CN115984077A - Causal detection method and device for abnormal traffic flow - Google Patents
Causal detection method and device for abnormal traffic flow Download PDFInfo
- Publication number
- CN115984077A CN115984077A CN202310160350.3A CN202310160350A CN115984077A CN 115984077 A CN115984077 A CN 115984077A CN 202310160350 A CN202310160350 A CN 202310160350A CN 115984077 A CN115984077 A CN 115984077A
- Authority
- CN
- China
- Prior art keywords
- flow
- abnormal
- traffic
- time
- links
- 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.)
- Granted
Links
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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
- Y02A30/60—Planning or developing urban green infrastructure
Abstract
The invention discloses a causal detection method and equipment for abnormal traffic flow, which are characterized in that after an urban area is subjected to area segmentation to obtain an urban area node map, links are created between nodes according to the obtained urban flow data, the monitoring of urban area flow change on the flow change of other areas in the city is realized through the links, and a space-time abnormal value is obtained through calculating the distance value of the links, so that an abnormal causal map is obtained through calculating the space-time abnormal value through an abnormal causal fruit tree algorithm, the normal state causal map under the normal flow state is combined, the normal state and the abnormal lower flow under the abnormal state are distinguished, the abnormal change characteristic of the traffic flow caused by the space-time abnormal value can be captured, the causal detection of the abnormal value and the transitivity along with the time change are realized, the potential interaction between different urban areas and roads is disclosed, the effective perception of the urban traffic state is promoted, and the more accurate decision is facilitated.
Description
Technical Field
The invention relates to the technical field of intelligent traffic, in particular to a causal detection method and device for abnormal traffic flow.
Background
The transportation trip is a core part in the development and construction of the smart city. As an innovative travel mode, the intelligent travel utilizes advanced technology fusion such as artificial intelligence, internet of things, space perception, big data and cloud computing, and accordingly all-around traffic construction management and whole traffic construction management processes are conducted on the traffic fields such as urban traffic management, transportation and people travel. The traffic exit system has excellent capacities of internet of things, perception, interconnection, monitoring, early warning, prevention and control and the like in even wider space ranges of regions and cities. Plays a great role in guaranteeing the safety of the trips, improving the high efficiency of the trips and improving the comprehensive management capability. Traffic is considered one of the most promising applications in all application scenarios of the internet of things.
With the development of urbanization, traffic problems become more serious, so that the traditional solution cannot meet new traffic problems, and therefore intelligent traffic is produced at the same time. The intelligent transportation means that advanced information technology, data transmission technology, computer processing technology and the like are effectively integrated into a transportation management system, so that people, vehicles and roads can be closely matched, the transportation environment is improved, the resource utilization rate is improved, and the like.
The anomaly detection and analysis can solve the urban scale problems of traffic, public safety, crime prevention, efficient resource utilization and the like, and improve the user experience of large public and private spaces. Outlier detection is also known as anomaly detection and refers to detecting data that deviates significantly from most instances of data. Its increasing demand and use in areas such as risk management, compliance, security, financial surveillance, health and medical risk, and artificial intelligence, where outlier detection plays an increasingly important role.
The increasing popularity of positioning technologies, including GPS and WIFI, has produced a large amount of spatio-temporal data, one of the main forms of which is spatio-temporal data in the form of trajectory data. Abnormal patterns of trajectory data generated by a large number of moving objects can generally reflect abnormal traffic flow patterns on a traffic road network, which may be caused by non-periodic events such as celebrations, large commercial promotions, temporary traffic controls, special large sporting events, and the like, or periodic traffic congestion due to a certain degree of unreasonable planning of traffic roads. Therefore, the abnormal value detection from the track data helps managers to sense abnormal events in time and reduce the abnormal influence of the abnormal events on traffic flow.
For the current outlier detection technology, the detection method includes Principal Component Analysis (PCA) and its variants, DPMM (DirichletProcess mix Model), and deep neural network Model. Principal component analysis and variants thereof are widely applied to detect outliers from spatial and temporal data, while DPMM focuses more on the study of outlier detection in traffic data. Meanwhile, due to the strong learning capability of the deep neural network model, researchers also have the field of using the deep neural network model for abnormal value detection.
Principal Component Analysis (PCA) and its variants are more popular methods of data dimension reduction and anomaly detection. The main idea of principal component analysis is to convert linearly correlated data into linearly uncorrelated data using orthogonal transformation. PCA is applied to a link time matrix that displays the traffic volumes on different roads in a time window to detect the root cause of abnormal traffic behavior.
