A mobile monitoring carrier networking optimization method

Abstract

The application discloses a mobile monitoring carrier networking optimization method, and belongs to the technical field of mobile computing-based monitoring carrier networking optimization. The method solves the problem of insufficient fixed monitoring network coverage in the traditional mobile monitoring networking method in the prior art. The method forms a vehicle operation track based on a map matching algorithm and a GPS observation sequence, determines a vehicle operation space-time distribution rule, and realizes vehicle track space matching. The method analyzes the spatial regularity of all vehicle operation tracks by using a clustering algorithm, obtains a clustering result, and realizes track identification. Based on the clustering result, a stability index is set, a typical track is defined, and vehicle fleet filtering is realized. The track overlap degree between different vehicles is quantitatively obtained, a mobile detection carrier selection model is constructed in combination with the networking target of a mobile monitoring system, and networking vehicle fleet selection is realized. The application effectively improves the fixed monitoring network coverage, reduces the cost, and can be applied to mobile monitoring system networking optimization.

Description

TECHNICAL FIELD

[0001] The present application relates to a mobile monitoring carrier networking optimization method, belonging to the technical field of mobile computing-based monitoring carrier networking optimization. BACKGROUND

[0002] City digital construction is highly dependent on real-time and comprehensive perception of city dynamic activity information (such as traffic flow, environmental parameters, crowd distribution, event occurrence, etc.). With the rapid development of intelligent networking and IOT technology, mobile intelligent vehicles such as intelligent networked vehicles are gradually popularized. However, the traditional mobile monitoring carrier networking method relies on the perception network constructed by fixed monitoring facilities (such as cameras, sensor stations, and environmental monitoring points), which faces the following challenges: (1) High investment in city fixed monitoring network facility construction: deploying fixed monitoring points with sufficient coverage and density requires huge upfront investment and ongoing operation and maintenance costs; (2) Insufficient spatial and temporal distribution coverage: fixed devices are limited to their physical location and can only monitor a limited area around them, resulting in spatial and temporal blind spots, and their lack of flexibility makes it difficult to dynamically adjust the monitoring focus area according to city activity hotspots or emergency needs.

[0003] In view of the above, there is a need for a mobile monitoring carrier networking optimization method. SUMMARY

[0004] In the following, a brief summary of the present application is given in order to provide a basic understanding of some aspects of the present application. It should be understood that this summary is not a comprehensive summary of the present application. It is not intended to determine the key or important parts of the present application, nor to limit the scope of the present application. Its purpose is only to give some concepts in a simplified form as a prelude to the more detailed description discussed later.

[0005] In view of the above, in order to solve the problem of insufficient coverage of the traditional mobile monitoring networking method in the prior art, the present application provides a mobile monitoring carrier networking optimization method.

[0006] The technical solution is as follows: a mobile monitoring carrier networking optimization method, comprising the following steps:

[0007] S1. Forming a vehicle running track based on a map matching algorithm and a GPS observation sequence, determining the spatio-temporal distribution rule of vehicle running, and realizing spatial matching of vehicle track;

[0008] S2. Analyzing the spatial regularity of all vehicle running tracks using a clustering algorithm to obtain clustering results and realize track recognition;

[0009] S3. Based on the clustering results, setting a stability index and defining a typical track to realize vehicle fleet filtering;

[0010] S4. Based on the vehicle trajectory space matching and trajectory recognition, the trajectory overlap degree between different vehicles is quantified, a mobile detection carrier selection model is constructed in combination with the networking target of the mobile monitoring system, a typical trajectory is input, a vehicle ID list is obtained through screening, and the selection of a networked vehicle fleet is realized.

[0011] Further, in S1, the following steps are included:

[0012] S11. Screening the trajectory in the GPS observation sequence, performing trajectory matching, and forming a vehicle running trajectory;

[0013] In S11, the information of the vehicle GPS points in the research area is collected, the hidden Markov chain hmm map matching algorithm is used, the GPS points are matched to the running road section according to the longitude and latitude information, a plurality of GPS point sequences are matched to a plurality of road section sequences, and the running trajectory of the vehicle is formed, which specifically includes the following steps:

[0014] S111. According to the GPS longitude and latitude information, the projection distance d1 of each GPS point vehicle distance road section g is calculated, the projection distance d1 is input into the observation probability model, and the observation probability of the GPS point falling on the road section g is calculated .

[0015] S112. According to the observation probability of each point, the length d2 of the projection point of the current GPS point on the road section g to the projection point of the next GPS on the road section g is calculated, and the state transition probability of the GPS point from the road section g to the road section g+1 is calculated by using the state transition probability model .

[0016] S113. The maximum possible path is solved based on the Viterbi algorithm, that is, the maximum possible probability of each GPS point falling on a road section is calculated according to the observation probability and the state transition probability of the GPS points.

[0017] In S113, the maximum possible probability is represented as:

[0018]

[0019] Wherein, is the projection road section state space of the current GPS point, is the maximum possible probability of the current GPS point from the projection road section g to the road section g+1, is the observation probability of the next GPS point falling on the road section g+1.

