High-speed illegal parking positioning method based on deep learning and user position big data

Abstract

The invention relates to the field of intelligent positioning, in particular to a high-speed illegal parking positioning method based on deep learning and user position big data, which comprises the following steps: acquiring basic data, fitting base station data with road point data based on an iterative nearest point algorithm to obtain a fitting point set matched with an actual road shape; binding the coverage direction of the base station with the highway pile; extracting illegal parking features, and constructing an illegal parking model through a deep learning algorithm recurrent neural network for training and evaluation; and deploying the trained illegal parking model in practical application, performing real-time highway illegal parking prediction, screening out illegal parking users, positioning base stations where the users are located, and analyzing a binding relationship between the range of the base stations and highway piles to accurately position the positions of the users. According to the invention, real-time data collection and analysis are realized, illegal parking behaviors are quickly found, and high coverage rate is realized with relatively low cost by utilizing the existing position base station network, so that the investment cost of the whole project is remarkably reduced while the positioning efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent positioning, and specifically to a high-speed illegal parking positioning method based on deep learning and user location big data. Background Art

[0002] At present, with the continuous expansion of my country's expressway network and the increasing traffic volume on expressways, the problem of expressway safety management has become increasingly prominent, especially illegal parking and traffic accidents on expressways. Due to the particularity of expressways, once they occur, they often have a serious impact on traffic flow and vehicle safety. Therefore, it is of great significance to quickly locate illegally parked vehicles and accident vehicles on expressways for expressway management.

[0003] At present, my country's highway monitoring system mainly relies on gantry cameras for monitoring, but these cameras have limited coverage and insufficient number, making it difficult to quickly locate illegally parked vehicles and accident vehicles on the highway. In addition, traditional monitoring methods often rely on manual inspections, which are inefficient and cannot be monitored in real time.

[0004] In order to solve this problem, there is an urgent need for a high-speed illegal parking positioning method based on deep learning and user location big data to achieve accurate identification and positioning of high-speed illegal parking behaviors and provide timely information support for rescue personnel. Summary of the invention

[0005] In order to overcome the problems existing in the prior art, the purpose of the present invention is to provide a high-speed illegal parking positioning method based on deep learning and user location big data.

[0006] To achieve the above purpose, the present invention provides the following technical solution: a high-speed illegal parking location method based on deep learning and user location big data, comprising the following steps:

[0007] S1: Collect base station data, highway pile data, highway road point data, highway illegal parking data and highway driving data;

[0008] S2: After removing the interference data, the base station data is fitted with the highway point data based on the iterative closest point algorithm to obtain a fitting point set that matches the actual road shape;

[0009] S3: Bind the base station coverage direction with the highway post, and establish a binding relationship between the base station range and the highway post;

[0010] S4: extracting illegal parking features through the highway illegal parking data, wherein the illegal parking features include base station switching during illegal parking, base station signal strength, distance of road pile switching, vehicle speed, and number of users;

[0011] S5: A parking violation model is constructed by combining base station data, highway driving data, user location data and parking violation features through a deep learning algorithm recurrent neural network for training and evaluation;

[0012] S6: Deploy the trained illegal parking model to actual applications, conduct real-time highway illegal parking prediction, screen out illegal parked users, locate the base station where the user is located, and analyze the binding relationship between the base station range and highway post to accurately locate the user's location.

[0013] The present invention is further configured as follows: the base station data collection is specifically to collect signal data generated by the user's mobile phone when a vehicle passes by through a mobile communication base station deployed on the road; the base station data includes the user number, the base station cell, the base station latitude and longitude, CGI and timestamp.

[0014] The present invention is further configured as follows: the highway road point data collection based on highway road posts is specifically to set up a number of highway road posts along both sides of the highway, the highway road post data includes the longitude and latitude corresponding to each highway road post; the highway road point data includes the longitude and latitude corresponding to each highway road point.

[0015] The highway waypoint data can be obtained through the Internet, for example, through Baidu Maps or Amap.

[0016] The present invention is further configured as follows: the illegal parking data includes a user number, a detection base station, longitude and latitude data, and illegal parking time; and the driving data is obtained by calculating the user's residence time and the moving speed.

[0017] The present invention is further configured as follows: in step S2, the iterative closest point algorithm comprises the following steps:

[0018] S21: using an iterative closest point algorithm to initialize the transformation model between the two sets of feature point sets, the longitude and latitude of the base station and the longitude and latitude of the waypoint;

[0019] S22: Setting a convergence condition to optimize the transformation model, and obtaining a fitting point set by minimizing the distance between two sets of feature point sets;

[0020] S23: Calculate the error between the fitting point in the fitting point set and the actual measurement point. If the error is within a preset range, the fitting is considered successful, and a fitting point set matching the actual road shape is output; if the error exceeds the preset range, reselect the feature point set or adjust the convergence condition of the iterative nearest point algorithm, and iterate again; the convergence condition is a preset maximum number of iterations or a preset transformation model error threshold.

[0021] The present invention is further configured as follows: the convergence condition is a preset maximum number of iterations or a preset transformation model error threshold. When the error of the algorithm result is less than the preset transformation model error threshold or reaches the maximum number of iterations, the algorithm is considered to have converged.

[0022] The present invention is further configured as follows: the interference data in step S2 includes base station data of toll stations, service areas, surrounding residential areas, highway waypoint data, highway illegal parking data and highway driving data.

[0023] The present invention is further configured as follows: Step S3 specifically includes confirming the coordinates of the road stake in combination with the longitude and latitude of the road stake, and comparing the coordinates of the road stake with the coverage range of the base station to determine the corresponding road stake number covered by the base station.

[0024] The coverage of the base station may be calculated based on the base station transmission power, antenna type, elevation angle, and orientation using a free space path loss model or considering the orientation and downtilt angle of the antenna.

[0025] The present invention is further configured as follows: Step S5 specifically utilizes a recurrent neural network (RNN) to construct a high-speed illegal parking model, firstly performs data preprocessing, cleans the data, extracts data features of base station switching frequency, signal strength, vehicle speed and number of users, divides the data set into a training set, a verification set and a test set, and completes the training, adjustment and evaluation of the model by constructing a recurrent neural network model.

[0026] The present invention is further configured as follows: in step S6, the binding relationship between the base station range and the highway post is analyzed to accurately locate the user position. Specifically, when the user is driving on the highway or illegally parked, the mobile phone signal is connected to the nearest base station, and the specific road point range is located through the fitting relationship between the base station data and the highway road point data, and the road post is located through the binding relationship between the base station and the highway post. The road post is associated with the longitude and latitude of the nearest road point through the highway illegal parking model, and the position of the user on the highway can be located.

[0027] The present invention is further configured as follows: the high-speed illegal parking location method further includes:

[0028] S7: Push the screened illegal parking user locations to the traffic platform.

[0029] In summary, the beneficial effects of the above technical solution of the present invention are as follows:

[0030] 1. The present invention avoids the cost of building a large number of high-speed gantry cameras by utilizing the existing location base station network. The location base stations are widely distributed, forming a dense monitoring network, achieving high coverage at a low cost, thereby significantly reducing the investment cost of the overall project while improving positioning efficiency.

[0031] 2. Wide coverage: The positioning system based on the location base station can span a wide area and is not limited by a single gantry position, so it can effectively monitor potential illegal parking behaviors in a larger range. This wide coverage capability is crucial for urban traffic management departments to better monitor traffic flow and parking conditions.

[0032] 3. High-precision positioning: The location base station combined with advanced positioning technology can provide high-precision location data. This high-precision positioning can not only accurately capture the location of the vehicle, but also distinguish the parking state and driving state of the vehicle when necessary, thereby realizing accurate identification of illegal parking on the highway.

[0033] 4. Real-time monitoring and response: After the model is deployed, real-time high-speed illegal parking predictions are made by inputting base station snapshot data in real time. The positioning system based on the location base station can realize real-time data collection and analysis, and quickly detect illegal parking behaviors. Timely monitoring and response can effectively reduce traffic congestion caused by illegal parking, improve road use efficiency, and ensure road traffic safety. Traffic management departments can more effectively deploy law enforcement resources, such as patrol cars and law enforcement personnel, to achieve timely handling of illegal parking behaviors, improve traffic order, and improve the overall traffic operation efficiency of the city. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0035] Figure 1 This is a flow chart of a method for locating illegal high-speed parking based on deep learning and user location big data in an embodiment. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is clearly and completely described below in conjunction with the accompanying drawings of the present invention. Based on the embodiments of the present invention, other similar embodiments obtained by ordinary technicians in the field without making any creative work should all fall within the scope of protection of the present invention.

[0037] The present invention will be further described below in conjunction with the accompanying drawings and preferred embodiments.

[0038] Example:

[0039] like Figure 1 As shown in the figure, a preferred embodiment of the present invention is a high-speed illegal parking location method based on deep learning and user location big data, comprising the following steps:

[0040] S1: Collect base station data, highway pile data, highway road point data, highway illegal parking data and highway driving data;

[0041] Through the mobile communication base stations deployed on the road, the signal data generated by the user's mobile phone when the vehicle passes by is collected; the base station data includes the user number, the base station cell, the base station latitude and longitude, CGI and timestamp.

[0042] The highway road point data collection based on highway road posts is specifically to set up a number of highway road posts along both sides of the highway, and the highway road post data includes the longitude and latitude corresponding to each highway road post; the highway road point data includes the longitude and latitude corresponding to each highway road point.