The DPMM may be used to derive outliers in the city traffic flow data. First, a set of all flow values is projected into an n-dimensional space. The dimensionality of the data is then reduced to a two-dimensional space by Principal Component Analysis (PCA). Then, the flow value is clustered into a plurality of classes according to the Chinese restaurant process (Chinese restaurant process). Each flow value is assigned to a new cluster with a probability proportional to the quantity parameter a, otherwise it will be assigned to the previously created cluster. Thereafter, all flow values belonging to the cluster having the largest number of elements are regarded as internal values, and the remaining flow values are regarded as outliers, i.e., outliers.
The method for detecting the abnormal value by the deep neural network is as follows: researchers have trained three independent networks of auto-encoders to learn three different features of video anomaly detection: appearance, motion, and appearance-motion combination features. A classifier integrated by three support vector machines is trained independently on each learned feature representation to perform anomaly scoring. First, outlier detection is performed on low-dimensional feature representations of high-dimensional raw data generated by deep neural networks using a Linear support vector machine (Linear SVM). After further optimization, an unsupervised classification method is used instead of SVM (SupportVector Machine) to achieve anomaly scoring in projection space. In this technique, they first cluster the low-dimensional features of the video frames generated by the convolution-based auto-encoder and then classify the cluster labels as pseudo-labels. The classification probabilities are used to calculate an anomaly score for each frame.
For causal relationship testing techniques, the traditional analytical causal relationship testing method is control variable testing, such as a/B testing, which is widely used in industry. The a/B test is a traditional causal analysis test that randomly assigns groups of variables to specific univariate treatment levels, compares the performance of one or more test groups to the control groups, and finally performs the test.
When the above method has the following disadvantages:
disadvantage 1: in terms of outlier detection, the results of Principal Component Analysis (PCA) are very sensitive to parameter settings that are highly data dependent. In some cases, extreme outliers can adversely affect the computation of PCA, resulting in false positives and false negatives of outliers. At the same time, the above-described techniques lack causal testing for outliers and transitivity over time.
And (2) disadvantage: the current space-time cause and effect detection methods are focused on discovering cause and effect relationships in abnormal situations, and some try to discover cause and effect relationships among general traffic roads from data, but do not distinguish road cause and effect graphs with different conditions in normal situations and abnormal situations.
Disadvantage 3: in the aspect of causal relationship verification, the traditional a/B test needs to obtain different test feedbacks of the same user or sample for different variable conditions to determine causal relationship between variables. In practice, it is very difficult to obtain data of the same sample under different conditions, and taking traffic data as an example, it is usually only possible to obtain the flow volume of outflow and inflow in a certain area at a certain time, but it is not possible to obtain the influence of the flow volume of outflow in several areas becoming larger or smaller at the same time on the flow volume of other areas. Meanwhile, the causal test relationship is simple correlation, but stronger correlation, so that the reliability of A/B detection is limited in reuse.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: a cause and effect detection method and equipment for traffic abnormal flow are disclosed, wherein a more accurate cause and effect relationship graph is established, so that the current road traffic mode can be better understood and a more timely coping strategy can be made;
in order to solve the technical problems, the invention adopts the technical scheme that:
a causal detection method for traffic abnormal flow comprises the following steps:
carrying out region segmentation on the urban region to obtain an urban region node map;
acquiring flow data, and constructing links between nodes according to the urban area node graph and the flow data;
calculating the distance values of all the links, and obtaining a space-time abnormal value according to the distance values;
judging the relation between the space-time abnormal values according to an abnormal cause-and-effect tree algorithm to obtain an abnormal cause-and-effect graph;
and obtaining a normal cause-and-effect graph according to the flow city region node graph and the relationship between the flow data judgment nodes and the nodes.
In order to solve the technical problem, the invention adopts another technical scheme as follows:
a causal detection device for traffic abnormal flow comprises a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the causal detection method for traffic abnormal flow.
The invention has the beneficial effects that: after an urban area node map is obtained by carrying out area segmentation on an urban area, a link is created between nodes according to obtained urban traffic data, monitoring of traffic change of other areas of the city by the urban area traffic change is realized through the link, a space-time abnormal value is obtained by calculating a distance value of the link, an abnormal causal map is obtained by calculating the space-time abnormal value through an abnormal cause-fruit algorithm, the lower traffic of a normal state and an abnormal state is distinguished by combining a normal causal map under the normal state of the traffic, abnormal change characteristics of the traffic flow caused by the space-time abnormal value can be captured, causal inspection of the abnormal value and transitivity along with time change are realized, potential interaction between different urban areas and roads is disclosed, effective perception of the urban traffic state is promoted, and more accurate decision is facilitated.