[0020] S114. Determine the maximum possible projection road segments of the GPS points according to the maximum possible probability, connect the maximum possible projection road segments of each GPS point, and the road segment numbers form the vehicle operation trajectory k, i.e. a sequence composed of multiple road segments g;

[0021] S12. Combine the GIS vector information of the vehicle operation trajectory and the GIS vector information of the road network of the study area, use GIS tools to perform spatial vector intersection operation on the marked road segment g in the vehicle operation trajectory k and the study space j, mark the affiliation of the GPS point and the road segment g, the affiliation of the road segment g and the vehicle operation trajectory k, and the affiliation of the vehicle operation trajectory k and the study area j, and realize the spatial matching of the vehicle trajectory.

[0022] Further, in the S2, the following steps are included:

[0023] S21. Define the clustering distance threshold as , the minimum number of neighbors as , and the cluster number as ;

[0024] S22. Based on the clustering definition of step S21, calculate the spatio-temporal distance between the vehicle operation trajectory k and the center trajectory k+1, and construct a trajectory similarity matrix;

[0025] In the S22, the trajectory similarity matrix is represented as:

[0026]

[0027] , wherein is the time similarity of the vehicle operation trajectory k and the vehicle operation trajectory k+1, is the spatial similarity of the vehicle operation trajectory k and the vehicle operation trajectory k+1, is the spatio-temporal joint similarity of the vehicle operation trajectory k and the vehicle operation trajectory k+1, is the time length of the vehicle operation trajectory k, is the spatial length of the vehicle operation trajectory k, is the time overlap part of the vehicle operation trajectory k and the vehicle operation trajectory k+1, is the spatial overlap part of the vehicle operation trajectory k and the vehicle operation trajectory k+1, is the spatio-temporal joint distance of the vehicle operation trajectory k and the vehicle operation trajectory k+1;

[0028] S23. Based on the spatio-temporal joint similarity of the trajectory similarity matrix, define the trajectory distance matrix D composed of the distance between the vehicle operation trajectory k and the vehicle operation trajectory k+1 , and determine the domain of the trajectory k;

[0029] The distance between the vehicle running trajectory k and the vehicle running trajectory k+1 in the S23 is expressed as:

[0030]

[0031] The field of the trajectory k is expressed as:

[0032]

[0033] wherein, the other trajectories satisfying the distance less than to the trajectory k are the set of trajectories reachable by the density of the trajectory k.

[0034] S24. The number of points in the field of the vehicle running trajectory k is calculated, if the number of points in the field of the vehicle running trajectory k is greater than the minimum neighbor number, that is, , the core point is marked, otherwise the noise point is marked, and the above step S24 is repeated until all points are traversed.

[0035] S25. The entire trajectory dataset composed of the vehicle running trajectory k of the vehicle i is traversed when an unvisited core point is found, the current core point is taken as the starting point, the expansion cluster operation is called, and the trajectory recognition is completed.

[0036] In the S25, the expansion cluster operation is specifically as follows:

[0037] Let wherein, the current cluster is denoted, and the vehicle running trajectory k is marked as visited.

[0038] The expansion queue of the vehicle running trajectory k is initialized the following judgment is performed on each neighbor in the expansion queue seed:

[0039] If the current neighbor is not marked, the field of the current neighbor is calculated, and it is judged whether the current neighbor is a core point.

[0040] If the current neighbor is a core point, the field of the current neighbor is added to the expansion queue seed, and the points not appearing in the expansion queue seed are added after being removed from the expansion queue seed until all the fields in the expansion queue seed are traversed and checked.

[0041] All k-density reachable trajectories, that is, , are classified into the current cluster until all trajectories are visited and classified or marked as noise points.

[0042] Further, in S3, after clustering the vehicle trajectory, the trajectory features of the vehicle will be divided into different clusters, wherein the cluster with the most number of samples is called the main cluster, and a stability index is set to preferentially select vehicles with high stability index, and the first 20% of the vehicle trajectories closest to the clustering center of the main cluster are defined as typical trajectories representing the normal operation path of the representative vehicle;

[0043] The stability index is represented as:

[0044]

[0045] wherein, is the number of clustering trajectories of vehicle i in the main cluster, is the number of trajectories of vehicle i, is the number of all trajectory clusters of vehicle i.

[0046] Further, in S4, the following steps are included:

[0047] S41. Based on the divided monitoring area, the key area and key section are determined by calculating the comprehensive score of the area;

[0048] In S41, the monitoring area is divided into N2 blocks, and based on the GDP, population size p, land use type l, and total mileage m of expressways and urban trunk roads, the key degree of the block is scored, and the comprehensive score of the area is calculated ;

[0049] The comprehensive score of the area is represented as:

[0050]

[0051] wherein, is the block number, are the weight coefficients of the four evaluation indexes, respectively;

[0052] The comprehensive score of the area is taken as the key area j', and a key area matrix is constructed, and according to the road grade and flow, the key section g' is selected in the key area;