[0043] The highway waypoint data can be obtained through the Internet, for example, through Baidu Maps or Amap.

[0044] The present invention is further configured as follows: the illegal parking data includes a user number, a detection base station, longitude and latitude data, and illegal parking time; and the driving data is obtained by calculating the user's residence time and the moving speed.

[0045] S2: Eliminate interference data, including base station data of toll stations, service areas, surrounding residential areas, highway road point data, highway illegal parking data, and highway driving data, and fit the base station data with the highway road point data based on the iterative closest point algorithm to obtain a fitting point set that matches the actual road shape;

[0046] S21: using an iterative closest point algorithm (ICP algorithm) to initialize the transformation model between the two sets of feature point sets, the longitude and latitude of the base station and the longitude and latitude of the waypoint;

[0047] The basic steps of the ICP algorithm are as follows:

[0048] Step 1: closest point correspondence;

[0049] For each point p in the source point set P, find the closest point q in the target point set Q:

[0050]

[0051] The source point set P is a base station longitude and latitude data set, the target point set Q is a highway point longitude and latitude data set; q' is any point in the target point set Q.

[0052] Step 2: Calculate the error function;

[0053] Define the error function E(T), which measures the difference between the source point set P and its corresponding point set Q′ under transformation T:

[0054]

[0055] in:

[0056] p i is the i-th point in the source point set P.

[0057] q i It is the closest point in the target point set Q corresponding to pi.

[0058] T is the transformation matrix, which usually includes rotation R and translation t, that is, T = [R|t].

[0059] n is the number of points in the point set P.

[0060] Step 3: Minimize the error function

[0061] Find the optimal transformation matrix T* to minimize the error function E(T):

[0062]

[0063] In practice, T* is usually solved by the following steps:

[0064] 1. Calculate the centroid of the corresponding point set Q′ and the centroid of P

[0065] 2. Calculate the covariance matrix H:

[0066]

[0067] 3. Solve the rotation matrix R and translation vector t through singular value decomposition (SVD).

[0068] Step 4: Update the transformation and iterate;

[0069] Apply the found optimal transformation matrix T* to update the source point set P, and repeat steps 1 to 3 until the convergence condition is met.

[0070] S22: Set convergence conditions to optimize the transformation model, gradually improve the accuracy of the model by minimizing the distance between the two sets of feature point sets, and obtain a set of fitting points; the convergence condition is a preset maximum number of iterations or a preset transformation model error threshold. When the error of the algorithm result is less than the preset transformation model error threshold or reaches the maximum number of iterations, the algorithm is considered to have converged. In this embodiment, the preset maximum number of iterations is 50, and the transformation model error threshold is 1e-2;

[0071] S23: Calculate the error between the fitting point and the actual measurement point. If the error is within the preset range, the fitting is considered successful and a fitting point set matching the actual road shape is output. If the error exceeds the preset range, reselect the feature point set or adjust the convergence condition of the iterative nearest point algorithm and iterate again.

[0072] S3: Bind the base station coverage direction with the highway post to establish a binding relationship between the base station range and the highway post; confirm the post coordinates based on the post latitude and longitude, and compare the post coordinates with the coverage range of the base station to determine the corresponding post number covered by the base station.

[0073] The coverage of the base station may be calculated based on the base station transmission power, antenna type, elevation angle, and orientation using a free space path loss model or considering the orientation and downtilt angle of the antenna.

[0074] S4: Extracting illegal parking features through the highway illegal parking data, wherein the illegal parking features include base station switching during illegal parking, base station signal strength, distance of road post switching, vehicle speed, and number of users.

[0075] S5: A parking violation model is constructed by combining the features of illegal parking in base station data, highway driving data and user location data through a deep learning algorithm recurrent neural network for training and evaluation;

[0076] By collecting base station data, vehicle driving data, and user location data on the highway, a highway illegal parking model is constructed using a recurrent neural network (RNN). First, data preprocessing is performed to extract data related to illegal parking features by cleaning the data, including base station switching frequency, signal strength, vehicle speed, and number of users. The data set is divided into a training set, a validation set, and a test set. The training, adjustment, and evaluation of the model are completed by constructing a recurrent neural network model.

[0077] S6: Deploy the trained illegal parking model to actual applications, conduct real-time highway illegal parking prediction, screen out illegal parking users, locate the base station where the user is located, and analyze the binding relationship between the base station range and the highway road pile to accurately locate the user's location. When the user is driving on the highway or illegally parking, the mobile phone signal connects to the nearest base station. Through the fitting relationship between the base station data and the highway road point data, the specific road point range is located, and then the road pile is located through the binding relationship between the base station and the highway road pile. The highway illegal parking model is used to associate the road pile with the longitude and latitude of the nearest road point, and the user's highway location can be located.