Drawings
FIG. 1 is a flow chart illustrating the steps of a causal detection method for abnormal traffic flow in an embodiment of the present invention;
FIG. 2 is a schematic diagram of a logic structure of a causal detection method for abnormal traffic flow according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a causal detection device for traffic abnormal flow in an embodiment of the present invention.
Detailed Description
In order to explain technical contents, achieved objects, and effects of the present invention in detail, the following description is made with reference to the accompanying drawings in combination with the embodiments.
Referring to fig. 1, a causal detection method for abnormal traffic flow includes the steps of:
carrying out region segmentation on the urban region to obtain an urban region node map;
acquiring flow data, and constructing links among nodes according to the urban area node graph and the flow data;
calculating the distance values of all the links, and obtaining a space-time abnormal value according to the distance values;
judging the relation between the space-time abnormal values according to an abnormal cause-and-effect tree algorithm to obtain an abnormal cause-and-effect graph;
and obtaining a normal cause-and-effect graph according to the flow city region node graph and the relationship between the flow data judgment nodes and the nodes.
As can be seen from the above description, the beneficial effects of the present invention are: after an urban area node map is obtained by carrying out area segmentation on an urban area, a link is created between nodes according to obtained urban traffic data, monitoring of traffic change of other areas of the city by the urban area traffic change is realized through the link, a space-time abnormal value is obtained by calculating a distance value of the link, an abnormal causal map is obtained by calculating the space-time abnormal value through an abnormal cause-fruit algorithm, the lower traffic of a normal state and an abnormal state is distinguished by combining a normal causal map under the normal state of the traffic, abnormal change characteristics of the traffic flow caused by the space-time abnormal value can be captured, causal inspection of the abnormal value and transitivity along with time change are realized, potential interaction between different urban areas and roads is disclosed, effective perception of the urban traffic state is promoted, and more accurate decision is facilitated.
Further, the performing region segmentation on the urban region to obtain the urban region node map includes:
dividing an urban area by adopting a regular hexagon to obtain an urban area node map;
each of the nodes represents an area;
the connecting lines between the nodes represent the traffic from region to region.
As can be seen from the above description, the cellular hexagon is used to divide the urban area into a plurality of areas to construct an area map, the nodes in the map represent the corresponding divided areas, and the edges of the hexagon represent the traffic between two areas, so that the traffic change of each node in the urban area and the traffic change between nodes can be accurately described.
Further, the constructing links between nodes according to the urban area node map and the traffic data includes:
and judging whether flow interaction exists between the nodes according to the flow data, and if so, establishing a link between the two groups of nodes with the flow interaction.
It can be known from the above description that when there is traffic interaction between two nodes, it indicates that there is an interaction relationship between the two nodes, and the association between the nodes is improved by establishing a link between the nodes having an interaction relationship.
Further, the calculating distance values of all the links and obtaining a space-time outlier according to the distance values comprises:
calculating distance characteristic values of all the links, and obtaining a time abnormal value according to the distance characteristic values;
calculating the distance between all different links, and obtaining a spatial abnormal value according to the distance between the links;
and obtaining the space-time abnormal value according to the intersection of the time abnormal value and the space abnormal value.
From the above description, it can be seen that the time abnormal value is obtained according to the distance values of the links, the space abnormal value is obtained according to the distance between the links, and finally the data satisfying both the time abnormal value and the space abnormal value is screened out to obtain the space-time abnormal value, so that the abnormal value in the traffic data can be accurately screened out.
Further, the calculating distance characteristic values of all the links and obtaining a time abnormal value according to the distance characteristic values comprises:
calculating the average value of the traffic distances between all the time frames in each link and any other time frame to obtain the average value of the distances between each time frame and other time frames;
and screening out an extreme value from all the distance average values, and taking the extreme value as the time abnormal value.
As can be seen from the above description, by calculating the average value of the distances between the flows of each link in different time frames, the time frames in which the abnormal flow data exists in each link can be screened out, and the time abnormal value corresponding to each link can be obtained.
Further, the calculating the distances between all the different links and obtaining the spatial outlier according to the distances between the links includes:
calculating the traffic distance values between any two groups of links in the same time frame to obtain the traffic distance values between each link and all other links in each time frame;
and screening out an extreme value from all the flow distance values, and taking the extreme value as the space abnormal value.
