CN115830866A - Traffic jam inference method, system, device and medium based on time sequence dynamic graph - Google Patents
Traffic jam inference method, system, device and medium based on time sequence dynamic graph Download PDFInfo
- Publication number
- CN115830866A CN115830866A CN202211447815.5A CN202211447815A CN115830866A CN 115830866 A CN115830866 A CN 115830866A CN 202211447815 A CN202211447815 A CN 202211447815A CN 115830866 A CN115830866 A CN 115830866A
- Authority
- CN
- China
- Prior art keywords
- congestion
- propagation
- time
- road
- road section
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000010586 diagram Methods 0.000 claims abstract description 19
- 230000001364 causal effect Effects 0.000 claims description 46
- 238000004364 calculation method Methods 0.000 claims description 34
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000012512 characterization method Methods 0.000 claims description 10
- 238000004891 communication Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000001914 filtration Methods 0.000 claims description 9
- 238000004140 cleaning Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 abstract description 19
- 238000009792 diffusion process Methods 0.000 abstract description 6
- 238000012546 transfer Methods 0.000 description 13
- 238000004422 calculation algorithm Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 238000012795 verification Methods 0.000 description 6
- 239000012634 fragment Substances 0.000 description 5
- 230000007246 mechanism Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008034 disappearance Effects 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000011524 similarity measure Methods 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000000551 statistical hypothesis test Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a traffic jam inference method, a system, equipment and a medium based on a time sequence dynamic graph, wherein the method comprises the following steps: the time sequence dynamic graph is adopted to represent the congestion state characteristics of the traffic flow time sequence data of each road section; calculating a congestion propagation road section set caused by congestion of each road section in the congestion state according to the obtained congestion state time sequence feature vector; calculating a road network congestion space-time propagation map according to the congestion propagation road section set; and deducing a key propagation road section of the congestion space-time propagation chain and the congestion propagation probability of the key propagation road section according to the road network congestion space-time propagation diagram. According to the invention, the representation of the time sequence unstable characteristics under the congestion state is realized by adopting the time sequence dynamic diagram, and then the key propagation road section of the congestion space-time propagation chain and the congestion propagation probability of the key propagation road section are deduced based on the road network congestion space-time propagation diagram, so that the diffusion dissipation process of the traffic congestion of the real road section is accurately inverted, and the space-time influence range of the traffic congestion is accurately quantized.
Description
Technical Field
The invention relates to the technical field of traffic management, in particular to a traffic jam inference method, a system, equipment and a medium based on a time sequence dynamic graph.
Background
A data representation mode (or called as a 'state') with physical significance is obtained from massive high-dimensional traffic space-time observation data, and accurate and interpretable decision support is provided for road system operation and traffic management planning. Independent traffic states on different road sections within discrete-time slices are usually easily obtained through traffic observation data, but the traffic operation process is continuous in time and space. Meanwhile, the traffic flow has obvious space-time instability and space-time correlation, namely the traffic state of a certain road section is possibly influenced by the preamble time interval or the surrounding road sections simultaneously so as to change the traffic state, and how to model the space-time instability and the space-time correlation is an important problem in the current research. Traffic congestion is a typical traffic state with strong propagation in space and uneven stability in time sequence, and not only the spatial mode and the time mode of the traffic congestion state are separately researched from the perspective of a road network cascade phenomenon and a time sequence correlation theory, but the current research lacks further exploration on the spatial propagation link-level path and probability of the road network traffic operation state in continuous time.
Disclosure of Invention
The invention aims to overcome the defects that a space propagation link section with traffic jam on a real road section cannot be accurately inverted and the jam propagation probability in the prior art, and provides a traffic jam inference method, a system, equipment and a medium based on a time sequence dynamic graph.
The invention solves the technical problems through the following technical scheme:
the invention provides a traffic jam inference method based on a time sequence dynamic graph, which comprises the following steps:
calculating traffic flow time series data of at least one road section;
characterizing the congestion state characteristics of the traffic flow time sequence data of each road section by adopting a time sequence dynamic graph to obtain a congestion state time sequence characteristic vector;
calculating a congestion propagation road section set caused by each road section in a congestion state according to the congestion state time sequence feature vector;
calculating to obtain a road network congestion space-time propagation map according to the congestion propagation road section set;
and deducing a key propagation road section of a congestion space-time propagation chain and the congestion propagation probability of the key propagation road section according to the road network congestion space-time propagation map.
Preferably, before the step of calculating the traffic flow time-series data of at least one road segment, the traffic congestion estimation method further includes:
acquiring a road network topology communication relation;
the step of calculating the traffic flow time-series data of at least one link includes:
calculating the traffic flow time sequence data of at least one road section according to the road network topological connection relation;
the step of calculating the road network congestion space-time propagation map according to the congestion propagation road section set comprises the following steps:
and calculating to obtain a road network congestion space-time propagation map according to the road network topological connection relation and the congestion propagation road section set.
Preferably, the step of calculating the time-series data of the traffic flow of at least one road segment according to the topological connection relationship of the road network comprises:
acquiring road network data of road network topological connectivity;
acquiring historical track data of a target vehicle;
cleaning and filtering the historical track data to obtain processed historical track data;
establishing a spatial index for the processed historical track data according to the road network data, and performing map road network matching on the indexed historical track data to obtain matched historical track data;
and calculating the traffic flow time sequence data of at least one road section according to the matched historical track data.
Preferably, the step of calculating a set of congestion propagation segments caused by congestion of each segment in the congestion state according to the congestion state time-series feature vector includes:
calculating a time-space causal relationship of each road section in the congestion state according to the congestion state time sequence feature vector;
and acquiring a congestion propagation road section set caused by congestion of each road section in a congestion state according to the spatiotemporal causal relationship.
Preferably, the step of obtaining the road network congestion space-time propagation map by calculating according to the road network topology connectivity relation and the congestion propagation road section set includes:
calculating to obtain a congestion propagation probability matrix according to the road network topology communication relation and the congestion propagation road section set;
and obtaining the road network congestion space-time propagation map according to the congestion propagation probability matrix.
Preferably, the step of deducing a key propagation segment of a congestion spatio-temporal propagation chain and a congestion propagation probability of the key propagation segment according to the road network congestion spatio-temporal propagation map comprises:
calculating to obtain a congestion space-time propagation chain according to the road network congestion space-time propagation diagram;
checking the congestion space-time propagation chain according to the space-time causal relationship to obtain a congestion key space-time propagation chain with the redundant space-time causal relationship removed;
and obtaining the key propagation road section and the congestion propagation probability of the key propagation road section according to the congestion key space-time propagation chain.
Preferably, after the step of deducing a key propagation segment of a congestion spatio-temporal propagation chain and a congestion propagation probability of the key propagation segment according to the road network congestion spatio-temporal propagation map, the traffic congestion deduction method further comprises the following steps:
and evaluating the key propagation road section of the congested space-time propagation chain to obtain an evaluation result.
