid area after the real-time trajectory data is rasterized; the matching success condition judgment parameters include: matching radius R e 、M c and matching ratio M p , where the matching radius R e : The upper limit of the position error between two points at the same time when calculating the matching degree of different trajectory data sequences; the number of matches M c and matching ratio M p :Calculate the matching degree of the trajectory data sequences corresponding to two different entity numbers. When the number of matches is not less than M c , and the matching ratio is not less than M p , the match succeeds, otherwise the match fails.

[0009] Step 2: rasterization cache processing of trajectory data, dividing the real-time cached trajectory data into different types of raster areas, completing the grouping process of cached data;

[0010] Step 3, perform life detection on the cached data of any group, and clear the invalid data in the cache in real time according to the trajectory data cache life set in step 1;

[0011] Step 4, trajectory data sequence matching calculation;

[0012] Step 5, matching is performed based on the calculation result of step 4 and the matching success condition judgment parameters set in step 1. If the match is successful, the successfully matched mobile entity trajectories are merged and fused into the same mobile entity; if the match fails, return to step 3 to perform the next grouping judgment process until all cached data are traversed.

[0013] Furthermore, the rasterization cache processing of the trajectory data in step 2 is implemented based on the nearest neighbor method, and the specific process is as follows:

[0014] Step 1: Given t i Time trajectory point T i =(x i ,y i ,z i ), where x i ,y i ,z i are the longitude, latitude and altitude of the trajectory points respectively; T i As a benchmark, determine the grid area R = {P(t)|P(t)≤T i +V max |tt i |}, where P(t) is any trajectory point at time t, V max To preset the maximum speed of the moving entity;

[0015] Step 2: Given t j Time trajectory point T j =(x j ,y j ,z j ), if the trajectory point T j If it is in grid region R, execute step ③, otherwise create a new region R * ={P(t)|P(t)≤T j +V max |tt j |}, continue to step ②;

[0016] Step ③: Update the reference points and grid area;

[0017] If t i ≤t j , the reference point is updated to T j , the grid area is updated to R = R * ={P(t)|P(t)≤T j +V max |tt j |}, otherwise the grid area R remains unchanged;

[0018] Step ④: Repeat this process to update the reference points and grid areas, and divide the real-time trajectory data into different types of grid areas.

[0019] Furthermore, in step 3, the survival time t of the trajectory data of the i-th (i=1, 2, 3, ...) grid-like region is detected in real time. i , when t i >T life When , all trajectory data in the i-th class grid area are cleared, otherwise all trajectory data in the i-th class grid area are retained, where T life Cache lifespan for track data.

[0020] Furthermore, in step 4, the matching degree calculation of the trajectory data sequence is implemented based on the interpolation algorithm, and the calculation steps are as follows:

[0021] Step ①: Sort the trajectory data sequences in the same type of grid area according to the number of point columns from large to small, denoted as s i (i=1,2,3,…);

[0022] Step ②: Take the trajectory data sequence with a larger number of points as the benchmark and i and j (i=1,2,3,…,j=1,2,3,…,i≠j) to calculate the matching degree;

[0023] Step 3: Calculate the data sequence s based on the interpolation algorithmi With data series s j The matching degree between them, that is, calculating the same time s i Data sequence trajectory points and s j The distance difference between the trajectory points of the data sequence is smaller than the matching radius R. e When the number of matches is m j Add 1.

[0024] Furthermore, in step 5, when the number of matches m j Greater than or equal to a given threshold M c , and the number of matches is m j With s j The ratio of the number of data sequences is greater than the given matching ratio M p When , the match is successful, and the successfully matched mobile entity trajectories are merged to become the same mobile entity; otherwise, the match fails and the next grouping decision process is carried out.

[0025] Beneficial effects:

[0026] This method significantly reduces the number of trajectory similarity calculations by pre-grouping the mobile entity trajectory data, thereby improving the efficiency of fusion processing. It is particularly suitable for scenarios with diverse trajectory data sources, large data traffic, and wide geographical distribution of trajectories.

