Improved particle filter map matching method based on rail transit 5G positioning

Abstract

The invention discloses an improved particle filter map matching method based on rail transit 5G positioning, and the method mainly comprises the steps: 1) collecting original trajectory data obtained through 5G positioning, carrying out the sliding average processing according to a preset sliding window, generating a smooth new trajectory point sequence, and reducing the data redundancy; 2) constructing an improved particle filter map matching algorithm on the basis of the new track points, combining with subway tunnel fine map structure generation, considering space prior information such as a track center line and a track boundary, endowing particles with differentiated weights through a track center line judgment function and a track area judgment function, and determining the track area according to the differentiated weights; high-robustness matching of the track points and the tunnel environment is realized; and 3) in a particle filtering iteration process, setting an effective particle number threshold value, triggering a resampling operation when particles are seriously degraded, selecting a new particle set in a probability mode, avoiding falling into a degradation endless loop, and outputting a state posteriori estimation result in combination with a particle weight to generate an accurate trajectory.

Description

Technical Field

[0001] The present invention relates to the technical field of map matching, and particularly relates to an improved particle filter map matching method based on rail transit 5G positioning. Background Art

[0002] With the rapid layout and development of the fifth-generation mobile communication technology (5G) in vertical industries, the application demand for high-precision positioning in fields such as smart cities and intelligent transportation is becoming increasingly urgent. Especially in the underground environment of urban rail transit systems, such as subway concourses, platforms, entrances and exits, equipment areas, tunnel sections and inside trains, there is a higher demand for real-time and high-precision position perception. Due to the general characteristics of the subway environment, such as being enclosed, having a complex structure, and serious signal occlusion, traditional satellite-based positioning systems, such as GPS (Global Positioning System) and BDS (Beidou Navigation Satellite System), cannot work properly in the underground space and it is difficult to meet the requirements of positioning accuracy, continuity and reliability for practical applications.

[0003] To solve the above problems, in recent years, non-GNSS high-precision positioning technologies based on 5G communication networks have emerged. 5G positioning can provide high-precision positioning capabilities without relying on satellite signals by utilizing its advantages of large bandwidth, low latency and high-density deployment, combined with various ranging and angle measurement methods such as TOA (Time of Arrival), TDOA (Time Difference of Arrival), DOA (Direction Of Arrival) / AOA (Angle Of Arrival). In the subway scenario, by deploying 5G base stations in areas such as concourses, platforms and tunnels, and cooperating with user terminals or sensing devices, continuous and stable position perception can be achieved, providing technical support for intelligent dispatching, emergency management, passenger navigation, etc.

[0004] However, in underground tunnels, leaky coaxial cables are usually used to achieve signal coverage, and it becomes extremely difficult to implement angle measurement through the DOA / AOA algorithm. The internal facilities in the subway project are complex, the metal reflection is strong, and the signal propagation path is variable. Even for 5G positioning systems, there will inevitably be a certain degree of error, causing some positioning points to deviate from the actual trajectory, and even abnormal situations such as derailment may occur. This positioning error will seriously affect the accuracy of subsequent location services and trajectory analysis, and high-precision 5G positioning cannot be achieved in the real environment. Therefore, it is necessary to introduce map matching technology to fuse the collected positioning trajectory data with the subway line map, correct the deviated positioning points, and improve the robustness and credibility of the trajectory data.

[0005] Currently, map matching technologies are mainly divided into two categories: The first category is the indoor map matching method represented by the Simultaneous Localization and Mapping (SLAM) algorithm for indoor robots; the second category is the trajectory correction algorithm for complex urban road environments, which is mainly used to correct positioning deviations when the GPS or Beidou system is interfered. The latter mostly performs rough matching based on the road network topology and can usually only determine that the target is on a certain road, making it difficult to further accurately locate to a specific position point and unable to meet the requirements of high-precision positioning scenarios such as the subway.