The invention provides a traffic jam inference system based on a time sequence dynamic graph, which comprises a first calculation module, a characterization module, a second calculation module, a third calculation module and an inference module;
the first calculation module is used for calculating the traffic flow time sequence data of at least one road section;
the characterization module is used for characterizing the congestion state characteristics of the traffic flow time sequence data of each road section by adopting a time sequence dynamic graph so as to obtain a congestion state time sequence characteristic vector;
the second calculation module is used for calculating a congestion propagation road section set caused by each road section in a congestion state according to the congestion state time sequence feature vector;
the third calculation module is used for calculating a road network congestion space-time propagation map according to the congestion propagation road section set;
the inference module is used for inferring a key propagation road section of a congestion space-time propagation chain and the congestion propagation probability of the key propagation road section according to the road network congestion space-time propagation diagram.
Preferably, the traffic congestion inference system further comprises an acquisition module;
the acquisition module is used for acquiring a road network topology communication relation;
the first calculation module is used for calculating the traffic flow time sequence data of at least one road section according to the road network topological connection relation;
and the third calculation module is used for calculating a road network congestion space-time propagation map according to the road network topology communication relation and the congestion propagation road section set.
Preferably, the first calculation module includes a first obtaining unit, a second obtaining unit, a processing unit, a matching unit and a first calculation unit;
the first acquisition unit is used for acquiring road network data of a road network topological connectivity relation;
the second acquisition unit is used for acquiring historical track data of the target vehicle;
the processing unit is used for cleaning and filtering the historical track data to obtain processed historical track data;
the matching unit is used for establishing a spatial index for the processed historical track data according to the road network data and carrying out map road network matching on the indexed historical track data to obtain matched historical track data;
the first calculating unit is used for calculating the traffic flow time series data of at least one road section according to the matched historical track data.
Preferably, the second calculating module includes a second calculating unit and a third obtaining unit;
the second calculating unit is used for calculating a time-space causal relationship of each road section in the congestion state according to the congestion state time sequence feature vector;
the third obtaining unit is configured to obtain a congestion propagation road segment set caused by congestion of each road segment according to the spatiotemporal causal relationship.
Preferably, the third calculation module includes a third calculation unit and a fourth acquisition unit;
the third calculating unit is used for calculating a congestion propagation probability matrix according to the road network topological connection relation and the congestion propagation road section set;
the fourth obtaining unit is configured to obtain the road network congestion space-time propagation map according to the congestion propagation probability matrix.
Preferably, the inference module comprises a fourth calculation unit, a verification unit and a fifth acquisition unit;
the fourth calculation unit is used for calculating a congestion space-time propagation chain according to the road network congestion space-time propagation diagram;
the inspection unit is used for inspecting the congestion space-time propagation chain according to the space-time causal relationship so as to obtain a congestion key space-time propagation chain with the redundant space-time causal relationship removed;
the fifth acquisition unit is used for acquiring the key propagation road section and the congestion propagation probability of the key propagation road section according to the congestion key space-time propagation chain.
Preferably, the traffic congestion inference system further comprises an evaluation module;
the evaluation module is used for evaluating the key propagation road section of the congested space-time propagation chain to obtain an evaluation result.
A third aspect of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for traffic congestion estimation based on a time-series dynamic graph according to the first aspect.
A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method for traffic congestion inference based on a time-series dynamic graph according to the first aspect.
The positive progress effects of the invention are as follows:
according to the method, the time sequence dynamic graph is adopted to represent the congestion state characteristics of the traffic flow time sequence data of each road section of each time sequence section, the representation of the time sequence unstable characteristics under the congestion state is realized, the road network congestion space-time propagation graph is obtained by combining the congestion propagation road section set of each road section calculated based on the congestion state time sequence characteristic vector, the key propagation road section of the congestion space-time propagation chain and the congestion propagation probability of the key propagation road section are deduced based on the network congestion space-time propagation graph, the diffusion dissipation process of the traffic congestion of the real road section can be accurately performed, and the space-time influence range of the traffic congestion is accurately quantized.
Drawings
Fig. 1 is a flowchart of a traffic congestion estimation method based on a time-series dynamic graph according to embodiment 1 of the present invention.
Fig. 2 is a schematic diagram of traffic flow time-series data on a single road segment according to embodiments 1 and 2 of the present invention, which is characterized by a dynamic graph structure for traffic congestion status.
Fig. 3 is a flowchart of step 1011 of the traffic congestion estimation method based on the time series dynamic graph according to embodiment 1 of the present invention.
Fig. 4 is a flowchart of step 103 of the traffic congestion estimation method based on the time sequence dynamic graph according to embodiment 1 of the present invention.
Fig. 5 is a schematic diagram of obtaining a road network congestion space-time propagation map according to embodiments 1 and 2 of the present invention.
Fig. 6 is a flowchart of step 1041 of the traffic congestion estimation method based on the time sequence dynamic graph according to embodiment 1 of the present invention.
FIG. 7 shows the route section r obtained in embodiments 1 and 2 of the present invention i Schematic diagram of a candidate congested space propagation chain subtree of (1).
Fig. 8 is a flowchart of step 105 of the traffic congestion estimation method based on the time sequence dynamic graph according to embodiment 1 of the present invention.
FIG. 9 shows the route section r obtained in embodiments 1 and 2 of the present invention i A schematic diagram of key propagation segments of a congested spatio-temporal propagation chain.
Fig. 10 is a schematic structural diagram of a traffic congestion estimation system based on a time-series dynamic graph according to embodiment 2 of the present invention.
Fig. 11 is a schematic structural diagram of an electronic device according to embodiment 3 of the present invention.
Detailed Description
The invention is further illustrated by the following examples, which are not intended to limit the scope of the invention.
Example 1
The embodiment provides a traffic congestion inference method based on a time sequence dynamic graph, as shown in fig. 1, the traffic congestion inference method includes:
102, representing the congestion state characteristics of the traffic flow time sequence data of each road section by adopting a time sequence dynamic graph to obtain a congestion state time sequence characteristic vector;
in this embodiment, a time sequence dynamic graph (i.e., a dynamic graph structure) is used to model the dynamic evolution of the traffic flow state feature of each time sequence segment, so that the time correlation between the state of the time sequence segment and the state of the time span segment is maintained, and the characterization of the time sequence unstable feature in the congestion state is realized.
For the speed time-series data on the single road section obtained in the step 101, in order to obtain a representative time-series segment with interpretability to characterize the traffic running state on the road, a dynamic graph structure is adopted to model the state change of adjacent time-series segments, and the dynamic change rule of the traffic time-series state along with the time is captured. Specifically, TS = { v ] for a time series t T =0, 1.., m }, where v t And representing a speed value at the t moment in the TS, firstly selecting different time intervals to perform up-sampling and down-sampling on sequence data to obtain a plurality of pieces of time sequence data with different time resolutions on a single road section, further obtaining a plurality of time sequence data training samples by a sliding window method, extracting a most significant representative waveform feature subsequence from the obtained plurality of pieces of time sequence sample data of the single road section by adopting a time classifier algorithm based on shape similarity, and obtaining an effective waveform feature set of a speed curve of a traffic flow time sequence of the single road section. For example, the resulting set of temporal feature fragments may be defined as Ω = { ω i ,i=0,1,...,n},That is, n feature classes are obtained from time series data samples, where ω is i The vector value of the feature class is obtained for the extraction.