[0053] S42. According to the vehicle trajectory space matching result, a vehicle-area matrix and a vehicle-section matrix are constructed to obtain the spatial correlation degree between the vehicle and the block;

[0054] In S42, according to the spatial subordination relationship between each section g of the vehicle trajectory k and the block j, the number of times that the vehicle passes through different key sections and the number of times that the vehicle passes through different key areas , the active frequency of vehicle i between key road segment g' and key region j' is obtained;

[0055] The active frequency of vehicle i between key road segment g' and key region j' is represented as:

[0056]

[0057] The number of vehicle passes of different candidate vehicles through different key regions is integrated , and is assembled into a vehicle-region matrix The number of vehicle passes of different candidate vehicles through different key road segments is integrated , and is assembled into a vehicle-road segment matrix ;

[0058] S43. Based on the trajectory similarity matrix, the trajectory overlap degree between different vehicles is quantified;

[0059] In the S43, according to the spatial similarity of covered road segments and regions between different vehicles in the same time period, the monitoring overlap degree between different vehicles is quantified, the redundancy degree of monitoring resources in the same spatio-temporal distribution is quantified, and the trajectory overlap degree between different vehicles is obtained.

[0060] The quantification process of the redundancy degree is represented as:

[0061]

[0062] , wherein represents the covered road segment set of vehicle i, which is composed of the number of vehicle passes of vehicle i through different key road segments , represents the covered region set of vehicle i, which is composed of the number of vehicle passes of vehicle i through different key regions , represents the overlap degree of covered key regions and key road segments between vehicle i and vehicle i+1, i.e. the trajectory overlap degree between different vehicles;

[0063] S44. According to the trajectory overlap degree between different vehicles and the three calibration parameters of the networking target of the mobile monitoring system, a mobile detection carrier selection model is established, the model input is the typical trajectory of N candidate vehicles, the model output is the vehicle ID list screened, and the networking optimization is completed;

[0064] In the S44, the networking target of the mobile monitoring system is to monitor the least investment, the most key region coverage, and the least monitoring resource redundancy, the least investment means the least deployment of equipment under the premise of meeting the regional coverage requirement, and the number of selected vehicle fleets is obtained ;

[0065]

[0066] wherein, selecting the model for the mobile detection carrier, denotes whether to select vehicle i, i = 1 indicates to select the current vehicle networking, i = 0 indicates not to select the current vehicle networking;

[0067] The key area is covered most, and the typical trajectory of the selected vehicle group is required to be evenly distributed in each key area, and the typical trajectory between different vehicles has time and space complementarity, and the sum of the typical trajectory coverage areas of the selected vehicle group is obtained ;

[0068] The sum of the typical trajectory coverage areas of the selected vehicle group is represented as:

[0069]

[0070] wherein, respectively, the number of vehicles, the number of key areas, and the number of key road sections;

[0071] The monitoring resource redundancy is minimum, that is, the typical trajectory of the selected vehicle group does not coincide, and the sum of the two-two overlap degrees between different vehicles of the selected vehicle group is obtained ;

[0072]

[0073] wherein, denotes the overlap degree of the key area and the key road section covered between vehicle i and vehicle k, denotes whether to select vehicle k, i = 1 indicates to select the current vehicle networking, i = 0 indicates not to select the current vehicle networking;

[0074] According to the number of the selected vehicle group , the sum of the typical trajectory coverage areas of the selected vehicle group , and the sum of the two-two overlap degrees between different vehicles of the selected vehicle group , a mobile detection carrier selection model is established, an enumeration algorithm is used for solving, and a mathematical expression of the mobile detection carrier selection model is obtained;

[0075] The mathematical expression of the mobile detection carrier selection model is represented as:

[0076]

[0077] wherein, is a maximum function, is a networking target weight coefficient.

[0078] The beneficial effects of the present application are as follows: the mobile monitoring carrier networking optimization method can realize typical trajectory identification based on map matching and density clustering fusion, use the cluster main cluster and the dispersion information of the cluster to express the vehicle operation law, and screen the typical vehicle operation trajectory; the correlation degree model of the vehicle and the space and the vehicle and the vehicle is established, the minimum cost mobile monitoring vehicle team networking under the key space and road coverage constraint is provided, and the limitation of insufficient coverage of the fixed monitoring network is made up; the mobile monitoring system networking optimization with small cost, large coverage rate and high stability is realized, and the cost investment and monitoring coverage target of the mobile monitoring system are balanced. BRIEF DESCRIPTION OF DRAWINGS

[0079] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0080] Figure 1 It is a flowchart of a mobile monitoring carrier networking optimization method. DETAILED DESCRIPTION

[0081] In order to make the technical solutions and advantages in the embodiments of the present application clearer, the exemplary embodiments of the present application are further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and are not an exhaustive enumeration of all embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0082] REFERENCE Figure 1 A mobile monitoring carrier networking optimization method is described in detail in the embodiment, which specifically includes the following steps:

[0083] S1. Based on the map matching algorithm and the GPS observation sequence, the vehicle operation trajectory is formed, the vehicle operation spatiotemporal distribution law is determined, the vehicle trajectory space matching is realized;

[0084] S2. The spatial regularity of all vehicle operation trajectories is analyzed by using the clustering algorithm, the clustering result is obtained, and the trajectory identification is realized;

[0085] S3. Based on the clustering result, the stability index is set, the typical trajectory is defined, and the vehicle team filtering is realized;

[0086] S4. Based on the vehicle trajectory space matching and the trajectory identification, the trajectory overlap degree between different vehicles is quantitatively obtained, the mobile detection carrier selection model is constructed in combination with the networking target of the mobile monitoring system, the typical trajectory is input, the vehicle ID list screened is obtained, and the networking vehicle team selection is realized.