From the above description, by calculating the corresponding distance between each link and other links in the same time frame on the same day, the links far away from other links can be screened out, that is, the links have spatial anomaly relative to other links, so as to obtain an accurate spatial anomaly value.
Further, the screening out an extreme value from all the flow distance values, and the taking the extreme value as the spatial abnormal value includes:
normalizing all the flow distance values;
and screening out an extreme value from the normalized flow distance values, and taking the extreme value as the spatial abnormal value.
According to the description, the flow distance value is normalized and then screened, so that the influence of different numerical values among different areas can be eliminated, and the screening precision is improved.
Further, the determining a relationship between a node and a node according to the flow urban area node map and the flow data to obtain a normal cause-and-effect map includes:
aggregating the input flow and the output flow of each node to obtain a flow time sequence;
obtaining a stable flow sequence by extracting features and differentially processing the flow time sequence;
and performing Glangel causal test on the steady flow sequence to obtain the relationship between the nodes and obtain the normal causal graph.
From the above description, it can be known that the periodicity of the flow time sequence can be eliminated by aggregating the input flow and the output flow of each node and performing feature extraction and difference processing on the flow time sequence obtained by aggregation, so that a stable flow sequence is obtained, the grand causal test is more favorably performed, and the validity of the test result is improved.
Further, the obtaining of the steady flow sequence by feature extraction and differential processing of the flow time sequence includes:
extracting the characteristics of the flow time sequence by adopting a non-negative matrix factorization method;
and performing the difference processing by fitting and combining the flow data to the nonnegative matrix factorization to obtain the stable flow sequence.
As can be seen from the above description, by performing feature extraction on the flow time series by using a non-negative matrix factorization method, only the additive combination of the flow time series in the flow time series is allowed, so that the effect of hiding the features is achieved, and the method is more suitable for processing traffic flow data, thereby improving the data processing effect.
Referring to fig. 3, another embodiment of the present invention provides a causal detection device for abnormal traffic flow, 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 steps of the causal detection method for abnormal traffic flow. And visualizing an effect graph obtained by the traffic abnormal flow causal detection method.
The method and the device for causally detecting the traffic abnormal flow can be suitable for urban intelligent traffic scenes, can better understand the current road traffic mode and make a more timely coping strategy by establishing an accurate causality graph, and are explained by a specific implementation mode as follows:
example one
Referring to fig. 1 and fig. 2, a causal detection method for traffic abnormal flow includes the steps of:
s1, carrying out region segmentation on an urban region to obtain an urban region node map;
the grid system is important for analyzing a large-scale spatial data set and dividing an urban area into identifiable grid units, and in the embodiment, the urban area is divided by adopting a honeycomb hexagon to obtain an urban area node map; constructing an area graph according to the track data, wherein each node in the area graph represents an area; the connecting line between the nodes represents the flow between the areas;
s2, acquiring flow data, and constructing links between nodes according to the urban area node graph and the flow data; specifically, whether flow interaction exists between the nodes is judged according to the flow data, and if yes, a link is established between the two groups of nodes with the flow interaction; if the flow data is taxi order track data; the track data comprises four characteristics of a taxi order number, a timestamp, a longitude and a latitude; sequencing the taxis according to the same order number of the same taxi and the time stamp sequence so as to obtain the taxi running track; for two pieces of track data of the taxi, if the two pieces of track data are continuous in time (namely continuous in timestamp) and continuous in space (namely continuous in longitude and latitude), namely the taxi moves from one area to an adjacent area, the interaction of flow is generated between the two areas; regarding two continuous track data, if they belong to two different areas respectively, a link exists between the two areas at the moment; namely, a link is formed from one divided area to the next area, and the link has directionality;
converting each of said trace data into a series of said links between pairs of regions by scanning the entire trace data set; if an interaction of traffic occurs between two areas, connecting the two areas through the link, wherein link = (total traffic, traffic out-degree, traffic in-degree); namely, one link is a vector containing three dimensional information; wherein the total flow is defined as the total flow on the link, i.e. the total number of vehicles passing through two zones within a certain time period; flow rate, i.e., the proportion of the flow rate flowing through the link over a certain period of time to the total flow rate from the region; flow in, which is the proportion of the flow that flows in through the link over a certain period of time to the total flow from the zone;