For a given time sequence section TS Δt And a time series data feature set omega using a shape distance D (TS) Δt ,ω i ) As time-sequential segments TS Δt Similarity measure P (Ω | TS) with all derived time-series feature classes Δt ) Taking the obtained similarity metric value as weight to realize the characteristic vector set omega of the characteristic space to the time sequence section TS Δt I.e. each time-sequential segment can be represented as a dot product of a different weight vector weight and a feature vector matrix. As shown in FIG. 2, for two consecutive time-sequential segmentsAndintroducing a directed graph structure G (n) = { V, E, M } to model the state transition relation of adjacent time sequence segments, wherein V represents a node set, E represents an edge set, and M represents a weight matrix, and specifically, obtaining a characteristic vector omega i As a single data point in the high-dimensional feature space, the time-series feature set Ω is used as a graph node set V of the dynamic graph structure, i.e. each node V in the graph corresponds to a feature vector ω i Edge e between nodes k-1,k For representing preamble time-sequential fragmentsWeight vector weight of k-1 Time-ordered fragmentCorresponding weight vector weight k The corresponding edge weight matrix is denoted as m k Since the weight calculation of each time sequence segment is independent, the edge weight matrix can be calculated by adopting independent event probability product. Thus, the speed time series conversion about a single road section can be obtainedAnd forming a dynamic graph structure.
Based on the dynamic graph structure of the traffic state time sequence evolution of the single road section, a graph neural network is adopted to extract the time sequence characteristics of the road section in the congestion state. Specifically, training sample data is constructed through the acquired track data and the corresponding road section congestion event records (for example, congestion road section marks observed through a bayonet video), and in a graph neural network, an obtained time sequence data feature set omega is defined as an initial state vector of each nodeThe information of a single picture is encoded as U i The method adopts a message transmission mechanism to aggregate local structure information of the graph, adopts a multilayer recurrent neural network mechanism to respectively propagate information of a graph node level and the whole graph level in time, and learns continuous time sequence segmentsToFinally, the feature characterization (h, U) of the single road section marked as the congestion state is obtained.
103, calculating a congestion propagation road section set caused by congestion of each road section in the congestion state according to the congestion state time sequence feature vector;
104, calculating to obtain a road network congestion space-time propagation map according to the congestion propagation road section set;
and 105, deducing a key propagation road section of the congestion space-time propagation chain and the congestion propagation probability of the key propagation road section according to the road network congestion space-time propagation map.
In this embodiment, based on the continuous state evolution of the traffic flow time series data and the link topology connectivity of the road, the congestion propagation probabilities of the key propagation road segment and the key propagation road segment of the time-space propagation link of the traffic congestion state are inferred, that is, how the congestion state of a certain road segment will reach a certain specific road segment at a certain moment and the duration time are obtained from the traffic flow time series data, so that the time-space continuous link-level traffic congestion propagation process is obtained.
In an implementation scenario, before step 101, the method for inferring traffic congestion further includes:
step 101 comprises:
step 1011, calculating time series data of the traffic flow of at least one road section according to the topological communication relation of the road network;
step 104 comprises:
and 1041, calculating to obtain a road network congestion space-time propagation map according to the road network topology connection relation and the congestion propagation road section set.
In one possible implementation, as shown in fig. 3, step 1011 includes:
step 1011-1, obtaining road network data of road network topological connectivity;
step 1011-2, acquiring historical track data of the target vehicle;
step 1011-3, cleaning and filtering the historical track data to obtain processed historical track data;
step 1011-4, establishing a spatial index for the processed historical track data according to the road network data, and performing map and road network matching on the indexed historical track data to obtain matched historical track data;
in this embodiment, a hidden markov model is used to perform map network matching on indexed historical track data.
And 1011-5, calculating the traffic flow time sequence data of at least one road section according to the matched historical track data.
In a specific implementation process, carrying out track matching on target vehicles (such as floating cars) based on road network data with topological connectivity (namely the road network data of the road network topological connectivity), and calculating traffic flow time series data of road sections according to the track data;
for example, taking a target vehicle as a floating vehicle as an example, specifically, historical track data of a GNSS (global navigation satellite system) of the floating vehicle is read, and the historical track data of a single floating vehicle (the historical track data includes, but is not limited to, a vehicle number, a timestamp, longitude and latitude, a driving state and a driving azimuth) is preprocessed, specifically, the preprocessing includes, but is not limited to, field cleaning and filtering (for example, effective motion track identification) processing, and the like. And (2) combining lane link level road network data with a topological connection relation (namely road network data of a road network topological connection relation), firstly establishing spatial index for historical track GNSS sequence point data, then performing map road network matching on the indexed historical track data by adopting a hidden Markov model, and smoothing the obtained spatial historical track data by adopting Gaussian filtering, thereby obtaining the historical track data of the floating car for reasonably calculating the motion parameters. And segmenting the complete single historical track data based on the road section, and establishing spatial connection between the segmented historical track data and the road section for calculating traffic flow parameters on the single road section. And calculating the motion tracks with different sampling frequencies on the single road section at the same moment according to the time stamps and the track points to obtain a track speed curve, and using a dynamic time warping algorithm to warp the speed curves aggregated on the single road section to finally obtain the time series data of the traffic flow of all the road sections of the road network, wherein the speed of all the road sections of the road network changes along with the time.
In one embodiment, as shown in fig. 4, step 103 comprises:
and 1032, acquiring a congestion propagation road section set caused by the congestion state of each road section according to the spatiotemporal causal relationship.
In this embodiment, based on the congestion state time sequence feature vector of a single road segment, a time sequence and causal combined relationship of congestion state evolution between the road segments is calculated (i.e., a spatiotemporal causal relationship of each road segment in a congestion state is calculated), and a potential influence propagation road segment set of each congestion road segment is obtained (i.e., a congestion propagation road segment set caused by each road segment in a congestion state is obtained);
specifically, examplesE.g. for a section r i And road section r j The obtained traffic state feature vector in the k time periodAndcalculation of road sections r within the same time window using transfer entropy TE i And road section r j The causal relationship of congestion states is determined by comparing the characteristics of asymmetry and non-aftereffect of transfer entropyAndto determine the direction of propagation of the congested road segment. Taking into account the propagation of the congestion state of the road section over space with a certain time delay, i.e. the road section r i Congestion during the kth time period may cause the section r j Congestion in the k + h time segment, therefore, adding a time lag variable h to the road segment r i And road section r j The cause-and-effect relationship of the congestion influence is calculated, namely a modified transfer entropy formula is adopted to calculate the road section r i For road section r j The spatiotemporal causal relationship that may generate congestion influence in the k-th to k + h-th time intervals is shown in formula (1):
wherein, a spectral clustering mode is adopted for the road section r i And road section r j State feature vector ofAndperforming feature classification, and estimating the path by using kernel densitySegment r i And road section r j Congestion state joint probability density over h time periods from kth to kth + hCalculating, and calculating conditional probability by using a Markov chain forward algorithm because the time-varying process of the traffic flow parameters meets Markov propertyAndby comparing all road segment pairs (e.g. road segment r) i And road section r j ) According to the congestion propagation direction of the road section, the congestion road section r is obtained in space i Potential impact congestion propagation road section set (result) Cr (caused) possibly caused in the k time period i As shown in fig. 5. Further, by stepping the sliding window, it is possible to obtain information about the section r in consideration of the fact that the traffic state on the section changes with time i The time sequence set of potentially affected congested road segments in the time duration series TS is Cr = { Cr = i ,i=1,2,...,m}。
According to the embodiment, the congestion state time sequence characteristic vector of the traffic flow time sequence data and the link topology connectivity of the road are combined, so that the influence of congestion on the road network along with the change of time can be better detected, and the spatiotemporal causal relationship among different congestion road sections can be better captured.
