CN111009126A

CN111009126A - Vehicle driving track prediction method, system and equipment based on bayonet snapshot data

Info

Publication number: CN111009126A
Application number: CN201911337818.1A
Authority: CN
Inventors: 熊永福; 吴涛
Original assignee: Chongqing Unisinsight Technology Co Ltd
Current assignee: Chongqing Unisinsight Technology Co Ltd
Priority date: 2019-12-23
Filing date: 2019-12-23
Publication date: 2020-04-14

Abstract

The invention discloses a vehicle running track prediction method, a system and equipment based on bayonet snapshot data, wherein the method comprises the following steps: acquiring historical track data of a vehicle; aggregating the vehicle historical track data to obtain track data of each vehicle; aggregating the vehicle historical track data to obtain a vehicle track data set of each time interval; constructing a Markov state transition matrix of the road network vehicles based on the vehicle track data set of each time interval; constructing a Markov state transition matrix of an individual vehicle based on the track data of each vehicle; and predicting the probability of passing through each gate according to the Markov state transition matrix of the individual vehicle and the Markov state transition matrix of the road network vehicle. The method can well make up for the defect that the existing vehicle track prediction method cannot perform personalized prediction.

Description

Vehicle driving track prediction method, system and equipment based on bayonet snapshot data

Technical Field

The invention relates to the field of intelligent traffic, in particular to a vehicle driving track prediction method, a system and equipment based on bayonet snapshot data.

Background

With the rapid development of cities and the popularization of private cars, traffic jam, illegal vehicle investigation and control and the like bring huge pressure on urban traffic. However, with the rapid development of the information technology industry, advanced image recognition, video monitoring and communication technologies are widely applied to the traffic field, and related departments gradually alleviate the current situation through a plurality of means such as a big data processing technology intelligent analysis technology and the like. At present, a lot of vehicle snapshot cameras are arranged at road gates in various cities and on expressways among the cities, so that a lot of road traffic information and vehicle traffic information can be generated, and the running tracks of all vehicles are recorded; in addition, the handheld mobile terminal or vehicle-mounted driving recording equipment with the positioning function, such as a smart phone and a tablet personal computer, collects the GPS position in real time and records the driving track of each collected vehicle. The method analyzes vehicle track information and is applied to a research hotspot in the field of intelligent transportation, and mainly focuses on vehicle track prediction, road section traffic condition analysis, road operation safety analysis and the like.

Generally, vehicle trajectory data has a time sequence and regularity. By processing the trajectory data of the vehicle and extracting the characteristics, the position and the spatial distribution of the vehicle in a future period can be effectively predicted. If the track of the vehicle can be well predicted, the urban congestion can be effectively relieved to a certain extent, the accuracy of vehicle distribution and control is improved, and the like. Therefore, the prediction of the track of the vehicle has important significance for efficient operation, safety management and the like of urban traffic.

However, in the existing large number of vehicle trajectory prediction methods, the positions and the spatial distribution of vehicles in the future time period are predicted only through vehicle GPS positioning data or vehicle track data of a road network, the driving preference of each vehicle is not utilized to perform personalized prediction, more importantly, the time spent on reaching the spatial position is not predicted, if the time spent on reaching the spatial position is predicted while the future spatial position of the vehicle is predicted, great effects are brought to various applications of intelligent traffic, such as real-time prediction of traffic flow at a gate, vehicle tracking, intelligent traffic lights, diversion commands and the like.

Disclosure of Invention

In view of the above-mentioned shortcomings of the prior art, the present invention provides a method, system and device for predicting a vehicle driving track based on bayonet snapshot data, which are used to solve the shortcomings of the prior art.

In order to achieve the above and other related objects, the present invention provides a vehicle driving track prediction method based on bayonet snapshot data, including:

acquiring historical track data of a vehicle;

aggregating the vehicle historical track data to obtain track data of each vehicle;

aggregating the vehicle historical track data to obtain a vehicle track data set of each time interval;

constructing a Markov state transition matrix of the road network vehicles based on the vehicle track data set of each time interval;

constructing a Markov state transition matrix of the individual vehicle based on the track data of each vehicle;

and predicting the probability of passing through each gate according to the Markov state transition matrix of the individual vehicle and the Markov state transition matrix of the road network vehicle.

Optionally, the vehicle driving track prediction method further includes:

and predicting the time point when the vehicle reaches each gate according to the predicted probability of passing each gate and the average travelling time of the adjacent gates in the historical simultaneous interval.