[0087] Further, the S1 includes the following steps:

[0088] S11. Screen the trajectory in the GPS observation sequence, perform trajectory matching, and form the vehicle running trajectory;

[0089] In the S11, the GPS observation sequence is a time sequence data composed of multiple GPS longitude and latitude, the information of the vehicle GPS point in the research area is collected, the hmm map matching algorithm is used, the GPS point is matched to the running road section according to the longitude and latitude information of the GPS point, multiple GPS point sequences are matched to multiple road section sequences, and the running trajectory of the vehicle is formed, which specifically includes the following steps:

[0090] S111. According to the GPS longitude and latitude information, the projection distance d1 of each GPS point vehicle from the road section g is calculated, the projection distance d1 is input into the observation probability model, and the observation probability of the GPS point falling on the road section g is calculated ;

[0091] S112. According to the observation probability of each point , the length d2 of the projection point of the current GPS point on the road section g to the projection point of the next GPS on the road section g is calculated, the state transition probability model is used, the state transition probability model is a matrix composed of the projection length d2 between each point in the sequence, and the state transition probability of the GPS point from the road section g to the road section g+1 is calculated ;

[0092] S113. The maximum possible path is solved based on the Viterbi algorithm, that is, according to the observation probability of each GPS point and the state transition probability , the maximum possible probability of each GPS point falling on a road section is calculated;

[0093] In the S113, the maximum possible probability is represented as:

[0094]

[0095] Among them, is the projection road section state space of the current GPS point, is the maximum possible probability of the current GPS point from the projection road section g to the road section g+1, is the observation probability of the next GPS point falling on the road section g+1;

[0096] S114. According to the maximum possible probability, the maximum possible projection road section of the GPS point is determined, the maximum possible projection road section of each GPS point is connected, and the road section number forms the vehicle running trajectory k, that is, a sequence composed of multiple road sections g;

[0097] S12. Combine the GIS vector information of the corresponding research area of the vehicle running trajectory with the GIS vector information of the road network, use GIS tools to perform spatial vector intersection operation on the marked road segment g in the vehicle running trajectory k and the research space j, mark the affiliation of the GPS point and the road segment g, the affiliation of the road segment g and the vehicle running trajectory k, and the affiliation of the vehicle running trajectory k and the research area j, and realize spatial matching of the vehicle trajectory.

[0098] Specifically, the mobile monitoring system aims to achieve maximum coverage, highest resolution, and most stable monitoring effect with minimum monitoring cost investment. The detection vehicle needs to have the characteristics of large coverage, high frequency, and regular running trajectory, and a model needs to be established to screen typical running routes that can represent the vehicle running regularity. The input of the typical trajectory identification model is the vehicle GPS positioning sequence, and the output is the vehicle;

[0099] In this embodiment, the research area is selected as follows: in the same quarter, regardless of special dates such as holidays, the vehicle running regularity usually does not change much, so the vehicle running trajectory of 3 months (one quarter) is taken as the research sample to screen the typical trajectory that can represent the vehicle running regularity. Before screening the typical trajectory, based on the GPS positioning sequence data, the total time of the sequence in which the vehicle running speed is less than 10 km / h and less than 5 km / h in 3 months is counted to filter low travel rate vehicles.

[0100] Further, in the S2, the following steps are included:

[0101] S21. Define the clustering distance threshold as , the minimum number of neighbors as , and the cluster number as ;

[0102] S22. Based on the clustering definition of step S21, calculate the spatiotemporal distance between the vehicle running trajectory k and the center trajectory k+1, and construct a trajectory similarity matrix;

[0103] In the S22, the trajectory similarity matrix is represented as:

[0104]

[0105] wherein, is the time similarity of the vehicle running trajectory k and the vehicle running trajectory k+1, is the spatial similarity of the vehicle running trajectory k and the vehicle running trajectory k+1, is the spatiotemporal joint similarity of the vehicle running trajectory k and the vehicle running trajectory k+1, is the time length of the vehicle running trajectory k, the spatial length of the vehicle running trajectory k, the time overlap part of the vehicle running trajectory k and the vehicle running trajectory k+1, the spatial overlap part of the vehicle running trajectory k and the vehicle running trajectory k+1, the spatio-temporal joint distance of the vehicle running trajectory k and the vehicle running trajectory k+1;

[0106] S23. Based on the spatio-temporal joint similarity of the trajectory similarity matrix, defining the trajectory distance matrix D composed of the distance of the vehicle running trajectory k and the vehicle running trajectory k+1 , and determining the field of the trajectory k;