In one embodiment, as shown in fig. 6, step 1041 comprises:
step 1041-1, calculating to obtain a congestion propagation probability matrix according to the road network topological connection relation and the congestion propagation road section set;
and 1041-2, obtaining a road network congestion space-time propagation map according to the congestion propagation probability matrix.
In the concrete implementation process, after the time sequence set of the potential influence propagation road sections of the single road section is obtained, the congestion has spatial continuity in the propagation on the spaceAnd then, establishing a spatial adjacency relation among all road sections of the road network through the topological connectivity relation of the road network, and screening candidate congestion road sections in the congestion road section set with potential influence according to the constraint of the spatial adjacency relation. As shown in the road network diagram of fig. 5, for a single road section r i According to the potentially-affected road segment set (i.e. the congestion propagation road segment set caused by congestion of each road segment) C obtained in step 103 and the corresponding transfer entropy, the normalized transfer entropy is usedConsidered as a link r in the k-th time segment i For any road section r in the set j Congestion state propagation probability ofFor the road sections which are not in the candidate road section set, the congestion propagation probability is regarded as 0, so that a two-dimensional matrix P can be obtained for the whole road network k To represent the congestion propagation probability between networks in the kth time period. The method comprises the steps of considering first-order connectivity of congestion in spatial transfer, namely a congestion propagation chain is continuous in space, and meanwhile, considering the spatial connectivity relation due to the fact that a time-lag variable is introduced when transfer entropy is calculated, and introducing multi-order adjacency relation when a road section spatial adjacency matrix is constructed. For example, a spatial adjacent variable o may be taken, and the inverse of the euclidean distance between links may be used as a spatial propagation probability attenuation coefficient, i.e., the propagation probability obtained in the original stateMultiplying the space attenuation coefficient by the basis to obtain the new road section congestion propagation probability with space constraintFor a segment that is not in its o-th order neighbor set, its propagation probability is set to 0. Further, as shown in fig. 7, a graph structure with road network abstracted as road segments (nodes) and congestion propagation relations (edges) between road segments is provided, wherein the weight matrix of the edges is a new probability matrix P with the addition of space propagation attenuation constraint ′k Thus obtaining the product. Therefore, by introducing the time dimension t, a group of space-time congestion probability directed graphs (namely, network congestion space-time propagation graphs) TG = { TG (t) with potential propagation paths can be obtained k ,k=1,2,...,t}。
In one possible implementation, as shown in fig. 8, step 105 includes:
and 1053, obtaining the key propagation road section and the congestion propagation probability of the key propagation road section according to the congestion key space-time propagation chain.
In this embodiment, a statistical inference based on spatiotemporal causal relationship is used to perform significance test on the congestion propagation subtree, remove redundant spatiotemporal causal relationship, and obtain a key link and a propagation probability of a congestion propagation chain.
In a specific implementation process, the congestion probability directed graph tg in the kth time interval obtained in step 104 is used k (i.e. road network congestion space-time propagation map) for calculating congested road sections r i The corresponding graph nodes are correspondingly found in the congestion space-time propagation graph of the road networkAnd the strongly connected components where its graph nodes are locatedFirstly, according to breadth-first search in the strongly connected componentGet the graph nodeGenerating a plurality of subtrees as the road segments r for the root node i For the graph tg k In (1)All nodes can recursively obtain the candidate space propagation chain set of all road sections. Then stepping the sliding window to propagate the chain for the candidate spaceCongestion propagation chain for candidate space by further adopting statistical hypothesis testThe significance of the time-space causal relationship of the parent and child nodes of the subtree is judged, at the moment, a time component t is introduced, and the probability represented by the weight of the connecting edge of the parent and child nodes is obtainedTime series curve ofCongestion propagation probability curve for candidate road segmentsReconstructing the original time sequence into a null hypothesis of hypothesis test by adopting a method of disturbing the original time sequence, and deleting the affected road sections without significant time-space causal relationship by calculating the sample mean value and standard deviation of the time-space causal relationship strength of a plurality of reconstructed sequence samples, thereby realizing the candidate space congestion propagation chainA spatial chain reduction operation. For a road segment r in the road network i Obtaining the candidate space jam key propagation chain with the redundancy elimination space-time causal relation in the kth time periodMoving the time window by step length s, and obtaining the obtained road section r i Time-ordered collection of congestion critical propagation chain subtrees for root nodesAre combined to form a newAbout a section r i Congestion spatiotemporal propagation subgraphWherein the edge weight value between graph nodes is updated by adopting the maximum probability in the time sequence set of the key propagation chain sub-tree, and then the obtained key propagation chain sub-graph with the time-space causal relationship at the same time is updatedBy road section r i And the nodes are root nodes, the weight values among the nodes are probabilities, and the congestion propagation key road section with the maximum propagation probability sum is obtained by adopting a maximum spanning tree algorithm. As shown in fig. 9, the congestion probability directed graph tg obtained in step 104 is obtained k Each graph node in the road network is calculated, and a time window is stepped, so that the key propagation road section with the maximum propagation possibility of all the congested road sections in the whole road network can be obtainedAnd a dynamic process of growth and disappearance of its critical propagation path segments in space over time can be observed.
In the embodiment, the space propagation chain of the congestion state is identified by adopting the temporal-spatial causal relationship inference, so that the key propagation road section of the congestion space-time propagation chain with interpretability and the congestion propagation probability of the credible key propagation road section are obtained, the diffusion dissipation process of the traffic congestion of the real road section can be accurately inverted, and the temporal-spatial influence range of the traffic congestion is accurately quantized.
In one implementation, the traffic congestion estimation method further includes:
and 106, evaluating the key propagation road section of the congested space-time propagation chain to obtain an evaluation result.
In this embodiment, the validity of the result of the detected congestion spatiotemporal propagation chain is evaluated according to the congestion duration time of the structured queuing data record or the artificial mark extracted by the video portal as verification data.
The verification data and the historical track data are data of the same time and same road section.
In the embodiment, a time sequence dynamic graph is adopted to represent the congestion state characteristics of the traffic flow time sequence data of each road section of each time sequence section, the representation of the time sequence unstable characteristics under the congestion state is realized, a road network congestion space-time propagation graph is obtained by combining a congestion propagation road section set of each road section obtained by calculation based on a congestion state time sequence characteristic vector, then a key propagation road section of a congestion space-time propagation chain and the congestion propagation probability of the key propagation road section are deduced based on the network congestion space-time propagation graph, the diffusion dissipation process of the traffic congestion of a real road section can be accurately performed, and the space-time influence range of the traffic congestion is accurately quantized.