s3, calculating the distance values of all the links, and obtaining a time-space abnormal value according to the distance values; specifically, defining an abnormal value in the distance values as the space-time abnormal value, and considering the abnormal value as the space-time abnormal value only if the abnormal value exists in both time and space;
s31, calculating distance characteristic values of all the links, and obtaining time abnormal values according to the distance characteristic values;
s311, calculating the average value of the traffic distances between all the time frames in each link and any other time frame to obtain the average value of the distances between each time frame and other time frames, namely obtaining the distance characteristic value of the link; in an alternative embodiment, the calculation is performed using euclidean distance, that is, data related to time such as the same time on different dates or the same day on different weeks can be obtained by the calculation, and special traffic data patterns of similar behaviors can be observed; for example, for link L1, it will have different traffic attributes { (flow 1, in1, out 1), (flow 2, in2, out 2), (flow 3, in3, out 3), (flow 4, in4, out 4) } at different time frames { t1, t2, t3, t4. }, where flow1, in1, and out1 represent total traffic, traffic out, and traffic in the link definition, respectively; the distance between any two time frames is calculated by the following formula:
namely t1t2, t1t3, t1t4., t2t1, t2t3, t2t4,. T3t1, t3t2, t3t 4.; for the t1 time frame, the distance between the t1 time frame and other time frames is t1t2+ t1t3+ t1t 4./n; by analogy, the distances from all the time frames to other time frames can be obtained to obtain the distance average value as the distance characteristic value;
s312, screening out an extreme value from all the distance average values, and taking the extreme value as the time abnormal value; the distance feature value calculated in step S311 is subjected to normalization processing by subtracting the minimum value from the distance feature value and dividing by the maximum value: (x-min)/max, the distance characteristic value is in the range of [0,1], and the influence of the numerical values of different areas is eliminated; screening distance characteristics of different time frames of each link according to a given threshold, wherein a maximum value, namely the time frame which is far away from all other links, is considered as a time abnormal value; as in an alternative embodiment, for the normalized distance feature, set threshold =0.95 to screen out the top 5 percent of distance features where the distance is very large and consider it to be the maximum of the distance features;
s32, calculating the distances among all the different links, and obtaining a spatial abnormal value according to the distances among the links; for the detection of the spatial outlier, the distance between two links is calculated in a time frame range to determine, that is, the maximum difference value between different links in the same time frame is searched to represent, specifically:
s321, calculating the traffic distance values between any two groups of links in the same time frame to obtain the traffic distance values between each link and all other links in each time frame; if the distance between two link flows in the same time frame of the same day is calculated, the mahalanobis distance is used for representing the distance; that is, for the links on the same time frame, for example, any two links may have different traffic attributes such as link { L1, L2, L3, L4. }, where L1 is { (flow 1, in1, out 1), L2 is (flow 2, in2, out 2), L3 is (flow 3, in3, out 3), and L4 is (flow 4, in4, out 4. }, the traffic distance of any two of the links is calculated by the following formula:
x and y are two different links, namely the distance between L1 and all other links at the time of t1 can be obtained, and the distance between any two links at the time of t1 can be obtained by analogy;
s322, screening out an extreme value from all the flow distance values, and taking the extreme value as the space abnormal value; such as normalizing all of the traffic distance values; screening out an extreme value from the normalized flow distance values, and taking the extreme value as the space abnormal value; if an extreme value in all links within the t1 time frame is detected, the link with the largest difference between the characteristics of the links and the spatial neighbors is a spatial outlier;
s33, obtaining the space-time abnormal value according to the intersection of the time abnormal value and the space abnormal value; that is, by intersecting the temporal outlier result obtained in step S31 with the spatial outlier result obtained in step S32, the intersection is the spatio-temporal outlier;
s4, judging the relation between the space-time abnormal values according to an abnormal cause-and-effect tree algorithm to obtain an abnormal cause-and-effect graph; looking up an abnormal value causal relationship based on an abnormal causal tree algorithm looking at a relationship between the abnormal values from an earliest time range to a last time range; when one of said spatio-temporal outliers STO1 occurs temporally before the other of said spatio-temporal outliers STO2 and they also have a head-to-tail relationship spatially, the outlier STO1 is considered to be the cause of the other outlier STO 2; finding the dependency relationship according to the obtained abnormal link value in the previous step, specifically:
for each of the spatio-temporal outliers, it includes a temporal attribute (i.e., the time at which the anomaly occurred), a spatial attribute (i.e., the two regions at which the anomaly link occurred); then for the exception link L1 (t 1, startgrid1, end grid 1) and the exception connection L2 (t 2, start grid2, end grid 2), if t1 and t2 are temporally consecutive (e.g. 10;
s5, judging the relationship between the nodes according to the flow city region node graph and the flow data to obtain a normal cause-and-effect graph;