[0107] In the S23, the distance of the vehicle running trajectory k and the vehicle running trajectory k+1 is expressed as:

[0108]

[0109] The field of the trajectory k is expressed as:

[0110]

[0111] Wherein, The set of other trajectories satisfying the distance less than to the trajectory K, that is, the set of trajectories density reachable to the trajectory k;

[0112] S24. Calculating the number of points in the field of the vehicle running trajectory k, if the number of points in the field of the vehicle running trajectory k is greater than the minimum neighbor number, that is, , it is marked as a core point, otherwise it is marked as a noise point, and the above step S24 is repeated until all points are traversed;

[0113] S25. Traversing the entire trajectory data set of the vehicle i composed of the vehicle running trajectory k , when an unvisited core point is found, taking the current core point as the starting point, calling the expansion cluster operation to complete the trajectory recognition;

[0114] In the S25, the expansion cluster operation is specifically as follows:

[0115] Let , wherein, The current cluster is denoted as C, and the vehicle running trajectory k is marked as visited;

[0116] Initializing the expansion queue of the vehicle running trajectory k , executing the following judgment on each neighbor in the expansion queue seed:

[0117] If the current neighbor is not marked, the domain of the current neighbor is calculated, and it is determined whether the current neighbor is a core point;

[0118] If the current neighbor is a core point, the current neighborhood is added to the expansion queue seed. To avoid duplication, add the points that do not appear in the expansion queue seed after deduplication, until all neighborhoods in the expansion queue seed are checked;

[0119] All trajectories reachable by k-density are classified into the current cluster , until all trajectories are accessed and classified or marked as noise points.

[0120] Further, in S3, after clustering the vehicle operation trajectory, the trajectory features of the vehicle are divided into different clusters, wherein the cluster with the most clustering samples is called the main cluster, a stability index is set, and vehicles with high stability index are preferentially selected, and the first 20% of the vehicle operation trajectories closest to the clustering center of the main cluster are defined as typical trajectories representing the normal operation path of the vehicle;

[0121] The stability index is represented as:

[0122]

[0123] wherein, is the number of clustering trajectories of the main cluster of vehicle i, is the number of trajectories of vehicle i, is the number of clusters of all trajectories of vehicle i.

[0124] Specifically, the number of clustering samples in the same cluster represents the repetition degree of the historical operation trajectory of the vehicle, and the number of clusters represents the dispersion degree of the operation trajectory of the vehicle. To ensure the stability of the mobile monitoring system, vehicles with dispersed and irregular operation ranges should be avoided as mobile monitoring vehicles. The greater the number of main cluster trajectories and the fewer the number of clusters, the more regular the operation features of the vehicle, and the more suitable the vehicle is as a candidate vehicle. Therefore, a stability index is set for outlier vehicles.

[0125] Further, in S4, the following steps are included:

[0126] S41. Based on the divided monitoring area, the key area and key section are determined by the calculated regional comprehensive score;

[0127] In S41, the monitoring area is divided into N2 blocks, and based on the evaluation indexes of GDP, population size p, land use type l, and total mileage of expressways and urban trunk roads m, the key degree of the block is scored, and the regional comprehensive score is calculated ;

[0128] The regional comprehensive score​ is represented as:

[0129]

[0130] wherein, is the block number, are the weight coefficients of the four evaluation indexes, respectively;

[0131] The regional comprehensive score is taken The top 50% of the regions are taken as key regions j', a key region matrix is constructed, and key road segments g' are selected in the key regions according to road grades, traffic flows, etc.

[0132] S42. According to the vehicle trajectory space matching result, a vehicle-region matrix and a vehicle-road segment matrix are constructed, and a spatial correlation degree between the vehicle and the block is obtained.

[0133] In the S42, according to the spatial subordination relationship between each road segment g of the vehicle running trajectory k and the block j, the number of times that the vehicle passes through different key road segments in different time periods is counted and the number of times that the vehicle passes through different key regions , so as to obtain the active frequency (for example, the number of times passed through per day) of the vehicle i between the key road segments g' and the key regions j';

[0134] The active frequency of the vehicle i between the key road segments g' and the key regions j' is represented as:

[0135]

[0136] The number of times that the vehicle passes through different key regions of different candidate vehicles is integrated to form a vehicle-region matrix The number of times that the vehicle passes through different key road segments of different candidate vehicles is integrated to form a vehicle-road segment matrix ;

[0137] S43. Based on the trajectory similarity matrix, the trajectory overlap degree between different vehicles is quantified;

[0138] In the S43, according to the spatial similarity of the covered road segments and regions between different vehicles in the same time period, the monitoring overlap degree between different vehicles is quantified, the redundancy degree of the monitoring resources in the same space-time distribution is quantified, and the trajectory overlap degree between different vehicles is obtained.