Optionally, aggregating the historical trajectory data of the vehicles to obtain trajectory data of each vehicle, including:

grouping and aggregating the historical track data of the vehicles according to the license plate numbers and sequencing the historical track data according to the snapshot time to obtain a gate sequence of each vehicle after sequencing the vehicles according to the time, namely the track data of each vehicle, wherein the track data of each vehicle comprises the license plate numbers, the snapshot time sequence and the gate number sequence.

Optionally, aggregating the historical trajectory data of the vehicle to obtain a vehicle trajectory data set for each time interval, including:

grouping and aggregating vehicle historical track data according to license plate numbers and time intervals and sequencing the vehicle historical track data according to snapshot time, firstly obtaining bayonet sequence data of each vehicle after sequencing in each time interval, and then grouping and aggregating the bayonet sequence data according to the time intervals to obtain a track data set of each time interval, wherein the track data set of each time interval comprises the time intervals, a snapshot time sequence set and a bayonet number sequence set.

Optionally, the constructing a markov state transition matrix of individual vehicles based on the trajectory data of each vehicle includes:

and calculating the transition probability of the vehicle between the checkpoints and between the checkpoint sequence and the checkpoints based on the track data of each vehicle, and combining to obtain the Markov state transition matrix of the individual vehicle.

Optionally, the constructing a markov state transition matrix of the road network vehicle based on the vehicle track data set of each time interval includes:

and calculating transition probabilities between gates and between gate sequences to gates based on the vehicle track data set of each time interval, and combining to obtain a Markov state transition matrix of the road network vehicles.

Optionally, predicting the probability of passing through each gate according to the markov state transition matrix of the individual vehicle and the markov state transition matrix of the road network vehicle includes:

acquiring a first transition probability of the vehicle reaching a next gate based on the current track of the vehicle and the Markov state transition matrix of the individual vehicle;

acquiring a second transition probability of the vehicle reaching the next gate based on the current track of the vehicle and the Markov state transition matrix of the road network vehicle;

and obtaining the probability of passing through each bayonet according to the first transition probability, the second transition probability and the proportion of the first transition probability and the second transition probability.

To achieve the above and other related objects, the present invention provides a vehicle travel track prediction system based on bayonet snapshot data, comprising:

the data acquisition module is used for acquiring historical track data of the vehicle;

the first aggregation module is used for aggregating the vehicle historical track data to obtain track data of each vehicle;

the second aggregation module is used for aggregating the vehicle historical track data to obtain a vehicle track data set of each time interval;

the first Markov state transition matrix construction module is used for constructing a Markov state transition matrix of the road network vehicle based on the vehicle track data set of each time interval;

the second Markov state transition matrix building module is used for building a Markov state transition matrix of the individual vehicle based on the track data of each vehicle;

and the checkpoint prediction module is used for predicting the probability of passing through each checkpoint according to the Markov state transition matrix of the individual vehicle and the Markov state transition matrix of the road network vehicle.

Optionally, the vehicle travel track prediction system further includes:

and the time prediction module is used for predicting the time point when the vehicle reaches each gate according to the predicted probability of passing each gate and the average running time of the adjacent gates in the historical simultaneous time interval.

To achieve the above and other related objects, the present invention provides an apparatus comprising: a processor and a memory;

the memory is configured to store a computer program and the processor is configured to execute the computer program stored by the memory to cause the apparatus to perform the method.

As described above, the vehicle driving track prediction method, system and device based on the bayonet snapshot data of the present invention have the following beneficial effects:

the invention fully utilizes vehicle track data, and can respectively construct a Markov state transition matrix between the road network vehicle and the individual vehicle about the gate and an average running time characteristic set of any adjacent gates of a plurality of subdivision time periods according to the historical track information of the road network vehicle and the historical track information of the individual vehicle. According to the current passing gate track information of the vehicle, the probability of the vehicle passing each gate is predicted by combining the constructed Markov state transition matrix, and the time point of the vehicle reaching each gate is predicted by combining the average running time characteristic set of the adjacent gates in the historical same time period. The method not only carries out personalized prediction on the vehicle track information, but also carries out prediction on the time of arriving at the next possible position on the basis, thereby well making up for the defect that the current vehicle track prediction method can not carry out personalized prediction and time point prediction, and bringing more application scenes for intelligent traffic.