[0027] Specifically: the time cost of judging the grouping is less than the trajectory similarity calculation. By pre-grouping, the number of trajectory similarity calculations is reduced, which greatly reduces the time cost, thereby improving the efficiency of trajectory fusion processing. Assuming that the time cost of the fusion processing program for similarity calculation is c, n entities are divided into p groups through preprocessing, where the number of groups with more than 1 entities in the group is p2, and the maximum number of entities in the group is q, then the maximum time cost required for one processing process is q×(q-1)×p2×c / 2. It can be deduced from the calculation formula that the time cost of this method is much less than that of the traditional method. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Mapping of geographic grids to 2D arrays;

[0029] Figure 2 Screenshot of the aircraft trajectory fusion test system;

[0030] Figure 3 Screenshot of the ship trajectory fusion test system. DETAILED DESCRIPTION

[0031] The main idea of ​​the present invention is to group the cached trajectory data according to the spatiotemporal correlation attributes, and put the mobile entities that meet the fusion conditions into the same group; then, the trajectory similarity calculation is performed in the group containing multiple mobile entities through an asynchronous thread at regular intervals.

[0032] The flowchart of the whole process is as follows Figure 1 As shown, the efficient trajectory fusion method based on local matching degree of the present invention includes the following steps:

[0033] 1. Initial parameter setting

[0034] According to the different confidence levels of real-time trajectory data from different reporting units, the corresponding initial parameters are set. This mainly includes the trajectory data cache life T life And the matching success condition judgment parameter, that is, the matching radius R e 、M c and matching ratio M p .

[0035] Trajectory data cache lifespan T life : The survival time of trajectory data in each grid area after real-time trajectory data is rasterized. The value range is generally [1,30] (in minutes). The longer the time, the more accurate the trajectory similarity calculation, but the time cost is also greater. It needs to be adjusted according to the data volume and hardware performance.

[0036] Matching radius R e : The upper limit of the position error between two points at the same time when calculating the matching degree of different trajectory data sequences. The value range is generally [1,1000] (unit: meter) and needs to be adjusted according to the positioning accuracy of the data source.

[0037] Matching number M c and matching ratio M p :Calculate the matching degree of the trajectory data sequences corresponding to two different entity numbers. When the number of matches is not less than M c , and the matching ratio is not less than M p When , the match is successful, otherwise the match fails. c The value range of M is [1,1000], which can be adjusted according to the strictness of the fusion judgment. The larger the value, the stricter the fusion judgment. p The value range is [0.8,0.99].

[0038] 2. Trajectory data rasterization cache processing

[0039] The rasterization cache processing of trajectory data is based on the nearest neighbor method. The specific process is as follows:

[0040] Step 1: Given t i Time trajectory point T i =(x i ,y i ,z i ), where x i ,y i ,zi are the longitude, latitude and altitude of the trajectory points respectively; T i As a benchmark, determine the grid area R = {P(t)|P(t)≤T i +V max |tt i |}, where P(t) is any trajectory point at time t, V max To pre-set the maximum speed of the moving entity, if the moving entity is an airplane, then |V max |=5*340m / s, if the moving entity is a ship, then |V max |=150*1852 / 3600m / s; if the moving entity is a vehicle, then |V max |=350000 / 3600m / s

[0041] Step 2: Given t j Time trajectory point T j =(x j ,y j ,z j ), if the trajectory point T j If it is in grid region R, execute step ③, otherwise create a new region R * ={P(t)|P(t)≤T j +V max |tt j |}, continue to step ②;

[0042] Step ③: Update the reference points and grid area. i ≤t j , the reference point is updated to T j , the grid area is updated to R = R * ={P(t)|P(t)≤T j +V max |tt j |}, otherwise the grid area R remains unchanged;

[0043] Step ④: Repeat this process to update the reference points and grid areas, and divide the real-time trajectory data into different types of grid areas.

[0044] This step is the process of grouping cached data, which is also a key step to improve fusion efficiency.

[0045] 3. Cache data life detection

[0046] Real-time detection of the survival time t of the trajectory data of the i-th (i=1,2,3,…) class grid area i , when t i >T lifeWhen , all trajectory data in the i-th class grid area are cleared, otherwise all trajectory data in the i-th class grid area are retained.

[0047] 4. Calculation of trajectory data sequence matching degree

[0048] The matching degree calculation of trajectory data sequence is realized based on interpolation algorithm. The calculation steps are as follows:

[0049] Step ①: Sort the trajectory data sequences in the same type of grid area according to the number of point columns from large to small, denoted as s i (i=1,2,3,…);

[0050] Step ②: Take the trajectory data sequence with a larger number of points as the benchmark and i and j (i=1,2,3,…,j=1,2,3,…,i≠j) to calculate the matching degree;

[0051] Step 3: Calculate the data sequence s based on the interpolation algorithm i (benchmark data series) and data series s j That is, calculate the matching degree between the same time s i Data sequence trajectory points and s j The distance difference between the trajectory points of the data sequence is smaller than the matching radius R. e When the number of matches is m j Add 1.