[0139] The quantification process of the redundancy degree is represented as:

[0140]

[0141] wherein, denotes the set of covered road segments of vehicle i, which is the number of times the vehicle passes through different key road segments comprises, denotes the set of covered areas of vehicle i, which is the number of times the vehicle passes through different key areas comprises, denotes the degree of overlap of the covered key areas, key road segments between vehicle i and vehicle i+1, i.e., the trajectory overlap degree between different vehicles;

[0142] S44. According to the trajectory overlap degree between different vehicles and the three calibration parameters of the networking target of the mobile monitoring system, a mobile detection carrier selection model is established, the model input is the typical trajectory of N candidate vehicles, the model output is a vehicle ID list screened, and the networking optimization is completed;

[0143] In the S44, the networking target of the mobile monitoring system is to monitor the least facility investment, the most key area coverage, and the least monitoring resource redundancy, the least investment means the least deployment of equipment on the premise of meeting the regional coverage requirement, and the number of selected vehicle groups is obtained ;

[0144]

[0145] wherein, is the independent variable of the mobile detection carrier selection model, denotes whether to select vehicle i, i=1 denotes to select the current vehicle for networking, and i=0 denotes not to select the current vehicle for networking;

[0146] The most key area coverage requires that the typical trajectories of the selected vehicle groups are evenly distributed in each key area as much as possible, and the typical trajectories between different vehicles have good time and space complementarity, and the sum of the typical trajectory coverage areas of the selected vehicle groups is obtained ;

[0147] The sum of the typical trajectory coverage areas of the selected vehicle groups is represented as:

[0148]

[0149] wherein, are the number of vehicles, the number of key areas, and the number of key road segments, respectively;

[0150] The least monitoring resource redundancy means that the typical trajectories of the selected vehicle groups do not overlap as much as possible, avoiding selecting the same path for repeated monitoring and reducing the waste of monitoring resources, and the sum of the two-by-two overlap degrees between different vehicles of the selected vehicle groups is obtained ;

[0151]

[0152] wherein, represents the degree of overlap between vehicle i and vehicle k covering the critical area, critical road section, represents whether to select vehicle k, i=1 represents selecting the current vehicle networking, i=0 represents not selecting the current vehicle networking;

[0153] According to the number of selected vehicle groups , the sum of the typical trajectory coverage areas of the selected vehicle groups , and the sum of the degrees of overlap between different vehicles of the selected vehicle groups , a mobile detection carrier selection model is established, an enumeration algorithm is used to solve, and a mathematical expression of the mobile detection carrier selection model is obtained;

[0154] The mathematical expression of the mobile detection carrier selection model is represented as:

[0155]

[0156] wherein, is a maximization function, is a networking target weight coefficient.

[0157] Specifically, the goal of the mobile monitoring system networking optimization is to complete the maximum range of monitoring area coverage with the minimum detection investment, to realize the minimum monitoring redundancy, to select x vehicles from the candidate vehicles to add monitoring equipment, and to require as many as possible to cover the key areas and critical road sections, so that the problem is converted into an integer linear programming problem.

[0158] Although the present application is described in terms of a limited number of embodiments, those skilled in the art, with the benefit of the above description, will appreciate that other embodiments are possible within the scope of the application described herein. In addition, it should be noted that the language used in the specification is chosen primarily for readability and instructional purposes and can not have been selected to convey the most technical meaning. Accordingly, numerous modifications and variations are possible in light of the above teachings without departing from the scope and spirit of the appended claims. The disclosure of the application is illustrative only and not limiting of the scope of the application, which is defined by the appended claims.

Claims

1. A mobile monitoring carrier networking optimization method, characterized in that, The method comprises the following steps: S1. Forming a vehicle running track based on a map matching algorithm and a GPS observation sequence, determining a vehicle running space-time distribution rule, and realizing vehicle track space matching; S2. Analyzing the spatial regularity of all vehicle running tracks by using a clustering algorithm to obtain a clustering result and realize track identification; S3. Based on the clustering result, setting a stability index, defining a typical track, and realizing vehicle fleet filtering; S4. Based on vehicle track space matching and track identification, quantitatively obtaining the track overlap degree between different vehicles, combining the networking target of the mobile monitoring system, constructing a mobile detection carrier selection model, inputting the typical track, obtaining a vehicle ID list screened, and realizing networking fleet selection; In the S2, the following steps are included: S21. define a clustering distance threshold as , a minimum number of neighbors as , a cluster number as ; S22. Based on the clustering definition of step S21, calculating the space-time distance between the vehicle running track k and the center track k+1, and constructing a track similarity matrix; S23. The spatio-temporal joint similarity based on the trajectory similarity matrix, defines the distance of the vehicle running trajectory k and the vehicle running trajectory k+1 The trajectory distance matrix D composed of the matrix, and determines the field of the trajectory k; S24. Calculate the number of points within the field of the vehicle running trajectory k, if the number of points in the field of the vehicle running trajectory k is greater than the minimum neighbor number, that is then mark as a core point, otherwise mark as a noise point, repeat the above step S24 until all points are traversed; S25. Traverse the entire trajectory dataset of vehicle i consisting of vehicle run trajectories k When an unvisited core point is found, the expansion cluster operation is invoked with the current core point as the starting point to complete trajectory recognition. In the S4, the following steps are included: S41. Based on the divided monitoring area, determining the key area and key section by calculating the regional comprehensive score; S42. According to the vehicle track space matching result, constructing a vehicle-area matrix and a vehicle-section matrix to obtain the spatial correlation degree between the vehicle and the block; S43. Based on the track similarity matrix, quantitatively obtaining the track overlap degree between different vehicles; S44. According to the track overlap degree between different vehicles and the three calibration parameters of the networking target of the mobile monitoring system, establishing a mobile detection carrier selection model, the model input being the typical track of N candidate vehicles, the model output being a vehicle ID list screened, and completing networking optimization.