Example 2
The embodiment provides a traffic congestion inference system based on a time sequence dynamic diagram, as shown in fig. 10, the traffic congestion inference system includes a first calculation module 21, a characterization module 22, a second calculation module 23, a third calculation module 24, and an inference module 25;
the first calculation module 21 is used for calculating the traffic flow time series data of at least one road section;
the characterization module 22 is configured to characterize the congestion state feature of the traffic flow time series data of each road segment by using a time series dynamic graph, so as to obtain a congestion state time series feature vector;
in this embodiment, a time sequence dynamic graph (i.e., a dynamic graph structure) is used to model the dynamic evolution of the traffic flow state feature of each time sequence segment, so that the time correlation between the state of the time sequence segment and the state of the time span segment is maintained, and the characterization of the time sequence unstable feature in the congestion state is realized.
For the obtained speed time series data on the single road section, in order to obtain a representative time series segment with interpretability to carry out feature description on the traffic running state on the road, a dynamic graph structure is adopted to model the state change of adjacent time series segments, and the dynamic change rule of the traffic time series state along with the time is captured. Specifically, TS = { v ] for a time series t T =0,1,. Said, m }, whereinv t And representing a speed value at the t moment in the TS, firstly selecting different time intervals to perform up-sampling and down-sampling on sequence data to obtain a plurality of pieces of time sequence data with different time resolutions on a single road section, further obtaining a plurality of time sequence data training samples by a sliding window method, extracting a most significant representative waveform feature subsequence from the obtained plurality of pieces of time sequence sample data of the single road section by adopting a time classifier algorithm based on shape similarity, and obtaining an effective waveform feature set of a speed curve of a traffic flow time sequence of the single road section. For example, the resulting set of temporal feature fragments may be defined as Ω = { ω i I =0, 1.. Multidot.n }, i.e. n feature classes are derived from the time series data samples, where ω is i The vector value of the feature class is obtained for the extraction.
For a given time sequence section TS Δt And a time series data feature set omega using a shape distance D (TS) Δt ,ω i ) As time-sequential segments TS Δt Similarity measure P (Ω | TS) with all derived time-series feature classes Δt ) Taking the obtained similarity metric value as weight to realize the characteristic vector set omega of the characteristic space to the time sequence section TS Δt I.e. each time-sequential segment can be represented as a dot product of a different weight vector weight and a feature vector matrix. As shown in fig. 2, for two consecutive time-sequential segmentsAndintroducing a directed graph structure G (n) = { V, E, M } to model the state transition relation of adjacent time sequence segments, wherein V represents a node set, E represents an edge set, and M represents a weight matrix, and specifically, obtaining a characteristic vector omega i As a single data point in the high-dimensional feature space, the time-series feature set Ω is used as a graph node set V of the dynamic graph structure, i.e. each node V in the graph corresponds to a feature vector ω i Edge e between nodes k-1,k For before representationSequential time sliceWeight vector weight of k-1 Time-ordered fragmentCorresponding weight vector weight k The corresponding edge weight matrix is denoted as m k Since the weight calculation of each time sequence segment is independent, the edge weight matrix can be calculated by adopting independent event probability product. Thus, a dynamic graph structure into which a speed time series about a single road section is converted can be obtained.
Based on the dynamic graph structure of the traffic state time sequence evolution of the single road section, a graph neural network is adopted to extract the time sequence characteristics of the road section in the congestion state. Specifically, training sample data is constructed through the acquired track data and the corresponding road section congestion event record (for example, a congestion road section mark observed through a bayonet video), and in a graph neural network, an obtained time sequence data feature set omega is defined as an initial state vector of each nodeThe information of a single picture is encoded as U i The information of the local structure of the graph is aggregated by adopting a message transmission mechanism, the information of the node level of the graph and the information of the whole graph level are respectively propagated on time by adopting a multilayer recurrent neural network mechanism, and continuous time sequence segments are learnedToFinally, the feature characterization (h, U) of the single road section marked as the congestion state is obtained.
The second calculating module 23 is configured to calculate a congestion propagation road segment set caused by each road segment in the congestion state according to the congestion state time sequence feature vector;
the third calculation module 24 is used for calculating a road network congestion space-time propagation map according to the congestion propagation road section set;
the inference module 25 is configured to infer a key propagation road segment of the congestion spatio-temporal propagation chain and a congestion propagation probability of the key propagation road segment according to the road network congestion spatio-temporal propagation map.
In this embodiment, based on the continuous state evolution of the traffic flow time series data and the link topology connectivity of the road, the congestion propagation probabilities of the key propagation road segment and the key propagation road segment of the time-space propagation link of the traffic congestion state are inferred, that is, how the congestion state of a certain road segment will reach a certain specific road segment at a certain moment and the duration time are obtained from the traffic flow time series data, so that the time-space continuous link-level traffic congestion propagation process is obtained.
In one possible implementation, as shown in fig. 10, the traffic congestion estimation system further includes an acquisition module 26;
the obtaining module 26 is configured to obtain a road network topology connection relationship;
the first calculation module 21 is configured to calculate time series data of traffic flow of at least one road segment according to the topological connectivity relationship of the road network;
the third calculating module 24 is configured to calculate a road network congestion space-time propagation map according to the road network topology connectivity relation and the congestion propagation road section set.
In an implementable scenario, as shown in fig. 10, the first calculation module 21 includes a first obtaining unit 211, a second obtaining unit 212, a processing unit 213, a matching unit 214, and a first calculation unit 215;
the first obtaining unit 211 is configured to obtain road network data of a road network topology connectivity relationship;
the second acquisition unit 212 is used for acquiring historical track data of the target vehicle;
the processing unit 213 is configured to perform cleaning and filtering processing on the historical track data to obtain processed historical track data;
the matching unit 214 is configured to establish a spatial index for the processed historical track data according to the road network data, and perform map-road network matching on the indexed historical track data to obtain matched historical track data;
in this embodiment, a hidden markov model is used to perform map network matching on indexed historical track data.
The first calculating unit 215 is configured to calculate traffic flow time-series data of at least one road segment according to the matched historical track data.
In a specific implementation process, carrying out track matching on target vehicles (such as floating cars) based on road network data with topological connectivity (namely the road network data of the road network topological connectivity), and calculating traffic flow time series data of road sections according to the track data;
for example, taking the target vehicle as a floating vehicle as an example, specifically, the GNSS historical trajectory data of the floating vehicle is read, and the historical trajectory data of the single floating vehicle (the historical trajectory data includes, but is not limited to, a vehicle number, a timestamp, longitude and latitude, a driving state, and a driving azimuth) is preprocessed, specifically, the preprocessing includes, but is not limited to, field cleaning and filtering (for example, effective motion trajectory identification) processing, and the like. And (2) combining lane link level road network data with a topological connection relation (namely road network data of a road network topological connection relation), firstly establishing spatial index for historical track GNSS sequence point data, then performing map road network matching on the indexed historical track data by adopting a hidden Markov model, and smoothing the obtained spatial historical track data by adopting Gaussian filtering, thereby obtaining the historical track data of the floating car for reasonably calculating the motion parameters. And segmenting the complete single historical track data based on the road section, and establishing spatial connection between the segmented historical track data and the road section for calculating traffic flow parameters on the single road section. And calculating the motion tracks with different sampling frequencies on the single road section at the same moment according to the time stamps and the track points to obtain a track speed curve, and using a dynamic time warping algorithm to warp the speed curves aggregated on the single road section to finally obtain the time series data of the traffic flow of all the road sections of the road network, wherein the speed of all the road sections of the road network changes along with the time.