s51, aggregating the input flow and the output flow of each node to obtain a flow time sequence; aggregating all traffic from the inputs and outputs of each region in order to obtain normal traffic information; wherein, for example, the starting area of the link is denoted by Rgno, and the arrival area of the link is denoted by Rgnd; then for any region Rgnk, the input flow: gnk. Inflow = Σ i =1 to nlink. Flow where line i.rgno = Rgnk; output flow rate: gnk. Outflow = Σ i =1 to nlink. Flow whherelinki. Rgnd = Rgnk;
s52, obtaining a stable flow sequence through feature extraction and differential processing of the flow time sequence, specifically:
s521, extracting the characteristics of the flow time sequence by adopting a nonnegative matrix factorization method; thereby removing the periodicity of the flow time series in step S51;
s522, performing the differential processing through fitting combination of the flow data on the non-negative matrix factorization to obtain the steady flow sequence; if the column number of W and the row number of H are set as the data volume of one day, the non-negative matrix factorization can extract features according to the day, the fitting result of the non-negative matrix factorization by using the original data is subjected to differential processing to obtain the differential fluctuation of the original data, and further the periodic influence is removed to obtain the stable flow sequence;
s53, performing Glangel causal test on the steady flow sequence to obtain the relationship between the nodes and obtain the normal causal graph;
if the Glan's causal test is carried out between every two of all the stable flow sequences, taking the result of the Glan's causal test as the edge between the region directed graphs, namely if the causal relationship exists between two regions, the edge exists, otherwise, the causal relationship does not exist; meanwhile, in order to further extract causality but not correlation, the Glan's cause and effect test is selected to be carried out on different time lengths, and finally a normal regional cause and effect graph is obtained; and improving the traffic flow through the obtained abnormal cause and effect diagram and the obtained normal cause and effect diagram.
Example two
Referring to fig. 3, a causal detection device for abnormal traffic flow 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 steps of the causal detection method for abnormal traffic flow according to the first embodiment.
In summary, according to the method and the device for causal detection of abnormal traffic flow provided by the present invention, after an urban area is subjected to area segmentation to obtain an urban area node map, a link is created between nodes according to the obtained urban traffic data, monitoring of traffic changes of other areas in the city by the urban area traffic changes is realized by the link, and a space-time abnormal value is obtained by calculating a distance value of the link, so that an abnormal causal map is obtained by calculating a space-time abnormal value by an abnormal causal fruit tree algorithm, and a causal graph in a normal traffic state is combined to distinguish normal traffic flows from abnormal traffic flows, so that abnormal change characteristics of traffic flows caused by the space-time abnormal value can be captured, causal detection of the abnormal value and transitivity along with time change are realized, potential interactions between different urban areas and roads are disclosed, effective perception of urban traffic states is promoted, and more accurate decision is facilitated.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all equivalent changes made by using the contents of the present specification and the drawings, or applied directly or indirectly to the related technical fields, are included in the scope of the present invention.
Claims (10)
1. A causal detection method for traffic abnormal flow is characterized by comprising the following steps:
carrying out region segmentation on the urban region to obtain an urban region node map;
acquiring flow data, and constructing links among nodes according to the urban area node graph and the flow data;
calculating the distance values of all the links, and obtaining a space-time abnormal value according to the distance values;
judging the relation between the space-time abnormal values according to an abnormal cause-and-effect tree algorithm to obtain an abnormal cause-and-effect graph;
and judging the relationship between the nodes according to the flow city region node graph and the flow data to obtain a normal cause-and-effect graph.
2. The causal detection method for traffic abnormal flow according to claim 1, wherein the step of performing region segmentation on the urban region to obtain an urban region node map comprises:
partitioning urban areas by adopting regular hexagons to obtain an urban area node map;
each of the nodes represents an area;
the connecting lines between the nodes represent the flow between the regions.
3. The method of claim 1, wherein the constructing links between nodes according to the urban area node map and traffic data comprises:
and judging whether flow interaction exists between the nodes or not according to the flow data, and if so, establishing a link between two groups of nodes with the flow interaction.
4. The causal detection method for traffic abnormal flow according to claim 1, wherein said calculating distance values of all said links and obtaining a space-time abnormal value according to said distance values comprises:
calculating distance characteristic values of all the links, and obtaining a time abnormal value according to the distance characteristic values;
calculating the distances among all different links, and obtaining a spatial abnormal value according to the distances among the links;
and obtaining the space-time abnormal value according to the intersection of the time abnormal value and the space abnormal value.