2. The mobile monitoring carrier networking optimization method of claim 1, wherein, In the S1, the following steps are included: S11. Screening the track in the GPS observation sequence, performing track matching, and forming a vehicle running track; In the S11, the information of the vehicle GPS point in the research area is collected, the hidden Markov chain hmm map matching algorithm is used, the longitude and latitude information of the GPS point is matched to the running section, a plurality of GPS point sequences are matched to a plurality of section sequences, and the vehicle running track is composed, which specifically comprises the following steps: S111. According to the GPS longitude and latitude information, the projected distance d1 of each GPS point vehicle to the road segment g is calculated, and the projected distance d1 is input into the observation probability model to calculate the observation probability of the GPS point falling on the road segment g ; S112. Based on the observed probabilities of each point The length d2 from the projection point of the current GPS point on road segment g to the projection point of the next GPS point on road segment g is calculated. Using the state transition probability model, the state transition probability of the GPS point moving from road segment g to road segment g+1 is calculated. ; S113. Solve the maximum possible path based on the Viterbi algorithm, i.e. calculate the maximum possible probability of each GPS point falling on a certain road segment according to the observation probability of the GPS points and the state transition probability ​​ In the S113, the maximum probability is represented as: ； wherein, is the projected road segment state space for the current GPS point, is the maximum possible probability of the current GPS point transitioning from projected road segment g to road segment g+1, is the observation probability of the next GPS point falling on road segment g+1; S114. According to the maximum probability, the maximum projection section of the GPS point is determined, the maximum projection section of each GPS point is connected, the section number forms the vehicle running track k, that is, a sequence composed of a plurality of sections g; S12. Combining the GIS vector information of the vehicle running track corresponding to the research area and the GIS vector information of the road network, using GIS tools to perform spatial vector intersection operation on the marked section g in the vehicle running track k and the research space j, marking the affiliation of the GPS point and the section g, the affiliation of the section g and the vehicle running track k, and the affiliation of the vehicle running track k and the research area j, and realizing vehicle track space matching.

3. The mobile monitoring carrier networking optimization method of claim 2, wherein, In the S2, the following steps are included: S21. define a clustering distance threshold as , a minimum number of neighbors as , a cluster number as ; S22. Based on the clustering definition of step S21, calculating the space-time distance between the vehicle running track k and the center track k+1, and constructing a track similarity matrix; In the S22, the track similarity matrix is represented as: ； wherein, is a temporal similarity of the vehicle movement trajectory k and the vehicle movement trajectory k+1, is a spatial similarity of the vehicle movement trajectory k and the vehicle movement trajectory k+1, is a spatio-temporal joint similarity of the vehicle movement trajectory k and the vehicle movement trajectory k+1, is a temporal length of the vehicle movement trajectory k, is a spatial length of the vehicle movement trajectory k, is a temporal overlap of the vehicle movement trajectory k and the vehicle movement trajectory k+1, is a spatial overlap of the vehicle movement trajectory k and the vehicle movement trajectory k+1, is a spatio-temporal joint distance of the vehicle movement trajectory k and the vehicle movement trajectory k+1 ; S23. The spatio-temporal joint similarity based on the trajectory similarity matrix, defines the distance between the vehicle trajectory k and the vehicle trajectory k+1 The trajectory distance matrix D composed of the matrix, and determines the field of trajectory k; In the S23, the distance between the vehicle operation trajectory k and the vehicle operation trajectory k+1 is represented as: ； The field of trajectories k is represented as: ； wherein, represents a set of other trajectories that satisfy a distance to the trajectory K less than i.e. the set of trajectories that are density reachable from the trajectory k; S24. Calculate the number of points within the field of the vehicle running trajectory k, if the number of points in the field of the vehicle running trajectory k is greater than the minimum neighbor number, that is then mark as a core point, otherwise mark as a noise point, repeat the above step S24 until all points are traversed; S25. Traverse the entire trajectory dataset of vehicle i consisting of vehicle run trajectories k When an unvisited core point is found, the expansion cluster operation is invoked with the current core point as the starting point to complete trajectory recognition. The S25, the extended cluster operation is as follows: Let where, denotes the current cluster, mark the vehicle run trajectory k as visited; Initialize an extended queue of vehicle operating trajectories k For each neighbor in the extended queue seed, perform the following judgement: If the current neighbor is not marked, the domain of the current neighbor is calculated, and it is judged whether the current neighbor is a core point; If the current neighbor is a core point, the current neighborhood is added to the expansion queue seed, and after de-duplication, points not appearing in the expansion queue seed are added until all neighborhoods in the expansion queue seed are checked; All trajectories that are k-density reachable, i.e. , are assigned to the current cluster until all trajectories are visited and either assigned to a cluster or labeled as noise points.

4. The mobile monitoring carrier networking optimization method of claim 3, wherein, In the S3, after clustering the vehicle running trajectory, the trajectory features of the vehicle will be divided into different clusters, wherein the cluster with the most clustering samples is called the main cluster, the stability index is set, the vehicle with high stability index is preferentially selected, and the first 20% of the vehicle running trajectory closest to the main cluster center is defined as the typical trajectory representing the normal operation path of the vehicle; The stability index is expressed as: ； wherein, is the number of clustered trajectories for vehicle i, is the number of trajectories for vehicle i, is the number of all trajectory clusters for vehicle i.

5. The mobile monitoring carrier networking optimization method of claim 4, wherein, The S4 includes the following steps: S41. Based on the divided monitoring area, the key area and the key section are determined based on the comprehensive score of the area calculated; In the S41, the monitoring area is divided into N2 blocks, the key degree of the blocks is scored based on the GDP, the population size p, the land use type l, and the total mileage m of the expressways and urban trunk roads, and the regional comprehensive score is calculated ; Regional composite score is represented as: ； wherein, is a block number, are weight coefficients of the four evaluation indexes, respectively. Taking regional comprehensive score The first 50% of the region as a key region j', a key region matrix is constructed, and a key section g' is selected according to the road level and traffic in the key region. S42. According to the vehicle trajectory space matching result, a vehicle-area matrix and a vehicle-section matrix are constructed to obtain the spatial correlation degree between the vehicle and the block; In S42, according to the spatial dependency of each road segment g of the vehicle running trajectory k and the block j, the number of times of passing through different key road segments by the vehicle in different time periods is counted and the number of times of passing through different key areas by the vehicle , to obtain the active frequency of the vehicle i between the key road segment g' and the key area j'. The active frequency of vehicle i between key section g' and key area j' is expressed as: ； Integrating the number of times that vehicles of different candidate vehicles pass through different key areas , into a vehicle-area matrix Integrating the number of times that vehicles of different candidate vehicles pass through different key road segments , into a vehicle-segment matrix ; S43. Based on the trajectory similarity matrix, the trajectory overlap degree between different vehicles is quantified; In the S43, according to the spatial similarity of the covered sections and areas between different vehicles in the same time period, the monitoring overlap degree between different vehicles is quantified, the redundancy degree of the monitoring resources in the same space-time distribution is quantified, and the trajectory overlap degree between different vehicles is obtained; The quantification process of the redundancy degree is expressed as: ； wherein, represents the set of covered road segments of vehicle i, which consists of the number of times the vehicle passes through different key road segments, represents the set of covered areas of vehicle i, which consists of the number of times the vehicle passes through different key areas, represents the degree of overlap of covered key areas, key road segments between vehicle i and vehicle i+1, i.e., the degree of trajectory overlap between different vehicles;​​ S44. According to the trajectory overlap degree between different vehicles and the three calibration parameters of the networking target of the mobile monitoring system, a mobile detection carrier selection model is established, the model input is the typical trajectory of N candidate vehicles, the model output is the vehicle ID list screened, and the networking optimization is completed; In the S44, the networking goal of the mobile monitoring system is to monitor the most key areas with the least facilities, the least monitoring resources, and the least selected vehicle quantity ; ； wherein selecting the model for the mobile detection carrier, denotes whether the vehicle i is selected, i = 1 denotes that the current vehicle is selected for networking, i = 0 denotes that the current vehicle is not selected for networking; The key area coverage is the most, requires the typical trajectory of the selected vehicle fleet to be evenly distributed in each key area, and the typical trajectory between different vehicles has time and space complementarity, and the sum of the typical trajectory coverage area of the selected vehicle fleet is obtained ; the sum of the typical trajectory coverage areas of the selected fleet is represented as: ； wherein, respectively, the number of vehicles, the number of key areas, and the number of key road segments; monitoring resource redundancy minimum, i.e. selected vehicle fleet typical trajectories do not coincide, obtaining the sum of the pairwise overlap between different vehicles of the selected vehicle fleet ; ； wherein, represents the degree of overlap of the key area, the key section covered by vehicle i and vehicle k, represents whether to select vehicle k, i = 1 represents to select the current vehicle networking, i = 0 represents not to select the current vehicle networking; According to the selected number of vehicle fleets , the sum of typical trajectory coverage areas of the selected vehicle fleets , and the sum of two-by-two overlap degrees between different vehicles of the selected vehicle fleets , a mobile detection carrier selection model is established, an enumeration algorithm is used for solving, and a mathematical expression of the mobile detection carrier selection model is obtained. The mathematical expression of the mobile detection carrier selection model is expressed as: ； wherein, is a maximization function, is a networking target weight coefficient.

Images