Drawings

Fig. 1 is a flowchart of a vehicle driving track prediction method based on bayonet snapshot data according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a feasible bayonet path for a vehicle;

FIG. 3 is a schematic diagram of the frequency of transfer between one-way bayonets;

FIG. 4 is a schematic diagram of transition probability between one-way checkpoints;

fig. 5 is a schematic diagram of a vehicle driving track prediction system based on bayonet snapshot data according to an embodiment of the present invention.

Detailed Description

The embodiments of the present invention are described below with reference to specific embodiments, and other advantages and effects of the present invention will be easily understood by those skilled in the art from the disclosure of the present specification. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.

It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number, shape and size of the components in actual implementation, and the type, quantity and proportion of the components in actual implementation may be changed freely, and the layout of the components may be more complicated.

As shown in fig. 1, a vehicle driving track prediction method based on bayonet snapshot data includes:

s11, acquiring vehicle historical track data;

s12, aggregating the historical track data of the vehicles to obtain the track data of each vehicle;

s13, aggregating the vehicle historical track data to obtain a vehicle track data set of each time interval;

s14, constructing a Markov state transition matrix of the road network vehicles based on the vehicle track data set of each time interval;

s15, constructing a Markov state transition matrix of the individual vehicle based on the track data of each vehicle;

and S16, predicting the probability of passing through each gate according to the Markov state transition matrix of the individual vehicle and the Markov state transition matrix of the road network vehicle.

According to the invention, a Markov state transition matrix of the road network vehicles and the individual vehicles and an average running time feature set of any adjacent gates of a plurality of subdivision time periods are respectively constructed according to the historical track data of the road network vehicles and the historical track data of the individual vehicles. According to the current passing gate track information of the vehicle, the probability of passing each gate of the vehicle is predicted by combining the constructed Markov state transition matrix, and the vehicle track information can be predicted in an individualized way.

In one embodiment, the vehicle travel track prediction method further includes:

According to the method, the time points of the vehicles reaching each gate are predicted by combining the average running time characteristic set of the adjacent gates in the historical same time period on the basis of personalized prediction of vehicle track information. The method not only carries out personalized prediction on the vehicle track information, but also carries out prediction on the time of arriving at the next possible position on the basis, thereby well making up for the defect that the current vehicle track prediction method can not carry out personalized prediction and time point prediction, and bringing more application scenes for intelligent traffic.

In step S11, vehicle history track data is acquired.

Specifically, vehicle snapshot data of a total mount in a given area, that is, historical travel track data of a vehicle, are acquired, and are collectively referred to as simply: vehicle historical track data; the vehicle historical track data mainly comprises: license plate number, bayonet serial number and snapshot time. In order to ensure the timeliness of the historical data, only the data in the latest time range of the vehicle, such as the latest 30 days, the latest 60 days, etc., is obtained, and in this embodiment, the data in the latest 30 days is not taken as an example. Suppose that there is a region Q as shown in fig. 2, where a-H represent the location of the gate points, and any two gates with a connecting line represent the traversable path of the vehicle. Firstly, vehicle snapshot data Raw _ data of all gates in the area Q in the last 30 days are obtained, and the main obtained data fields are as follows: license plate number, bayonet serial number and snapshot time.

In step S12, aggregating the vehicle historical trajectory data to obtain trajectory data of each vehicle, including: and grouping and aggregating the historical track data of the vehicles according to the license plate numbers and sequencing the historical track data according to the snapshot time to obtain a bayonet sequence of each vehicle after sequencing according to the time, namely the track data of each vehicle. In the present embodiment, the trajectory data of each vehicle is denoted as GJ _ 1. The track data of each vehicle comprises a license plate number, a snapshot time sequence and a gate number sequence.

Specifically, the data Raw _ data obtained in step S11 are grouped and aggregated (group) according to the license plate number, and then sorted according to the snapshot time, so that the gate passing time sequence of each vehicle in the last 30 days and the gate sequence corresponding thereto can be obtained, and then the data of each vehicle are aggregated (String _ Agg), so as to obtain the track data G of each vehicle_p＝{(p,(g₁,g₂,g₃,...,g_i),(t₁,t₂,t₃,...,t_i) -) }, where p denotes the number plate of the car, (t)₁,t₂,t₃,...,t_i) Represents a time series, (g)₁,g₂,g₃,...,g_i) And representing a bayonet sequence corresponding to the time sequence, namely the driving track of the vehicle in the last 30 days of the area. G is ═ G_pDenotes the set of all vehicle trajectories.

In step S13, aggregating the vehicle history trajectory data to obtain a vehicle trajectory data set for each time interval, including: the vehicle historical track data are grouped and aggregated according to license plate numbers and time intervals and are sorted according to snapshot time, the checkpoint sequence data of each vehicle after being sorted in each time interval are firstly obtained, then the checkpoint sequence data are grouped and aggregated according to the time intervals, and a track data set of each time interval is obtained. The time interval category may be every ten minutes, every hour, etc.

Specifically, the small time intervals are set to be two dimensions of hourly and 10 minutes, respectively, for the division with hourly time intervals, the small time intervals are divided into 24 intervals per day, and for the division with 10 minutes time intervals, the small time intervals are divided into 144 intervals per day.

In this embodiment, only the division calculation with each hour as a time interval is described, the data Raw _ data obtained in step S11 are grouped and aggregated (Groupby) according to the license plate number, date and time interval, then sorted according to the snapshot time, and aggregated (String _ Agg) according to the time interval, so as to obtain the vehicle track data set T in the time interval of the last 30 days_ho_ur，

Wherein, ts_jRepresents the interval of time j, and the value range of j is [1,24 ]]；

A snapshot time sequence representing a vehicle passing through the region Q within the time interval j, then

A sequence set which is formed by snapshot time sequences of all vehicles passing through the gate in the region Q in the time interval j;

is shown and

corresponding bayonet numbering sequences (track sequences).

Similarly, a vehicle trajectory data set T with time intervals of every 10 minutes can be obtained_10minutesWherein j has a value in the range of [1,144 ]]。

In step S14, constructing a markov state transition matrix of the road network vehicle based on the vehicle trajectory data set of each time interval, including: and calculating transition probabilities between gates and between gate sequences to gates based on the vehicle track data set of each time interval, and combining to obtain a Markov state transition matrix of the road network vehicles. In this embodiment, the markov state transition matrix of the road network vehicle can be represented as M _2k, where k represents a unique identifier (e.g., license plate number) of the vehicle.

Specifically, according to the set G formed by all vehicle tracks obtained in step S12, statistics are accumulated first on the transfer frequency from the gate to the gate and from the gate sequence to the gate, then the transfer frequency from the gate to the gate and from the gate sequence to the gate is calculated, and then the markov state transfer matrix of the road network vehicle is obtained.

Fig. 3 shows a transfer frequency from one-way path bayonet to bayonet, and fig. 4 shows a transfer frequency from one-way path bayonet to bayonet. As shown in fig. 3, the transition frequency from the bayonet to the bayonet is accumulated and counted, then the transition probability from the bayonet to the bayonet is calculated, as shown in fig. 4, and then the markov state transition matrix of the road network vehicle is obtained, wherein only the one-way path is shown, as shown only a->The transition probability in the B direction can be calculated to obtain B->The transition probability in the A direction and the transition probabilities among other checkpoints are the same. Then the 1 st-order bayonet-to-bayonet transition probability matrix M can be obtained from FIG. 4₁The dictionary shows that:

{A->B:0.5,A->C:0.2,A->D:0.3,B->C:0.3,B->E:0.5,B->G:0.2,C->G:0.6,D->E:0.3,D->F:0.7,E->H:0.6,E->G:0.4,F->E:0.2,F->H:0.8}

similarly, it can be obtained that the transition probability matrix M from the 2 nd order bayonet sequence to the bayonet is assumed₂And the dictionary can be used as follows:

{AB->C:0.1,AB->E:0.6,AB->G:0.3,AC->G:0.4,AD->F:0.6,BC->G:0.1,BE->H:0.4,BE->G:0.4,DF->H:0.6,DE->H:0.2,DE->G:0.3,FE->G:0.1}

similarly, it can be obtained that the transition probability matrix M from 3-order bayonet sequence to bayonet is assumed₃And the dictionary can be used as follows:

{ABC->G:0.2,ABE->H:0.3,ABE->G:0.2,ADE->G:0.15,ADE->H:0.2,ADF->H:0.3,ADF->E:0.3}

note that M here₂、M₃Fig. 4 shows only the transition probability from 1 st-order bayonet to bayonet, assuming that such transition probability is present (which can be calculated from trajectory data in actual cases). Here, M is₁、M₂、M₃And are combined into a dictionary M _2 k.

In step S15, constructing a markov state transition matrix for the individual vehicle based on the trajectory data for each vehicle, comprising: and calculating the transition probability of the vehicle between the checkpoints and between the checkpoint sequence and the checkpoints based on the track data of each vehicle, and combining to obtain the Markov state transition matrix of the individual vehicle. In the present embodiment, the markov state transition matrix for an individual vehicle may be represented as M _ 1.

For each G_pE G, the dictionary M composed of 3 transfer matrixes for each vehicle can be obtained by adopting the same method as the step S14_p1，M_p2，M_p3Wherein p represents different license plate numbers, the sameThe treatment can merge M_p1，M_p2，M_p3Is M _ 1.

In step S16, predicting the probability of passing through each gate based on the markov state transition matrix of the individual vehicle and the markov state transition matrix of the road network vehicle includes:

Specifically, assume that a certain number of vehicles is p_lWhen the vehicle (a) is driven to the gate (a), the probability of the vehicle possibly reaching the next gate and the time point of reaching the gate need to be predicted. Since only the 1 st track (i.e. the gate A) of the car is known at present, it is first checked in M _1 whether there is a number p of the car_lIf the data of (1) is existed, the transition probabilities (namely the first transition probabilities) from the gate A to all other gates are continuously searched, and then the data are fused with the corresponding transition probabilities (the second transition probabilities) in the Markov state transition matrix M _2k of the road network vehicle according to a certain proportion, so that the probabilities of passing through all the gates can be obtained. In this embodiment, the ratio of the first transition probability to the second transition probability is 4: 1, according to 4: 1 (with greater generalization ability). Suppose M_pIn the presence of a number plate of p_lThe transition probability at the bayonet A is: { A->B:0.6,A->C:0.2,A->D:0.2, corresponding to a transition probability of { A->B:0.5,A->C:0.2,A->And D:0.3}, the probability that the vehicle possibly reaches the next gate can be predicted as follows: { B:0.6 × 0.8+0.5 × 0.2, C:0.2 × 0.8+0.2, D:0.2 × 0.8+0.3 × 0.2} - { B:0.58, C:0.2, D:0.22 }; if M is_pAbsence of license plate number p_lThe probability that the vehicle reaches the next gate can beIs taken as the transition probability in M _2k, which is { A->B:0.5,A->C:0.2,A->D:0.3}, namely the probability of reaching the next checkpoint.

Thus, the probability that the vehicle will reach the next gate is obtained in steps S11 to S16.

In another embodiment, the vehicle travel track prediction method further includes:

Specifically, the average travel time may be calculated by the following method:

the average time spent by the vehicles at the adjacent gates in each time interval is calculated from the trajectory data set GJ _2 for each time interval, and the construction may be based on a dictionary Ti having the time interval + a certain gate number + an adjacent gate number as a lookup key and the average time as a value, where i represents the kind of the time interval.

Specifically, for T obtained in S13_hourAnd calculating the average travel time of adjacent checkpoints in all track sequences in each time interval. For example, in interval 1(00:00:00-00:59:59), a track sequence [ (A, B), (A, B), (A, B, C), (A, B, E), (A, B, G), (A, C), (A, D) is assumed]The corresponding snapshot time sequence is [ (1572279093,1572279125), (1572279322,1572279357), (1572279101,1572279131,1572279182), (1572279066,1572279098,1572279132), (1572279122,1572279158,1572279176) (1572279188,1572279242), (1572279262,1572279332), (1572279388,1572279366)]Here, the time series is constituted by time stamps. The average travel time of the adjacent gates with tracks can be calculated, e.g. for a->B，

AvgTime(A->B)＝[(1572279125-1572279093)+(1572279357-1572279322)+(1572279131-1572279101)+(1572279098-1572279066)+(1572279158-1572279122)]/5＝(32+35+30+32+36)/5＝33s；

Can calculate by the same way

AvgTime (B- > C) ═ 51s, AvgTime (B- > E) ═ 34s, AvgTime (B- > G) ═ 18s, AvgTime (a- > C) ═ 54s, and AvgTime (a- > D) ═ 74 s. The dictionary representation may be:

T_{hour_t1}＝{A->B:33,B->C:51,B->E:34,B->G:18,A->C:54,A->D:74}

therefore, the average travel time dictionary set T of the adjacent gates of all the tracks of each section can be obtained_{hour_t}＝{T_{hour_tj}J e [1,24 ]]. In the same way, T can be obtained_{10minutes_t}＝{T_{10minutes_tj}J e [1,144 ]]。

And according to the predicted next possible bayonet, predicting the corresponding arrival time point. Given that the possible arrival bayonet is B, C, D, assuming the current time point 00:10:11, it is at T_{10minutes_t}The corresponding time interval is T_{10minutes_t2}Let T be_{10minutes_t2}In which there is data { A->B:31,A->C:52,A->D:69, the time points B:00:10:42, C:00:11:03 and C:00:11:20 which respectively reach B, C, D can be predicted, and the transition probability (p) is obtained in Step8_lThe existing case), the predicted trajectory can be found to be:

if T_{10minutes_t2}In the absence of a catalyst such as A->The average travel time of B can be spread to mean value fill through adjacent intervals, e.g. T_{10minutes_t1}And T_{10minutes_t3}Wherein A->B mean value substitution of average travel time, spread up to three times, if not still present, may be at corresponding T_{hour_t1}Middle search A->Supplementing the value corresponding to B, and if not, continuing to increase the value at T_{hour_tj}Diffusion until found (always found, since the transition probability has been predicted from the historical data, there must be a historical travel time for the adjacent trajectory gate); when a plurality of values do not exist or do not exist in whole, the same principle can be supplemented.

If it is known that a certain vehicle runs to the gate B and the track is A- > B, only the search for A in M _1 is replaced by the search for the transition probability from AB to the adjacent gate in M _1, and the probability of the vehicle possibly reaching the next gate and the time point of reaching the gate can be predicted in the same way. Similarly, if the track is known to be A- > B-C, the search value is changed to ABC. If the search sequence length is larger than 3, such as A- > D- > F- > E, the nearest three bayonet sequences, namely D- > F- > E, are directly taken as the known tracks. If the known track A- > B- > C sequence does not exist, deleting the bayonet at the top of the sequence to continue searching, if B- > C is continuously searched, repeatedly searching until the B- > C is found, and if the B- > C is not found, returning a null value to indicate that the history track or the bayonet point does not exist. Thus, the track prediction can be completed under all possible conditions.

As shown in fig. 5, a vehicle travel track prediction system based on bayonet snapshot data includes:

a data acquisition module 51 for acquiring vehicle historical track data;

the first aggregation module 52 is configured to aggregate the vehicle historical track data to obtain track data of each vehicle;

the second aggregation module 53 is configured to aggregate the vehicle historical trajectory data to obtain a vehicle trajectory data set of each time interval;

a first markov state transition matrix constructing module 54, configured to construct a markov state transition matrix of the road network vehicle based on the vehicle trajectory data set of each time interval;

a second markov state transition matrix construction module 55, configured to construct a markov state transition matrix of the individual vehicle based on the trajectory data of each vehicle;

and the gate prediction module 56 is used for predicting the probability of passing through each gate according to the Markov state transition matrix of the individual vehicle and the Markov state transition matrix of the road network vehicle.

The data acquisition module 51 is used for acquiring vehicle historical track data. Specifically, vehicle snapshot data of a total mount in a given area, that is, historical travel track data of a vehicle, are acquired, and are collectively referred to as simply: vehicle historical track data; the vehicle historical track data mainly comprises: license plate number, bayonet serial number and snapshot time. In order to ensure the timeliness of the historical data, only the data in the latest time range of the vehicle, such as the latest 30 days, the latest 60 days, etc., is obtained, and in this embodiment, the data in the latest 30 days is not taken as an example. Suppose that there is a region Q as shown in fig. 2, where a-H represent the location of the gate points, and any two gates with a connecting line represent the traversable path of the vehicle. Firstly, vehicle snapshot data Raw _ data of all gates in the area Q in the last 30 days are obtained, and the main obtained data fields are as follows: license plate number, bayonet serial number and snapshot time.

The first aggregation module 52 aggregates the vehicle historical track data to obtain track data of each vehicle, and includes: and grouping and aggregating the historical track data of the vehicles according to the license plate numbers and sequencing the historical track data according to the snapshot time to obtain a bayonet sequence of each vehicle after sequencing according to the time, namely the track data of each vehicle. In the present embodiment, the trajectory data of each vehicle is denoted as GJ _ 1. The track data of each vehicle comprises a license plate number, a snapshot time sequence and a gate number sequence.

Specifically, the data Raw _ data acquired by the data acquisition module are grouped and aggregated (Groupby) according to the license plate number, then the data Raw _ data are sequenced according to the snapshot time, the passing time sequence of each vehicle in the nearest 30 days and the corresponding gate sequence can be obtained, and then the data of each vehicle are aggregated (String _ Agg) to obtain the track data G of each vehicle_p＝{(p,(g₁,g₂,g₃,...,g_i),(t₁,t₂,t₃,...,t_i) -) }, where p denotes the number plate of the car, (t)₁,t₂,t₃,...,t_i) Represents a time series, (g)₁,g₂,g₃,...,g_i) And representing a bayonet sequence corresponding to the time sequence, namely the driving track of the vehicle in the last 30 days of the area. G is ═ G_pDenotes the set of all vehicle trajectories.

The first aggregation module 52 aggregates the vehicle historical trajectory data to obtain a vehicle trajectory data set for each time interval, and includes: the vehicle historical track data are grouped and aggregated according to license plate numbers and time intervals and are sorted according to snapshot time, the checkpoint sequence data of each vehicle after being sorted in each time interval are firstly obtained, then the checkpoint sequence data are grouped and aggregated according to the time intervals, and a track data set of each time interval is obtained. The time interval category may be every ten minutes, every hour, etc.

is shown and

corresponding bayonet numbering sequences (track sequences).

The first markov state transition matrix constructing module 54 is configured to construct a markov state transition matrix of the road network vehicle based on the vehicle trajectory data set for each time interval, and includes: and calculating transition probabilities between gates and between gate sequences to gates based on the vehicle track data set of each time interval, and combining to obtain a Markov state transition matrix of the road network vehicles. In this embodiment, the markov state transition matrix of the road network vehicle can be represented as M _2k, where k represents a unique identifier (e.g., license plate number) of the vehicle.

Specifically, according to a set G formed by all vehicle tracks, accumulating and counting the transfer frequency from the gate to the gate and from the gate sequence to the gate, calculating the transfer frequency from the gate to the gate and from the gate sequence to the gate, and then obtaining a Markov state transfer matrix of the road network vehicles.

Fig. 3 shows a transfer frequency from one-way path bayonet to bayonet, and fig. 4 shows a transfer frequency from one-way path bayonet to bayonet. As shown in fig. 3, the transition frequency from the bayonet to the bayonet is accumulated and counted, then the transition probability from the bayonet to the bayonet is calculated, as shown in fig. 4, and then the markov state transition matrix of the road network vehicle is obtained, wherein only the one-way path is shown, as shown only a->The transition probability in the B direction can be calculated to obtain B->The transition probability in the A direction and the transition probabilities between other checkpoints are the same. Then the 1 st-order bayonet-to-bayonet transition probability matrix M can be obtained from FIG. 4₁The dictionary shows that:

{ABC->G:0.2,ABE->H:0.3,ABE->G:0.2,ADE->G:0.15,ADE->H:0.2,ADF->H:0.3,ADF->E:0.3}

The second markov state transition matrix constructing module 55 is configured to construct a markov state transition matrix of the individual vehicle based on the trajectory data of each vehicle, and includes: and calculating the transition probability of the vehicle between the checkpoints and between the checkpoint sequence and the checkpoints based on the track data of each vehicle, and combining to obtain the Markov state transition matrix of the individual vehicle. In the present embodiment, the markov state transition matrix for an individual vehicle may be represented as M _ 1.

For each G_pE G, the dictionary M composed of 3 transfer matrixes for each vehicle can be obtained by adopting the same method as the step S14_p1，M_p2，M_p3Where p represents different license plate numbers, and M may be combined in the same way_p1，M_p2，M_p3Is M _ 1.

The predicting module 55 predicts the probability of passing through each gate according to the markov state transition matrix of the individual vehicle and the markov state transition matrix of the road network vehicle, and includes:

Specifically, assume that a certain number of vehicles is p_lWhen the vehicle (a) is driven to the gate (a), the probability of the vehicle possibly reaching the next gate and the time point of reaching the gate need to be predicted. Since only the 1 st track (i.e. the gate A) of the car is known at present, it is first checked in M _1 whether there is a number p of the car_lIf the data of (1) is existed, the transition probabilities (namely the first transition probabilities) from the gate A to all other gates are continuously searched, and then the data are fused with the corresponding transition probabilities (the second transition probabilities) in the Markov state transition matrix M _2k of the road network vehicle according to a certain proportion, so that the probabilities of passing through all the gates can be obtained. In this embodiment, the ratio of the first transition probability to the second transition probability is 4: 1, according to 4: 1 (with greater generalization ability). Suppose that there is a license plate number p in M _1_lThe transition probability at the bayonet A is: { A->B:0.6,A->C:0.2,A->D:0.2, corresponding to a transition probability of { A->B:0.5,A->C:0.2,A->And D:0.3}, the probability that the vehicle possibly reaches the next gate can be predicted as follows: { B:0.6 × 0.8+0.5 × 0.2, C:0.2 × 0.8+0.2, D:0.2 × 0.8+0.3 × 0.2} - { B:0.58, C:0.2, D:0.22 }; if M _1 does not exist, the number plate is p_lThe probability that the vehicle reaches the next gate can be the transition probability in M, which is { A->B:0.5,A->C:0.2,A->D:0.3}, that is, the next bayonet can be reachedThe probability of (c).

Thus, the probability that the vehicle reaches the next gate is obtained.

In another embodiment, the vehicle driving track prediction system further comprises a time prediction module, which is used for predicting the time point when the vehicle reaches each gate according to the predicted probability of passing each gate and the historical average driving time of the adjacent gates in the same time interval.

According to the method, the time points of the vehicles reaching each gate are predicted by combining the average running time characteristic set of the adjacent gates in the historical same time period on the basis of personalized prediction of vehicle track information. The vehicle track prediction method based on the time point prediction not only can perform personalized prediction on vehicle track information, but also can predict the time for reaching the next possible position on the basis, thereby well making up for the defect that the current vehicle track prediction method cannot perform personalized prediction and time point prediction, and bringing more application scenes for intelligent transportation.

Can calculate by the same way

T_{hour_t1}＝{A->B:33,B->C:51,B->E:34,B->G:18,A->C:54,A->D:74}

if T_{10minutes_t2}In the absence of a catalyst such as A->The average travel time of B can be spread to mean value fill through adjacent intervals, e.g. T_{10minutes_t1}And T_{10minutes_t3}Wherein A->B mean value substitution of average travel time, spread up to three times, if not still present, may be at corresponding T_{hour_t1}Middle search A->Supplementing the value corresponding to B, and if not, continuing to increase the value at T_{hour_tj}Diffusion until found (always found, since the transition probability has been predicted from historical data, then there must beHistorical travel time of the adjacent trajectory gate); when a plurality of values do not exist or do not exist in whole, the same principle can be supplemented.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other ways. For example, the above-described embodiments of the apparatus/terminal device are merely illustrative, and for example, the division of the modules or units is only one logical division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method embodiments may be implemented. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may comprise any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a Random Access Memory (RAM), an electrical carrier signal, a telecommunications signal, a software distribution medium, etc.

The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.

Claims

1. A vehicle driving track prediction method based on bayonet snapshot data is characterized by comprising the following steps:

acquiring historical track data of a vehicle;

constructing a Markov state transition matrix of an individual vehicle based on the track data of each vehicle;

2. The vehicle travel track prediction method based on bayonet snapshot data as recited in claim 1, further comprising:

3. The vehicle driving track prediction method based on bayonet snapshot data as claimed in claim 1, wherein the step of aggregating the vehicle historical track data to obtain track data of each vehicle comprises:

4. The vehicle driving track prediction method based on bayonet snapshot data as claimed in claim 1, wherein aggregating the vehicle historical track data to obtain a vehicle track data set for each time interval comprises:

5. The vehicle driving track prediction method based on bayonet snapshot data as claimed in claim 1, wherein the building of the Markov state transition matrix of the individual vehicle based on the track data of each vehicle comprises:

6. The vehicle driving track prediction method based on the bayonet snapshot data as claimed in claim 1, wherein the constructing a Markov state transition matrix of the road network vehicle based on the vehicle track data set of each time interval comprises:

7. The vehicle driving track prediction method based on the checkpoint snapshot data as claimed in claim 1, wherein predicting the probability of passing through each checkpoint according to the markov state transition matrix of the individual vehicle and the markov state transition matrix of the road network vehicle comprises:

8. A vehicle travel track prediction system based on bayonet snapshot data, the vehicle travel track prediction system comprising:

9. The vehicle travel track prediction system based on bayonet snapshot data as recited in claim 8, further comprising:

10. An apparatus, comprising: a processor and a memory;

the memory is used for storing a computer program, and the processor is used for executing the computer program stored by the memory to cause the device to execute the method of any one of claims 1-8.