[0006] Currently, map matching algorithms are mainly used to handle the problem that the positioning points deviate from the actual road or track. According to their spatial matching relationships, they can be divided into two categories: deterministic map matching algorithms and non-deterministic map matching algorithms. Among them, deterministic map matching algorithms such as the projection method mainly project the positioning points onto the line segments or nodes on the map based on geometric relationships. However, this method requires that the positioning points themselves must fall within the map, with poor fault tolerance and limited ability to adapt to complex environments. Therefore, non-deterministic map matching algorithms based on probability information are usually adopted to improve the robustness of map matching. Among non-deterministic map matching algorithms, the particle filter algorithm is a typical representative. It is an approximate Bayesian filtering algorithm based on Monte Carlo simulation. Its core idea is to use a number of discrete random samples (i.e., particles) to replace the integral calculation, and approximate the posterior probability density of the system by continuously updating the weights and states of the particles, so as to achieve the estimation of the target state. In the subway environment, the trajectory is usually long and the error changes are complex. Traditional particle filters are prone to particle degeneracy and resampling dilemmas when dealing with such scenarios, resulting in a decrease in algorithm accuracy or even falling into a computational dead loop.

[0007] The currently more common solution is to increase the number of particles to enhance system stability, but this will significantly increase the computational burden and affect the real-time performance and usability of the positioning system, especially in subway applications with high requirements for response speed. Therefore, there is an urgent need for a particle filter map matching algorithm that optimizes data processing and computational resource consumption while maintaining accuracy to improve the overall performance of the positioning system in the subway environment. Summary of the Invention

[0008] This application provides an improved particle filter map matching method based on rail transit 5G positioning to solve the problems of insufficient accuracy and excessive computational burden of existing map matching algorithms in real complex subway environments.

[0009] According to the first aspect, an embodiment provides an improved particle filter map matching method based on rail transit 5G positioning, and the method includes: Collect positioning trajectory data based on 5G positioning, and perform moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data; For each trajectory point in the new positioning trajectory data, use the improved particle filter map matching algorithm model for map matching and position correction, including: a. Particle initialization; b. Calculate importance sampling; c. Particle weight update, and the particle weight update strategy includes: for particles outside the track area, the weight is set to 0; for particles inside the track area, particles within the preset distance range from the track centerline are given the maximum weight, and the remaining particles inside the track area are given equal weights; d. Judge whether resampling is needed according to the number of effective particles. If so, perform resampling; e. Calculate the posterior state estimate; f. Obtain the next trajectory point in the positioning trajectory point sequence, and repeat steps b - e until all trajectory points are processed.

[0010] Furthermore, collecting positioning trajectory data based on 5G positioning specifically includes: Deploy multiple 5G base stations in the track area. The terminal device set on the vehicle obtains positioning trajectory data by communicating with surrounding 5G base stations, and transmits the positioning trajectory data to the data collection and processing terminal in real time.

[0011] Furthermore, performing moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data specifically includes: Set the threshold of the number of positioning trajectory points , if the number of collected positioning trajectory points is , then judge whether it holds. If it holds, perform weighted average processing on several consecutive positioning points within the preset sliding window to generate a smooth new trajectory point sequence.

[0012] Furthermore, particle initialization specifically includes: Let the time , according to the distribution of the prior probability density function, generate particles and form the initial particle set , and the initial weight of each particle is assigned as , is the initial state estimate value of the i th particle.

[0013] Furthermore, calculating importance sampling specifically includes: The selected sampling function is a probability density function with the same distribution as the prior probability density function. After sampling, the particle importance weights are calculated and normalized.

[0014] Further, the particle weight update specifically includes: Determine the boundary of the orbit region and the orbit center line; Establish an orbit region determination function and an orbit center line determination function according to the particle weight update strategy; Update the weights of the particles according to the orbit region determination function and the orbit center line determination function, and normalize the updated weights to obtain a new particle set and the corresponding normalized weights, denotes the state vector of the k -th particle at i time, N being the total number of current particles; Among them, first, the weight update is performed according to the orbit region determination function: ; In the formula, is the weight of the i -th particle after update at k time; indicates that the particle is not within the orbit region; Then, the weight update is performed according to the orbit center line determination function: ; In the formula, is the weight of the i -th particle after update at k time; means that the particle is within a preset range from the orbit center line.

[0015] Further, determine whether resampling is required based on the number of effective particles, specifically including: The number of effective particles is: ; If is satisfied, then resampling is required.

[0016] Further, resampling specifically includes: Resample from the particle set according to the updated and normalized particle weights to obtain a new particle set , and reassign the weight of each particle to , N being the total number of current particles.

[0017] Further, the posterior state estimate is calculated, specifically including: According to the particle set and the weights the posterior state estimate is obtained : ; wherein, represents the state vector of the i th particle after resampling at time k , represents the weight of the i th particle after resampling at time k , N is the total number of current particles.

[0018] According to a second aspect, an improved particle filter map matching system based on rail transit 5G positioning is provided in an embodiment. The system includes: A positioning trajectory acquisition module, configured to collect positioning trajectory data based on 5G positioning, and perform moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data; A map matching module, configured to perform map matching and position correction on each trajectory point in the new positioning trajectory data by using an improved particle filter map matching algorithm model, including: a. Particle initialization; b. Calculate importance sampling; c. Particle weight update, and the particle weight update strategy includes: for particles outside the track area, the weight is set to 0; for particles inside the track area, particles within a preset distance range from the track centerline are given the maximum weight, and the remaining particles inside the track area are given equal weights; d. Determine whether resampling is required according to the number of effective particles. If so, perform resampling; e. Calculate the posterior state estimate; f. Obtain the next trajectory point in the positioning trajectory point sequence, and repeat steps b - e until all trajectory points are processed.

[0019] The present application provides an improved particle filter map matching method based on rail transit 5G positioning, which is applicable to scenarios where there is a need for high-precision positioning trajectory correction in underground spaces such as rail transit tunnels, and has the following beneficial effects: (1) The present invention improves the efficiency of algorithm operation. Existing map matching technologies are mainly divided into two categories: one is the SLAM algorithm for indoor robot scenarios, mainly used for dynamic mapping and positioning; the other is the map matching algorithm for correcting GPS / BDS trajectories in complex urban road environments. The above existing methods can usually only determine the road section to which the target belongs, and it is difficult to further accurately locate to a specific position point, unable to meet the actual needs of high-precision positioning scenarios such as subways. In occasions with complex spatial structures and high positioning accuracy requirements such as subway concourses, platforms, and tunnels, traditional map matching methods face problems of poor adaptability and low efficiency. For this reason, the present invention proposes an improved particle filter map matching technology based on 5G positioning in the rail transit environment. By performing moving average processing on the trajectory data collected by the 5G positioning system, reducing the data dimension and the number of particles, the calculation efficiency and operation speed of the algorithm are significantly improved while ensuring the retention of trajectory features, which is applicable to real-time trajectory correction and position perception in large and complex environments such as subways.

[0020] (2) The present invention improves the accuracy of positioning, making the positioning result more accurate. The present invention fully combines the structural environment of the subway tunnel and the trajectory laws of personnel or trains, and introduces two types of spatial determination functions: the track area determination function and the track centerline determination function, to replace the observation likelihood model in traditional filtering. During the particle filtering process, track constraint information is introduced. By setting the particle weight strategy: the weight of particles outside the track is set to 0, the particles near the track centerline are given the maximum weight, and the remaining particles inside the track are given equal weights, the final particle filtering result is more consistent with the real trajectory. Through experimental verification, the positioning trajectory processed by the algorithm of the present invention is not only smoother, with smaller error fluctuations, but also the average relative error with the real trajectory is significantly reduced, having good accuracy performance and application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flowchart of an improved particle filter map matching method based on rail transit 5G positioning provided by an embodiment of the present invention; Figure 2 It is a specific algorithm flowchart in the improved particle filter map matching method based on rail transit 5G positioning provided by an embodiment of the present invention; Figure 3 It is an overall architecture diagram of an improved particle filter map matching method based on rail transit 5G positioning provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The present invention will be further described in detail below in conjunction with specific embodiments and the accompanying drawings. Similar elements in different embodiments are labeled with related similar element numbers. In the following embodiments, many detailed descriptions are provided to enable a better understanding of the present application. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, or methods. In some cases, some operations related to the present application are not shown or described in the specification to avoid overwhelming the core part of the present application with excessive descriptions. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the description in the specification and the general technical knowledge in the field.

[0023] In addition, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can also be reordered or adjusted in a manner obvious to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for clearly describing a certain embodiment and do not mean that they are the necessary sequences, unless it is stated that a certain sequence must be followed.

[0024] An improved particle filter map matching method based on rail transit 5G positioning provided by the first embodiment of the present invention is applicable to scenarios where there is a need for high-precision positioning trajectory correction in underground spaces such as rail transit tunnels. This technology addresses the problems of fluctuating positioning accuracy and trajectory deviation from the actual track centerline in the current rail transit environment. It integrates core modules such as data dimensionality reduction preprocessing, trajectory probability matching, and particle filter state estimation, aiming to improve trajectory matching accuracy, suppress positioning noise, and reduce the computational burden of the particle filter. The following will be described in detail in conjunction with Figure 1 , Figure 2 and Figure 3 for detailed description.

[0025] As Figure 1 shown, in step S100, positioning trajectory data is collected based on 5G positioning, and the collected positioning trajectory data is preprocessed by moving average to generate new positioning trajectory data.

[0026] Specifically, by collecting the original trajectory data obtained by the 5G positioning system and performing weighted average processing on a continuous number of positioning points within a preset sliding window, a smooth new trajectory point sequence can be generated, which can reduce data redundancy, improve the subsequent processing efficiency, and form a low-dimensional input suitable for filtering and matching.

[0027] The above steps specifically include: S110. Deploy multiple 5G small base stations at key positions in the track area to be responsible for providing real-time positioning services. The terminal device on the vehicle obtains location information by communicating with the surrounding 5G base stations, and the positioning data is transmitted in real time to the laptop responsible for data collection and processing.

[0028] S120. Set the threshold of the number of positioning track points. , if the number of collected positioning track points is , then judge whether it holds. If it holds, perform weighted average processing on several consecutive positioning points within the preset sliding window to generate a smooth new track point sequence. Specifically, when , in order to retain the characteristics of each positioning point track, select the positioning point at the current moment k , and take the subsequent positioning points as a group. Give different weights to these M positioning points in the group, and take the weighted average to obtain a new positioning point. The magnitude of the weight can be obtained according to the relative error magnitude of the positioning points, or the same weight can be directly assigned.

[0029] In this embodiment, by performing sliding average preprocessing on the trajectory data obtained based on 5G positioning technology, several consecutive positioning points in the original trajectory are grouped and averaged to generate a group of new trajectory points for map matching processing, thereby significantly reducing the data volume while retaining the trajectory form characteristics, reducing the particle filter calculation overhead, and improving the system operation efficiency. This method is particularly suitable for trajectory correction and high-precision positioning applications in 5G system signal coverage environments such as subway concourses, platforms, and tunnels.

[0030] As Figure 1 shown, in step S200, for each trajectory point in the new positioning trajectory data, use the improved particle filter map matching algorithm model to perform map matching and position correction.

[0031] As Figure 2 shown, the improved particle filter map matching algorithm specifically includes the following steps: a. Particle initialization.

[0032] Specifically, at the initial time when the moment , according to the distribution of the prior probability density function , generate N particles to form the initial particle set . The initial weight of each particle is assigned as , is the initial state estimate value of the i -th particle.

[0033] Wherein kIndicates a time index; i Indicates the i th particle, used as an index in the particle set ; Indicates the i th particle's state vector at k moment ; , where is the k moment's two-dimensional position for locating the target, Indicates the lateral position coordinate of the particle at the k th moment, Indicates the longitudinal position coordinate of the particle at the k th moment; is the k moment's motion direction angle relative to k-1 moment; Indicates the k moment's observation vector, , where Indicates the k moment's two-dimensional position measured based on 5G positioning, is the lateral position coordinate, is the longitudinal position coordinate; is the k moment's motion direction angle relative to k-1 moment derived based on 5G positioning.

[0034] And update the moment according to .

[0035] b. Calculate importance sampling.

[0036] Specifically, sample the state: The sampling function is , which is a probability density function with the same distribution as the prior probability density function , also known as the importance density function. After sampling, a new particle set at the moment is obtained; Calculate the importance weights: ; ; Among them, Indicates the i th particle's normalized weight at k-1 moment; Indicates the i th particle's unnormalized weight at k moment, and further perform normalization processing.

[0037] c. Particle weight update. The particle weight update strategy includes: for particles outside the orbit region, the weight is set to 0; for particles inside the orbit region, particles within the preset distance range from the orbit centerline are given the maximum weight, and the remaining particles inside the orbit region are given equal weights.

[0038] Specifically, update the weights of the particles according to the orbit region determination function and the orbit centerline determination function, and normalize the weights to obtain a new particle set and the corresponding weights , denotes the state vector of the k -th particle at i time, N being the total number of current particles; Among them, first update the weights according to the orbit region determination function: ; In the formula, is the weight of the updated i -th particle at k time; indicates that the particle is not within the orbit region; Then update the weights according to the orbit centerline determination function: ; In the formula, is the weight of the updated i -th particle at k time; means that the particle is within the preset range from the orbit centerline; And further normalize the updated weights obtained according to the orbit region determination function and the orbit centerline determination function.

[0039] d. Judge whether resampling is needed according to the number of effective particles. If needed, perform resampling.

[0040] Specifically, judge whether to perform resampling: Resampling is mainly to solve the phenomenon that particle depletion occurs, resulting in an unsatisfactory particle filtering effect. Determine whether to perform resampling by calculating the number of effective particles. The judgment criterion is as follows: ; Among them, represents the number of effective particles. If is satisfied, resampling is required.

[0041] Resampling specifically includes: Select from the particle set Resample according to the corresponding updated and normalized particle weights to obtain a new particle set , and re-assign the weight of each particle to , N is the total number of current particles.

[0042] e. Calculate the posterior state estimate; According to the particle set and the weight obtain the posterior probability estimate and the posterior state estimate : ; ; wherein, represents the state vector of the system at time k , represents all the observation vector sequences from time 1 to k , N is the total number of particles, represents the state vector of the i -th particle after resampling at time k (i.e., the estimate of the current state by this particle), represents the normalized weight of the i -th particle after resampling at time k , is the Dirac function, is the estimated system state (the output of the particle filter), N is the total number of current particles.

[0043] f. Obtain the next trajectory point in the positioning trajectory point sequence, and repeat steps b - e until all trajectory points are processed. Specifically, when the next measurement value (the next trajectory point in the positioning trajectory point sequence) is imported, repeat steps b - d. If there is no next measurement value, end the task.

[0044] In summary, in this embodiment, an improved particle filter map matching model is constructed based on the new trajectory point. The particle initialization is generated in combination with the subway tunnel digital precision map structure, considering spatial prior information such as the track center line and tunnel boundary. Differentiated weights are assigned to the particles through the track center line determination function and the track area determination function to achieve a highly robust matching between the trajectory point and the tunnel environment; during the particle filter iteration process, an effective particle number threshold is set, and when the particles degenerate severely, a resampling operation is triggered, and a new particle set is selected in a probabilistic manner to avoid falling into a degenerate dead loop, and the state posterior estimation result is output in combination with the particle weights to generate an accurate trajectory.

[0045] Next, in combination with the experimental data, according toFigure 3 The architecture diagram shown below further illustrates the technical effects of the present invention: 1. Experimental conditions: The experimental environment is within a certain residential community in a certain city, and the test area includes a complete road structure.

[0046] The hardware devices are as follows: 20 5G small base stations are evenly deployed on both sides of the community roads as wireless positioning base stations; multiple test vehicles are equipped with 5G terminal positioning modules to communicate with the base stations to obtain real-time positions; each test vehicle is additionally equipped with an RTK (Real - time kinematic) high-precision positioning module to collect real trajectory data; a laptop computer: serves as a data aggregation and processing platform, responsible for receiving and analyzing positioning results.

[0047] The software platform is: Windows10 operating system and Keil uVision5 software.

[0048] 2. Experimental content and result analysis: To verify the operation efficiency of the present invention, a running time comparison experiment of the method of this embodiment (improved particle filter map matching algorithm) and the prior art (traditional particle filter map matching algorithm) is carried out using 3 groups of trajectory data, as shown in Table 1: Table 1 Comparison of running times of different algorithms

[0049] As can be seen from Table 1, the running time of the method of this embodiment is significantly reduced compared with the prior art. By analyzing the data of the three groups of trajectory data, it is found that the average running time of the algorithm is reduced by 42.9%, improving the running efficiency of the algorithm.

[0050] According to the size of the 5G positioning trajectory data volume, an algorithm for adaptively taking the average is designed to obtain new positioning data. The function of this algorithm is to reduce the data volume as much as possible and preferably maintain some digital characteristics of the original data (the original data refers to the 5G positioning data without any processing), such as the obtained average relative positioning error and the approximate position of the trajectory points are basically the same as the original data. After calculation, the error analysis is carried out by taking the average of 2 - 5 data and comparing it with the real positioning data. The results are shown in Table 2: Table 2 Error results of taking the average of several data

[0051] From the results in Table 2 and the results obtained according to the improved particle filter algorithm, it is found that when The result error after the improved particle filter is minimized when [conditions are met]. Its average relative positioning error is 0.2744m, the minimum relative error is 0.0002m, and the maximum relative error is 0.5857m. It can be obtained that the average relative error after the improved particle filter map matching algorithm ( ) is reduced by 47.7% compared to the average relative error of the original 5G positioning trajectory, improving the relative positioning accuracy.

[0052] In summary, in this embodiment, a tunnel scenario is simulated in a cell environment, a 5G base station positioning system is constructed, and it is verified in combination with RTK data. The improved particle filter map matching technology based on 5G positioning in a rail transit environment realizes the error correction of the 5G positioning trajectory, significantly improving the positioning accuracy. The experimental results show that the method proposed in this embodiment effectively reduces the error and improves the operation efficiency while maintaining the data characteristics. In this way, the practicability and stability of the positioning system are improved, providing new ideas and methods for the research of subsequent high-precision positioning algorithms.

[0053] Corresponding to the improved particle filter map matching method based on rail transit 5G positioning disclosed above, an embodiment of the present invention also discloses an improved particle filter map matching system based on rail transit 5G positioning, which specifically includes: A positioning trajectory acquisition module, configured to collect positioning trajectory data based on 5G positioning, and perform moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data; A map matching module, configured to perform map matching and position correction on each trajectory point in the new positioning trajectory data by using an improved particle filter map matching algorithm model, including: a. Particle initialization; b. Calculate importance sampling; c. Particle weight update, and the particle weight update strategy includes: for particles outside the track area, the weight is set to 0; for particles inside the track area, particles within a preset distance range from the track centerline are given the maximum weight, and the remaining particles inside the track area are given equal weights; d. Determine whether resampling is required based on the number of effective particles. If so, perform resampling; e. Calculate the posterior state estimate; f. Obtain the next trajectory point in the positioning trajectory point sequence, and repeat steps b - e until all trajectory points are processed.

[0054] It should be noted that for the detailed description of an improved particle filter map matching system based on rail transit 5G positioning provided in an embodiment of the present invention, reference can be made to the relevant description of an improved particle filter map matching method based on rail transit 5G positioning provided in an embodiment of the present application, which will not be elaborated here.

[0055] Those skilled in the art can understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the programs can be stored in a computer-readable storage medium, which can include: read-only memory, random access memory, magnetic disks, optical discs, hard disks, etc. The above functions can be realized by a computer executing these programs. For example, the program is stored in the memory of the device, and when the processor executes the program in the memory, all or part of the above functions can be realized. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the programs can also be stored in storage media such as servers, other computers, magnetic disks, optical discs, flash drives or external hard drives, and saved to the memory of the local device by downloading or copying, or the system of the local device can be updated. When the processor executes the program in the memory, all or part of the functions in the above embodiments can be realized.

[0056] The above uses specific examples to illustrate the present invention, which is only for helping to understand the present invention and is not used to limit the present invention. For those skilled in the art of the present invention, according to the idea of the present invention, several simple deductions, deformations or substitutions can also be made.

Claims

1. An improved particle filter map matching method based on rail transit 5G positioning, characterized in that The method includes: Collecting positioning trajectory data based on 5G positioning, and performing moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data; For each trajectory point in the new positioning trajectory data, using an improved particle filter map matching algorithm model for map matching and position correction, including: a. Particle initialization; b. Calculating importance sampling; c. Particle weight update, and the particle weight update strategy includes: for particles outside the track area, the weight is set to 0; for particles inside the track area, particles within the preset distance range of the track centerline are given the maximum weight, and the remaining particles inside the track area are given equal weights; d. Judging whether resampling is needed according to the number of effective particles, and if so, performing resampling; e. Calculating the posterior state estimate; f. Obtaining the next trajectory point in the positioning trajectory point sequence, and repeating steps b - e until all trajectory points are processed.