In an implementation scenario, as shown in fig. 10, the second calculating module 23 includes a second calculating unit 231 and a third obtaining unit 232;
the second calculating unit 231 is used for calculating a spatiotemporal causal relationship of each road section in the congestion state according to the congestion state time sequence feature vector;
the third obtaining unit 232 is configured to obtain a congestion propagation road segment set caused by each road segment in a congestion state according to a spatiotemporal causal relationship.
In this embodiment, based on the congestion state time sequence feature vector of a single road segment, a time sequence and causal combined relationship of congestion state evolution between the road segments is calculated (i.e., a spatiotemporal causal relationship of each road segment in a congestion state is calculated), and a potential influence propagation road segment set of each congestion road segment is obtained (i.e., a congestion propagation road segment set caused by each road segment in a congestion state is obtained);
specifically, for example, for the section r i And road section r j The obtained traffic state feature vector in the kth time periodAndcalculation of road sections r within the same time window using transfer entropy TE i And road section r j The causal relationship of congestion states is determined by comparing the characteristics of asymmetry and non-aftereffect of transfer entropyAndto determine the direction of propagation of the congested road segment. Taking into account the propagation of the congestion state of the road section over the space with a certain time delay, i.e. the road section r i Congestion during the kth time period may cause the section r j Congestion in the k + h time segment, therefore, adding a time lag variable h to the road segment r i And road section r j Cause and effect correlation of congestion impactThe relation is calculated, i.e. the corrected transmission entropy formula is adopted to calculate the road section r i For road section r j The spatiotemporal causal relationship that may produce congestion effects in the k-th to k + h-th time intervals is shown in equation (1) in example 1:
wherein, a spectral clustering mode is adopted for the road section r i And road section r j State feature vector ofAndperforming feature classification, and estimating the r road section by using the kernel density i And road section r j Congestion state joint probability density over h time periods from kth to kth + hCalculating, and calculating conditional probability by using a Markov chain forward algorithm because the time-varying process of the traffic flow parameters meets Markov propertyAndby comparing all road segment pairs (e.g. road segment r) i And road section r j ) According to the congestion propagation direction of the road section, the congestion road section r is obtained in space i Potential impact congestion propagation road section set (result) Cr (caused) possibly caused in kth time period i As shown in fig. 5. Further, by stepping the sliding window, it is possible to obtain information about the section r in consideration of the fact that the traffic state on the section changes with time i The time sequence set of potentially affected congested road segments in the time duration series TS is Cr = { Cr = i ,i=1,2,...,m}。
According to the embodiment, the congestion state time sequence characteristic vector of the traffic flow time sequence data and the link topology connectivity of the road are combined, so that the influence of congestion on the road network along with the change of time can be better detected, and the spatiotemporal causal relationship among different congestion road sections can be better captured.
In an implementation scenario, as shown in fig. 10, the third calculating module 24 includes a third calculating unit 241 and a fourth obtaining unit 242;
the third calculating unit 241 is configured to calculate a congestion propagation probability matrix according to the road network topology connectivity relation and the congestion propagation road section set;
the fourth obtaining unit 242 is configured to obtain a road network congestion space-time propagation map according to the congestion propagation probability matrix.
In a specific implementation process, after a time sequence set of potential influence propagation road sections of a single road section is obtained, due to the fact that congestion propagates in the space continuously, a spatial adjacency relation between all road sections of a road network is established through a topological connection relation of the road network, and candidate congestion road sections in the set of potential influence congestion road sections are screened according to constraints of the spatial adjacency relation. As shown in the road network diagram of fig. 5, for a single road section r i According to the obtained potential influence road section set (namely the congestion propagation road section set caused by the congestion state of each road section) C and the corresponding transfer entropy, the normalized transfer entropy is obtainedConsider the road section r in the kth time period i For any road section r in the set j Probability of propagation of congestion stateFor the road sections which are not in the candidate road section set, the congestion propagation probability is regarded as 0, so that a two-dimensional matrix P can be obtained for the whole road network k To represent the probability of congestion propagation between the networks during the kth time period. The method comprises the steps of considering first-order connectivity of congestion in spatial transfer, namely a congestion propagation chain is continuous in space, and meanwhile, considering the spatial connectivity relation due to the fact that a time-lag variable is introduced when transfer entropy is calculated, and introducing multi-order adjacency relation when a road section spatial adjacency matrix is constructed. For example, can take spaceThe adjacent variable o takes the reciprocal of the Euclidean distance between the road sections as the space propagation probability attenuation coefficient, namely the propagation probability obtained originallyMultiplying the space attenuation coefficient by the basis to obtain the new road section congestion propagation probability with space constraintFor a segment that is not in its o-th order neighbor set, its propagation probability is set to 0. Further, as shown in fig. 7, a graph structure with road network abstracted as road segments (nodes) and congestion propagation relations (edges) between road segments is provided, wherein the weight matrix of the edges is composed of a new probability matrix p added with space propagation attenuation constraint ′k Thus obtaining the product. Therefore, by introducing the time dimension t, a group of space-time congestion probability directed graphs (namely, network congestion space-time propagation graphs) TG = { TG (t) with potential propagation paths can be obtained k ,k=1,2,...,t}。
In an implementation scenario, as shown in fig. 10, the inference module 25 includes a fourth calculation unit 251, a verification unit 252, and a fifth acquisition unit 253;
the fourth calculating unit 251 is configured to calculate a congestion space-time propagation chain according to the road network congestion space-time propagation map;
the inspection unit 252 is configured to inspect the congestion spatio-temporal propagation chain according to the spatio-temporal causal relationship to obtain a congestion key spatio-temporal propagation chain with the redundant spatio-temporal causal relationship removed;
the fifth obtaining unit 253 is configured to obtain the key propagation link and the congestion propagation probability of the key propagation link according to the congestion key spatiotemporal propagation chain.
In this embodiment, a statistical inference based on spatiotemporal causal relationship is used to perform significance test on the congestion propagation subtree, remove redundant spatiotemporal causal relationship, and obtain a key link and a propagation probability of a congestion propagation chain.
In a specific implementation process, the obtained congestion probability directed graph tg in the kth time interval k (i.e., road network congestion space-time propagation map) for calculating congested road sections r i The congestion propagation chain of (2), namely correspondingly finding the corresponding graph node in the congestion space-time propagation graph of the road networkAnd the strongly connected components where its graph nodes are locatedFirstly, the strongly connected component is searched according to breadth firstGet the graph nodeGenerating a plurality of subtrees as links r for a root node i For the graph tg k All the nodes in the set can recursively obtain the candidate space propagation chain set of all the road sections. Then stepping the sliding window to propagate the chain for the candidate spaceCongestion propagation chain for candidate space by further adopting statistical hypothesis testThe significance of the time-space causal relationship of the parent and child nodes of the subtree is judged, at the moment, a time component t is introduced, and the probability represented by the weight of the connecting edge of the parent and child nodes is obtainedTime sequence curve ofCongestion propagation probability curve for candidate road segmentsReconstructing a null hypothesis for hypothesis testing by a method of scrambling an original time sequence, and calculating a space-time causal effect of a plurality of reconstructed sequence samplesDeleting the affected road sections without significant time-space causal relation by using the sample mean value and standard deviation of the relation strength, thereby realizing the candidate space congestion propagation chainA spatial chain reduction operation. For a road segment r in the road network i Obtaining the candidate space jam key propagation chain with the redundancy elimination space-time causal relation in the kth time periodMoving the time window by step length s, and obtaining the obtained road section r i Time-ordered collection of congestion critical propagation chain subtrees for root nodesMerge to form a new relevant road section r i Congestion spatiotemporal propagation subgraphWherein the edge weight value between graph nodes is updated by adopting the maximum probability in the time sequence set of the key propagation chain sub-tree, and then the obtained key propagation chain sub-graph with the time-space causal relationship at the same time is updatedBy road section r i And the nodes are root nodes, the weight values among the nodes are probabilities, and the congestion propagation key road section with the maximum propagation probability sum is obtained by adopting a maximum spanning tree algorithm. As shown in fig. 9, the congestion probability directed graph tg obtained as described above is used k Each graph node in the road network is calculated, and a time window is stepped, so that the key propagation road section with the maximum propagation possibility of all the jammed road sections in the whole road network can be obtainedAnd a dynamic process of growth and disappearance of its critical propagation path segments in space over time can be observed.
In the embodiment, the space propagation chain of the congestion state is identified by adopting the temporal-spatial causal relationship inference, so that the key propagation road section of the congestion space-time propagation chain with interpretability and the congestion propagation probability of the credible key propagation road section are obtained, the diffusion dissipation process of the traffic congestion of the real road section can be accurately inverted, and the temporal-spatial influence range of the traffic congestion is accurately quantized.
In one implementation, as shown in fig. 10, the traffic congestion inference system further includes an evaluation module 27;
the evaluation module 27 is configured to evaluate a critical propagation segment of the congested spatio-temporal propagation chain to obtain an evaluation result.
In this embodiment, the validity of the result of the detected congestion spatiotemporal propagation chain is evaluated according to the congestion duration time of the structured queuing data record or the artificial mark extracted by the video portal as verification data.
The verification data and the historical track data are data of the same time and same road section.
In the embodiment, a time sequence dynamic graph is adopted to represent the congestion state characteristics of the traffic flow time sequence data of each road section of each time sequence section, the representation of the time sequence unstable characteristics in the congestion state is realized, a road network congestion space-time propagation graph is obtained by combining the congestion propagation road section set of each road section calculated based on the congestion state time sequence characteristic vector, the key propagation road section of a congestion space-time propagation chain and the congestion propagation probability of the key propagation road section are deduced based on the network congestion space-time propagation graph, the diffusion dissipation process of the traffic congestion of the real road section can be accurately performed, and the space-time influence range of the traffic congestion is accurately quantized.
Example 3
Fig. 11 is a schematic structural diagram of an electronic device according to embodiment 3 of the present invention. The electronic device comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to implement the traffic congestion inference method based on the time sequence dynamic graph in embodiment 1. The electronic device 30 shown in fig. 11 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiment of the present invention.
As shown in fig. 11, the electronic device 30 may be embodied in the form of a general purpose computing device, which may be, for example, a server device. The components of the electronic device 30 may include, but are not limited to: the at least one processor 31, the at least one memory 32, and a bus 33 connecting the various system components (including the memory 32 and the processor 31).
The bus 33 includes a data bus, an address bus, and a control bus.
The memory 32 may include volatile memory, such as Random Access Memory (RAM) 321 and/or cache memory 322, and may further include Read Only Memory (ROM) 323.
The processor 31 executes various functional applications and data processing, such as the traffic congestion estimation method based on the time-series dynamic graph in embodiment 1 of the present invention, by executing the computer program stored in the memory 32.
The electronic device 30 may also communicate with one or more external devices 34 (e.g., keyboard, pointing device, etc.). Such communication may be through an input/output (I/O) interface 35. Also, model-generating device 30 may also communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via network adapter 36. As shown in FIG. 11, network adapter 36 communicates with the other modules of model-generating device 30 via bus 33. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the model-generating device 30, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems, etc.
In addition, the electronic device may also be implemented in the form of an electronic chip, on which a memory, processor-related electronic components, and an operating program stored on the memory and executable on the processor are provided.
It should be noted that although in the above detailed description several units/modules or sub-units/modules of the electronic device are mentioned, such a division is merely exemplary and not mandatory. Indeed, the features and functionality of two or more of the units/modules described above may be embodied in one unit/module according to embodiments of the invention. Conversely, the features and functions of one unit/module described above may be further divided into embodiments by a plurality of units/modules.
Example 4
The present embodiment provides a computer-readable storage medium on which a computer program is stored, which when executed by a processor, implements the time-series dynamic graph-based traffic congestion inference method of embodiment 1.
More specific examples, among others, that the readable storage medium may employ may include, but are not limited to: a portable disk, a hard disk, random access memory, read only memory, erasable programmable read only memory, optical storage device, magnetic storage device, or any suitable combination of the foregoing.
In a possible implementation, the invention may also be implemented in the form of a program product comprising program code for causing a terminal device to perform a method for implementing a time series dynamic graph-based traffic congestion inference method according to example 1, when the program product is run on the terminal device.
Where program code for carrying out the invention is written in any combination of one or more programming languages, the program code may execute entirely on the user's device, partly on the user's device, as a stand-alone software package, partly on the user's device, partly on a remote device or entirely on the remote device.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.
Claims (10)
1. A traffic jam inference method based on a time sequence dynamic graph is characterized by comprising the following steps:
calculating traffic flow time series data of at least one road section;
characterizing the congestion state characteristics of the traffic flow time sequence data of each road section by adopting a time sequence dynamic graph to obtain a congestion state time sequence characteristic vector;
calculating a congestion propagation road section set caused by each road section in a congestion state according to the congestion state time sequence feature vector;
calculating to obtain a road network congestion space-time propagation map according to the congestion propagation road section set;
and deducing a key propagation road section of a congestion space-time propagation chain and the congestion propagation probability of the key propagation road section according to the road network congestion space-time propagation map.
2. The time-series dynamic graph-based traffic congestion inference method according to claim 1, wherein, before the step of calculating the traffic-flow time-series data of at least one link, the traffic congestion inference method further comprises:
acquiring a road network topology communication relation;
the step of calculating the traffic flow time-series data of at least one link includes:
calculating the traffic flow time sequence data of at least one road section according to the road network topological connection relation;
the step of calculating the road network congestion space-time propagation map according to the congestion propagation road section set comprises the following steps:
and calculating to obtain a road network congestion space-time propagation map according to the road network topology communication relation and the congestion propagation road section set.