5. The method as claimed in claim 4, wherein said calculating distance eigenvalues of all said links and obtaining time outliers based on said distance eigenvalues comprises:
calculating the average value of the traffic distances between all the time frames in each link and any other time frame to obtain the average value of the distances between each time frame and other time frames;
and screening out an extreme value from all the distance average values, and taking the extreme value as the time abnormal value.
6. The causal detection method for traffic abnormal flow according to claim 4, wherein said calculating distances between all different said links and obtaining a spatial outlier according to the distances between said links comprises:
calculating the traffic distance values between any two groups of links in the same time frame to obtain the traffic distance values between each link and all other links in each time frame;
and screening out an extreme value from all the flow distance values, and taking the extreme value as the space abnormal value.
7. The causal detection method for traffic abnormal flow according to claim 6, wherein said screening out an extreme value from all said flow distance values, and said taking said extreme value as said spatial abnormal value comprises:
normalizing all the flow distance values;
and screening out an extreme value from the normalized flow distance values, and taking the extreme value as the space abnormal value.
8. The method of claim 1, wherein the step of obtaining a normal cause-and-effect map according to the flow urban area node map and the relationship between nodes and flow data judgment nodes comprises:
aggregating the input flow and the output flow of each node to obtain a flow time sequence;
obtaining a stable flow sequence by extracting features and differentially processing the flow time sequence;
and performing Glangel causal test on the steady flow sequence to obtain the relationship between the nodes and obtain the normal causal graph.
9. The causal detection method for traffic abnormal flow according to claim 8, wherein said obtaining a steady flow sequence by feature extraction and differential processing of said flow time sequence comprises:
extracting the characteristics of the flow time sequence by adopting a non-negative matrix factorization method;
and performing the differential processing by fitting and combining the flow data to the non-negative matrix factorization to obtain the steady flow sequence.
10. A causal detection device for traffic abnormal flow, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of a causal detection method for traffic abnormal flow as claimed in any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310160350.3A CN115984077B (en) | 2023-02-24 | 2023-02-24 | Traffic abnormal flow causal detection method and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310160350.3A CN115984077B (en) | 2023-02-24 | 2023-02-24 | Traffic abnormal flow causal detection method and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115984077A true CN115984077A (en) | 2023-04-18 |
| CN115984077B CN115984077B (en) | 2023-06-13 |
Family
ID=85970759
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310160350.3A Active CN115984077B (en) | 2023-02-24 | 2023-02-24 | Traffic abnormal flow causal detection method and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115984077B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110161261A1 (en) * | 2009-12-28 | 2011-06-30 | Nec(China) Co., Ltd. | Method and system for traffic prediction based on space-time relation |
| EP2590151A1 (en) * | 2011-11-03 | 2013-05-08 | Deutsche Telekom AG | A framework for the systematic study of vehicular mobility and the analysis of city dynamics using public web cameras |
| US20130166188A1 (en) * | 2011-12-21 | 2013-06-27 | Microsoft Corporation | Determine Spatiotemporal Causal Interactions In Data |
| CN103258427A (en) * | 2013-04-24 | 2013-08-21 | 北京工业大学 | Urban expressway traffic real-time monitoring system and method based on information physical network |
| CN105023044A (en) * | 2015-07-21 | 2015-11-04 | 清华大学 | Traffic flow cause and effect relation excavation method based on large number of time sequences |
| CN109643485A (en) * | 2016-12-30 | 2019-04-16 | 同济大学 | A kind of urban highway traffic method for detecting abnormality |
| CN110111575A (en) * | 2019-05-16 | 2019-08-09 | 北京航空航天大学 | A kind of Forecast of Urban Traffic Flow network analysis method based on Complex Networks Theory |
| CN111179592A (en) * | 2019-12-31 | 2020-05-19 | 合肥工业大学 | Urban traffic prediction method and system based on spatio-temporal data flow fusion analysis |
| US10902337B1 (en) * | 2020-04-24 | 2021-01-26 | Jun Tang | Method and device of trajectory outlier detection, and storage medium thereof |