[0078] Because the base station coverage area is relatively large and the coverage directions are different, the user's travel may deviate based on the base station positioning. By binding the relationship between the base station and the road post, the user can be quickly located at which road post.

[0079] S7: Push the location of the users who have been screened out for illegal parking to the traffic platform. The traffic platform will further process the situation, such as issuing illegal parking notices and arranging law enforcement personnel for on-site processing. The real-time and stability of data push must be ensured to ensure the effectiveness of traffic management.

[0080] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A high-speed illegal parking location method based on deep learning and user location big data, characterized in that: The following steps are involved: S1: Collect base station data, highway pile data, highway road point data, highway illegal parking data and highway driving data; S2: After removing the interference data, the base station data is fitted with the highway point data based on the iterative closest point algorithm to obtain a fitting point set that matches the actual road shape; S3: Bind the base station coverage direction with the highway post, and establish a binding relationship between the base station range and the highway post; S4: extracting illegal parking features through the highway illegal parking data, wherein the illegal parking features include base station switching during illegal parking, base station signal strength, distance of road pile switching, vehicle speed, and number of users; S5: A parking violation model is constructed by combining base station data, highway driving data, user location data and parking violation features through a deep learning algorithm recurrent neural network for training and evaluation; S6: Deploy the trained illegal parking model to actual applications, conduct real-time highway illegal parking prediction, screen out illegal parked users, locate the base station where the user is located, and analyze the binding relationship between the base station range and highway post to accurately locate the user's location.

2. According to claim 1, a high-speed illegal parking location method based on deep learning and user location big data is characterized in that: The base station data collection is specifically to collect signal data generated by the user's mobile phone when a vehicle passes by through the mobile communication base stations deployed on the road; the base station data includes the user number, the base station cell, the base station latitude and longitude, CGI and timestamp.

3. According to claim 2, a high-speed illegal parking location method based on deep learning and user location big data is characterized in that: A number of highway posts are arranged along both sides of the highway, and the highway post data includes the longitude and latitude corresponding to each highway post; and the highway waypoint data includes the longitude and latitude corresponding to each highway waypoint.

4. According to claim 3, a high-speed illegal parking location method based on deep learning and user location big data is characterized in that: The illegal parking data includes user number, detection base station, longitude and latitude data and illegal parking time; the driving data is obtained by calculating the user's residence time and moving speed.

5. According to claim 4, a high-speed illegal parking location method based on deep learning and user location big data is characterized in that: In step S2, the fitting process includes the following steps: S21: using an iterative closest point algorithm to initialize the transformation model between the two sets of feature point sets, the longitude and latitude of the base station and the longitude and latitude of the waypoint; S22: Setting a convergence condition to optimize the transformation model, and obtaining a fitting point set by minimizing the distance between two sets of feature point sets; S23: Calculate the error between the fitting points in the fitting point set and the actual measurement points. If the error is within a preset range, the fitting is considered successful, and a fitting point set matching the actual road shape is output; If the error exceeds a preset range, the feature point set is reselected or the convergence condition of the iterative closest point algorithm is adjusted, and iteration is performed again; the convergence condition is a preset maximum number of iterations or a preset transformation model error threshold.

6. The high-speed illegal parking location method based on deep learning and user location big data according to claim 1 is characterized in that: The interference data in step S2 includes base station data of toll stations, service areas, surrounding residential areas, highway waypoint data, highway illegal parking data and highway driving data.

7. The high-speed illegal parking location method based on deep learning and user location big data according to claim 5 is characterized in that: Step S3 specifically includes confirming the coordinates of the road post in combination with the longitude and latitude of the road post, and comparing the coordinates of the road post with the coverage range of the base station to determine the corresponding road post number covered by the base station.

8. The high-speed illegal parking location method based on deep learning and user location big data according to claim 7 is characterized in that: Specifically, step S5 uses a recurrent neural network to build a high-speed illegal parking model, extracts data features such as base station switching frequency, signal strength, vehicle speed, and number of users, divides the data set into a training set, a validation set, and a test set, and completes the training, adjustment, and evaluation of the model by building a recurrent neural network model.

9. The high-speed illegal parking location method based on deep learning and user location big data according to claim 8 is characterized in that: In step S6, the binding relationship between the base station range and the highway post is analyzed to accurately locate the user's position. Specifically, when the user is driving on the highway or illegally parked, the mobile phone signal connects to the nearest base station, and the specific road point range is located through the fitting relationship between the base station data and the highway road point data. The road post is then located through the binding relationship between the base station and the highway post. The road post is associated with the longitude and latitude of the nearest road point through the highway illegal parking model, and the user's position on the highway can be located.

10. The high-speed illegal parking location method based on deep learning and user location big data according to claim 9 is characterized in that: The high-speed illegal parking location method also includes: S7: Push the screened illegal parking user locations to the traffic platform.

Images