3. The method for traffic congestion estimation according to claim 2, wherein the step of calculating time series data of traffic flow of at least one road segment according to the topological connectivity of the road network comprises:
acquiring road network data of road network topological connectivity;
acquiring historical track data of a target vehicle;
cleaning and filtering the historical track data to obtain processed historical track data;
establishing a spatial index for the processed historical track data according to the road network data, and performing map road network matching on the indexed historical track data to obtain matched historical track data;
and calculating the traffic flow time sequence data of at least one road section according to the matched historical track data.
4. The method for traffic congestion estimation based on time series dynamic graph as claimed in claim 1, wherein said step of calculating a set of congestion propagation road segments caused by congestion state of each road segment according to the congestion state time series feature vector comprises:
calculating a time-space causal relationship of each road section in the congestion state according to the congestion state time sequence feature vector;
and acquiring a congestion propagation road section set caused by the congestion state of each road section according to the spatiotemporal causal relationship.
5. The method for traffic congestion inference based on time-series dynamic graph as claimed in claim 2, wherein said step of calculating a road network congestion spatiotemporal propagation graph according to said road network topological connectivity relation and said set of congestion propagation road segments comprises:
calculating to obtain a congestion propagation probability matrix according to the road network topological communication relation and the congestion propagation road section set;
and obtaining the road network congestion space-time propagation map according to the congestion propagation probability matrix.
6. The method as claimed in claim 4, wherein the step of deducing a key propagation segment of a congestion spatiotemporal propagation chain and a congestion propagation probability of the key propagation segment according to the road network congestion spatiotemporal propagation map comprises:
calculating to obtain a congestion space-time propagation chain according to the road network congestion space-time propagation diagram;
checking the congestion space-time propagation chain according to the space-time causal relationship to obtain a congestion key space-time propagation chain with the redundant space-time causal relationship removed;
and acquiring the key propagation road section and the congestion propagation probability of the key propagation road section according to the congestion key space-time propagation chain.
7. The time-series dynamic graph-based traffic congestion inference method of claim 1, wherein after said step of inferring a key propagation segment of a congestion spatiotemporal propagation chain and a congestion propagation probability for the key propagation segment from the road network congestion spatiotemporal propagation map, the traffic congestion inference method further comprises:
and evaluating the key propagation section of the congested space-time propagation chain to obtain an evaluation result.
8. A traffic jam inference system based on a time sequence dynamic graph is characterized by comprising a first calculation module, a characterization module, a second calculation module, a third calculation module and an inference module;
the first calculation module is used for calculating the traffic flow time sequence data of at least one road section;
the characterization module is used for characterizing the congestion state characteristics of the traffic flow time sequence data of each road section by adopting a time sequence dynamic graph so as to obtain a congestion state time sequence characteristic vector;
the second calculation module is used for calculating a congestion propagation road section set caused by each road section in a congestion state according to the congestion state time sequence feature vector;
the third calculation module is used for calculating a road network congestion space-time propagation map according to the congestion propagation road section set;
the inference module is used for inferring a key propagation road section of a congestion space-time propagation chain and the congestion propagation probability of the key propagation road section according to the road network congestion space-time propagation diagram.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method for traffic congestion inference based on a time series dynamic graph according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a method for traffic congestion inference based on a time-series dynamic graph according to any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211447815.5A CN115830866A (en) | 2022-11-18 | 2022-11-18 | Traffic jam inference method, system, device and medium based on time sequence dynamic graph |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211447815.5A CN115830866A (en) | 2022-11-18 | 2022-11-18 | Traffic jam inference method, system, device and medium based on time sequence dynamic graph |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115830866A true CN115830866A (en) | 2023-03-21 |
Family
ID=85529221
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211447815.5A Pending CN115830866A (en) | 2022-11-18 | 2022-11-18 | Traffic jam inference method, system, device and medium based on time sequence dynamic graph |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115830866A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116151493A (en) * | 2023-04-24 | 2023-05-23 | 之江实验室 | Traffic jam prediction method and device based on head effect and cyclic neural network |
CN117037465A (en) * | 2023-05-24 | 2023-11-10 | 东北师范大学 | Traffic jam propagation mode sensing and visual analysis method |
-
2022
- 2022-11-18 CN CN202211447815.5A patent/CN115830866A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116151493A (en) * | 2023-04-24 | 2023-05-23 | 之江实验室 | Traffic jam prediction method and device based on head effect and cyclic neural network |
CN117037465A (en) * | 2023-05-24 | 2023-11-10 | 东北师范大学 | Traffic jam propagation mode sensing and visual analysis method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Djenouri et al. | Adapted k-nearest neighbors for detecting anomalies on spatio–temporal traffic flow | |
CN110827544B (en) | Short-term traffic flow control method based on graph convolution recurrent neural network | |
CN115830866A (en) | Traffic jam inference method, system, device and medium based on time sequence dynamic graph | |
Zhang et al. | A generative adversarial network for travel times imputation using trajectory data | |
US7296018B2 (en) | Resource-light method and apparatus for outlier detection | |
JP6525002B2 (en) | Maintenance time determination apparatus, deterioration prediction system, deterioration prediction method, and recording medium | |
Su et al. | Uncertainty quantification of collaborative detection for self-driving | |
US20220405682A1 (en) | Inverse reinforcement learning-based delivery means detection apparatus and method | |
Liang et al. | Dynamic spatiotemporal graph convolutional neural networks for traffic data imputation with complex missing patterns | |
CN111158964A (en) | Disk failure prediction method, system, device and storage medium | |
Zhang et al. | Robust KPI anomaly detection for large-scale software services with partial labels | |
James | Citywide estimation of travel time distributions with Bayesian deep graph learning | |
Peña et al. | A novel imputation method for missing values in air pollutant time series data | |
Basak et al. | Data-driven detection of anomalies and cascading failures in traffic networks | |
Bogdanov et al. | Sktr: Trace recovery from stochastically known logs | |
Ailliot et al. | Dynamical partitioning of directional ocean wave spectra | |
US20240012875A1 (en) | Method, system and computer readable medium for probabilistic spatiotemporal forecasting | |
Qiao et al. | Topology-aware robust optimization for out-of-distribution generalization | |
Berkani et al. | Spatio-temporal forecasting: A survey of data-driven models using exogenous data | |
Lorbeer et al. | Anomaly detection with hmm gauge likelihood analysis | |
Bhatt et al. | Online variational Bayesian subspace filtering | |
Wan et al. | Efficient data association in visual sensor networks with missing detection | |
Yang et al. | Efficient hidden trajectory reconstruction from sparse data | |
Parlitz et al. | Identification of equivalent dynamics using ordinal pattern distributions | |
Phung et al. | Unsupervised Air Quality Interpolation with Attentive Graph Neural Network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB02 | Change of applicant information |
Address after: 200092 Siping Road 1239, Shanghai, Yangpu District Applicant after: TONGJI University Applicant after: Yuance Information Technology Co.,Ltd. Address before: 200092 Siping Road 1239, Shanghai, Yangpu District Applicant before: TONGJI University Applicant before: SUZHOU INDUSTRIAL PARK SURVEYING MAPPING AND GEOINFORMATION Co.,Ltd. |
|
CB02 | Change of applicant information |