[0052] 5. Determination of successful matching conditions

[0053] When the number of matches m j Greater than or equal to a given threshold M c , and the number of matches is m j With s j The ratio of the number of data sequences is greater than the given matching ratio M p When , the match is successful, and the successfully matched mobile entity trajectories are merged to become the same mobile entity; otherwise, the match fails and the next grouping decision process is performed.

[0054] The simulation test verifies that the method of the present invention can run stably under the data flow of 100,000 trajectory points per minute. The screenshot of the simulation test system is as follows: Figure 2 and Figure 3 As shown, Figure 2 is the trajectory of the aircraft that successfully achieved fusion, Figure 3 is the trajectory of a ship that has successfully achieved integration.

[0055] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An efficient trajectory fusion method based on local matching, characterized in that: The following steps are involved: Step 1: Initial parameter setting, including trajectory data cache life and matching success condition determination parameters; Step 2: rasterization cache processing of trajectory data, dividing the real-time cached trajectory data into different types of raster areas, completing the grouping process of cached data; The rasterization cache processing of the trajectory data class in step 2 is implemented based on the nearest neighbor method, and the specific process is as follows: Step ①: Given Time track point ,in are the longitude, latitude and altitude of the trajectory point respectively; by As a benchmark, determine the grid area , where for Any trajectory point at any time, To preset the maximum speed of the moving entity; Step 2: Given Time track point , if the trajectory point In the grid area , execute step ③, otherwise create a new area , continue to step ②; Step ③: Update the reference points and grid area; like , the reference point is updated to , the grid area is updated to , otherwise the grid area remain unchanged; Step ④: Repeat this process to update the reference points and grid areas, and divide the real-time trajectory data into different types of grid areas; Step 3, perform life detection on the cached data of any group, and clear the invalid data in the cache in real time according to the trajectory data cache life set in step 1; Step 4, trajectory data sequence matching calculation; Step 5: Match the calculation result of step 4 with the matching success condition judgment parameters set in step 1. If the match is successful, merge the successfully matched mobile entity trajectories and merge them into the same mobile entity. If the match fails, return to step 3 to perform the next grouping judgment process until all cached data are traversed. Wherein, in said step 1: Track data cache lifespan Defined as: the survival time of trajectory data in each grid area after the real-time trajectory data class is rasterized; The parameters for determining successful matching conditions include: matching radius , Matching number and matching ratio , where the matching radius : When calculating the matching degree of different trajectory data sequences, the upper limit of the position error between two points at the same time; the number of matches and matching ratio :Calculate the matching degree of the trajectory data sequences corresponding to two different entity numbers. When the number of matches is not less than , and the matching ratio is not less than , the match succeeds, otherwise the match fails.

2. The efficient trajectory fusion method based on local matching degree as claimed in claim 1, characterized in that: In step 3, the survival time of the trajectory data of the i-th grid region is detected in real time. , where i=1,2,3,..., when When All trajectory data in the grid area of ​​the class, otherwise keep the All trajectory data in the grid area of ​​each class, among which, Cache lifespan for track data.

3. The efficient trajectory fusion method based on local matching degree as claimed in claim 1, characterized in that: In step 4, the matching degree calculation of the trajectory data sequence is implemented based on the interpolation algorithm, and the calculation steps are as follows: Step ①: Sort the trajectory data sequences in the same type of grid area according to the number of point columns from large to small, denoted as s i , where i=1,2,3,…; Step ②: Take the trajectory data sequence with a larger number of points as the benchmark and and Calculate the matching degree, where i=1,2,3,...,j=1,2,3,...,i≠j; Step 3: Calculate the data sequence based on the interpolation algorithm With data series The matching degree between them, that is, calculating the same time Data series trajectory points and The distance difference between the trajectory points of the data sequence is smaller than the matching radius. When the number of matches Add 1.

4. The efficient trajectory fusion method based on local matching degree as claimed in claim 3, characterized in that: In step 5, when the number of matches Greater than or equal to a given threshold , and the number of matches and The ratio of the number of data sequences is greater than the given matching ratio When , the match is successful, and the successfully matched mobile entity trajectories are merged to become the same mobile entity; otherwise, the match fails and the next grouping decision process is carried out.

Images