2. An improved particle filter map matching method based on rail transit 5G positioning according to claim 1, characterized in that, Collecting positioning trajectory data based on 5G positioning specifically includes: Deploying multiple 5G base stations in the track area, and the terminal device set on the vehicle obtains positioning trajectory data by communicating with the surrounding 5G base stations, and transmits the positioning trajectory data to the data acquisition and processing terminal in real time.

3. An improved particle filter map matching method based on rail transit 5G positioning according to claim 1, characterized in that, Performing moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data specifically includes: Set the threshold for the number of positioning trajectory points , if the number of collected positioning trajectory points is , then judge Whether it holds. If it holds, perform weighted average processing on several consecutive positioning points within the preset sliding window to generate a smooth new trajectory point sequence.

4. An improved particle filter map matching method based on rail transit 5G positioning according to claim 1, characterized in that, Particle initialization specifically includes: Let the moment , according to the distribution of the prior probability density function, generate particles and form an initial particle set . The initial weight of each particle is assigned as , i being the initial state estimate of the 5. An improved particle filter map matching method based on rail transit 5G positioning according to claim 4, characterized in that, Calculating importance sampling specifically includes: The selected sampling function is a probability density function with the same distribution as the prior probability density function, and after sampling, the particle importance weights are calculated and normalized.

6. An improved particle filter map matching method based on rail transit 5G positioning as claimed in claim 1, characterized in that, Particle weight update specifically includes: Determining the track area boundary and the track centerline; Establishing a track area determination function and a track centerline determination function according to the particle weight update strategy; Update the weights of the particles according to the orbital region determination function and the orbital centerline determination function, and normalize the updated weights to obtain a new set of particles and the corresponding normalized weights, denote the state vector of the k -th particle at time i , N is the total number of current particles; Among them, first, the weight is updated according to the track area determination function: ； In the formula, is the weight of the i -th updated particle at k moment; indicates that the particle is not within the orbit area; Then, the weight is updated according to the track centerline determination function: ； In the formula, is the weight of the i -th updated particle at k moment; means that the particle is within a preset range from the center line of the orbit.

7. An improved particle filter map matching method based on rail transit 5G positioning according to claim 6, characterized in that, Judging whether resampling is needed according to the number of effective particles specifically includes: The number of valid particles is as follows: ； If the following condition is met , resampling is required.

8. An improved particle filter map matching method based on rail transit 5G positioning according to claim 7, characterized in that Resampling specifically includes: From the particle set resample according to the corresponding updated and normalized particle weights to obtain a new particle set , and reassign the weight of each particle to be , N where is the total number of current particles.

9. An improved particle filter map matching method based on rail transit 5G positioning according to claim 8, characterized in that Calculating the posterior state estimate specifically includes: Based on the particle set and the weights obtain the posterior state estimate : ； Among them, represents the state vector of the i -th particle after resampling at k moment, represents the weight of the i -th particle after resampling at k moment, N is the total number of current particles.

10. An improved particle filter map matching system based on rail transit 5G positioning, characterized in that, The system includes: A positioning trajectory acquisition module, which is used to collect positioning trajectory data based on 5G positioning, and perform moving average preprocessing on the collected positioning trajectory data to generate new positioning trajectory data; A map matching module, which is used to perform map matching and position correction on each trajectory point in the new positioning trajectory data by using an improved particle filter map matching algorithm model, including: a. Particle initialization; b. Calculating importance sampling; c. Particle weight update, and the particle weight update strategy includes: for particles outside the track area, the weight is set to 0; for particles inside the track area, particles within the preset distance range of the track centerline are given the maximum weight, and the remaining particles inside the track area are given equal weights; d. Judging whether resampling is needed according to the number of effective particles, and if so, performing resampling; e. Calculating the posterior state estimate; f. Obtaining the next trajectory point in the positioning trajectory point sequence, and repeating steps b - e until all trajectory points are processed.

Images