| CN115204478A (en) * | 2022-06-30 | 2022-10-18 | 北京航空航天大学 | Public traffic flow prediction method combining urban interest points and space-time causal relationship |
-
2023
- 2023-02-24 CN CN202310160350.3A patent/CN115984077B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110161261A1 (en) * | 2009-12-28 | 2011-06-30 | Nec(China) Co., Ltd. | Method and system for traffic prediction based on space-time relation |
| EP2590151A1 (en) * | 2011-11-03 | 2013-05-08 | Deutsche Telekom AG | A framework for the systematic study of vehicular mobility and the analysis of city dynamics using public web cameras |
| US20130166188A1 (en) * | 2011-12-21 | 2013-06-27 | Microsoft Corporation | Determine Spatiotemporal Causal Interactions In Data |
| CN103258427A (en) * | 2013-04-24 | 2013-08-21 | 北京工业大学 | Urban expressway traffic real-time monitoring system and method based on information physical network |
| CN105023044A (en) * | 2015-07-21 | 2015-11-04 | 清华大学 | Traffic flow cause and effect relation excavation method based on large number of time sequences |
| CN109643485A (en) * | 2016-12-30 | 2019-04-16 | 同济大学 | A kind of urban highway traffic method for detecting abnormality |
| CN110111575A (en) * | 2019-05-16 | 2019-08-09 | 北京航空航天大学 | A kind of Forecast of Urban Traffic Flow network analysis method based on Complex Networks Theory |
| CN111179592A (en) * | 2019-12-31 | 2020-05-19 | 合肥工业大学 | Urban traffic prediction method and system based on spatio-temporal data flow fusion analysis |
| US10902337B1 (en) * | 2020-04-24 | 2021-01-26 | Jun Tang | Method and device of trajectory outlier detection, and storage medium thereof |
| CN115204478A (en) * | 2022-06-30 | 2022-10-18 | 北京航空航天大学 | Public traffic flow prediction method combining urban interest points and space-time causal relationship |
Non-Patent Citations (1)
| Title |
|---|
| 杨森炎: ""基于时空数据挖掘的城市路网交通态势分析及应用研究"", 《中国博士学位论文全文数据库 工程科技Ⅱ辑》, no. 02, pages 9 - 10 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115984077B (en) | 2023-06-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Urban anomaly analytics: Description, detection, and prediction | |
| Hamdi et al. | Spatiotemporal data mining: a survey on challenges and open problems | |
| Lin et al. | A novel variable selection method based on frequent pattern tree for real-time traffic accident risk prediction | |
| CN108764085B (en) | Crowd counting method based on generation of confrontation network | |
| Sun et al. | DxNAT—Deep neural networks for explaining non-recurring traffic congestion | |
| US20180293464A1 (en) | Method for understanding machine-learning decisions based on camera data | |
| CN113011322B (en) | Detection model training method and detection method for monitoring specific abnormal behavior of video | |
| CN110084201A (en) | A kind of human motion recognition method of convolutional neural networks based on specific objective tracking under monitoring scene | |
| Usmani et al. | A review of unsupervised machine learning frameworks for anomaly detection in industrial applications | |
| Hu et al. | Parallel spatial-temporal convolutional neural networks for anomaly detection and location in crowded scenes | |
| Liu et al. | A binary-classification-tree based framework for distributed target classification in multimedia sensor networks | |
| Jin et al. | A visual analytics system for improving attention-based traffic forecasting models | |
| Loy et al. | Modelling activity global temporal dependencies using time delayed probabilistic graphical model | |
| CN112101132B (en) | Traffic condition prediction method based on graph embedding model and metric learning | |
| Zhang et al. | Off-deployment traffic estimation—a traffic generative adversarial networks approach | |
| CN115984077B (en) | Traffic abnormal flow causal detection method and equipment | |
| Attanasi et al. | A hybrid method for real-time short-term predictions of traffic flows in urban areas | |
| CN109871469A (en) | Tuftlet crowd recognition method based on dynamic graphical component | |
| Liu et al. | Attention based spatio-temporal graph convolutional network with focal loss for crash risk evaluation on urban road traffic network based on multi-source risks | |
| Hu et al. | Multi-level trajectory learning for traffic behavior detection and analysis | |
| Grigoropoulos et al. | Detection and Classification of Bicyclist Group Behavior for Automated Vehicle Applications | |
| Tan et al. | Vehicle detection using YOLO and mobility tracking during COVID-19 pandemic lockdowns | |
| Prezioso et al. | Integrating Object Detection and Advanced Analytics for Smart City Crowd Management | |
| Bhattacharyya et al. | Semantic-aware anomaly detection in real time parking data | |
| CN110738692A (en) | spark cluster-based intelligent video identification method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |