A traffic warning method for intelligent traffic control system
By optimizing the European-style distance and DTW distance and combining the node impact factor, the untimely and inaccurate problems of traffic congestion warning in the smart traffic control system are solved, more accurate traffic pattern recognition and dynamic threshold adjustment are achieved, and the traffic congestion warning is improved Timeliness and accuracy.
Patent Information
- Application Number
- CN202411186786.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-08-28
AI Technical Summary
In the smart traffic control system, when traffic congestion warning is conducted based on the clustering method, the spatial correlation between traffic nodes is ignored, resulting in untimely and inaccurate early warnings.
By obtaining the traffic timing data of each traffic node and the location distance between each two traffic nodes, the European-style distance is optimized; combining the DTW distance and node impact factor, secondary optimization is performed to finally obtain a more accurate European-style distance for clustering Dynamic adjustment of class and traffic congestion thresholds.
It improves the timeliness and accuracy of traffic congestion warnings, is more in line with changes in actual traffic patterns, adjusts traffic congestion thresholds in a timely manner, and enhances the early warning capabilities of the smart traffic control system.
Smart Images

Figure CN118711376B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of traffic control, and in particular to a traffic warning method for an intelligent traffic control system. Background Art
[0002] In the smart traffic control system of the city, it is necessary to issue traffic congestion warnings for all traffic nodes in the city. In the existing traffic congestion warning process, by setting the traffic congestion threshold of the traffic node, the real-time traffic flow of each traffic node and the real-time average vehicle speed are judged by threshold. When the monitoring value exceeds the set threshold, the traffic congestion warning is issued for the traffic node, and the real-time information of the traffic node is updated in real time in the smart traffic control system of the city.
[0003] However, in the city's intelligent traffic control system, the traffic pattern of each traffic node is different, that is, it has different traffic conditions at different times (during peak hours in the morning and evening, some main roads will have high traffic volume, and if a traffic accident occurs, traffic congestion is more likely to occur. And there will be differences between traffic nodes due to the different vehicles that the roads can accommodate.) Therefore, when conducting traffic congestion warnings in the intelligent traffic control system, it is necessary to identify the traffic patterns of all traffic nodes in the city and divide all traffic nodes in the city into different traffic patterns, so as to conduct differentiated traffic congestion warning processes based on the differences in traffic patterns presented by traffic nodes.
[0004] The clustering method is used to identify the traffic mode of all traffic nodes. Specifically, the traffic monitoring time series data of each traffic node within a period of time is obtained, and the distance between traffic nodes is measured in the clustering process through the Euclidean distance between the traffic monitoring time series data, which is used to judge similar traffic mode changes, so that traffic nodes with similar traffic mode changes can be divided into the same cluster, and then judged by the same traffic warning threshold. However, in this process, it only analyzes the traffic mode through the traffic monitoring time series data between traffic nodes, and the spatial correlation between traffic nodes is missing. This will lead to the neglect of the traffic impact between traffic nodes in the process of traffic mode division, resulting in the untimely traffic congestion warning based on the clustering results.
[0005] Therefore, how to improve the timeliness and accuracy of urban traffic congestion warning based on clustering methods has become an urgent problem to be solved. Summary of the invention
[0006] In view of this, an embodiment of the present invention provides a traffic warning method for an intelligent traffic control system to solve the problem of how to improve the timeliness and accuracy of urban traffic congestion warning based on a clustering method.
[0007] An embodiment of the present invention provides a traffic warning method for an intelligent traffic control system, the method comprising the following steps:
[0008] Obtain the traffic time series data of each traffic node monitored by the intelligent traffic control system respectively;
[0009] The position distance between each two of the traffic nodes is obtained respectively, and the Euclidean distance between the traffic time series data of each two of the traffic nodes in the clustering process is optimized for the first time by using the position distance between each two of the traffic nodes to obtain the optimized Euclidean distance;
[0010] According to the traffic time series data of each two traffic nodes, the DTW distance and the node influence factor between each two traffic nodes are respectively obtained, and the optimized Euclidean distance between each two traffic nodes is optimized twice by using the DTW distance and the node influence factor between each two traffic nodes to obtain the final Euclidean distance;
[0011] All traffic nodes are clustered based on the final Euclidean distance to obtain corresponding clustering results, and the traffic congestion threshold of each traffic node is dynamically adjusted according to the clustering results to obtain an adjusted traffic congestion threshold, and traffic warnings are issued to each traffic node according to the adjusted traffic congestion threshold.
[0012] Preferably, the method of using the position distance between each two of the traffic nodes to perform a first optimization on the Euclidean distance between the traffic time series data of each two of the traffic nodes in the clustering process to obtain the optimized Euclidean distance comprises:
[0013] For any two traffic nodes, the position distance between the two traffic nodes is normalized to obtain a corresponding normalized value, and the difference between a constant 1 and the normalized value is used as the position proximity degree;
[0014] The product of the Euclidean distance between the traffic time series data of the two traffic nodes and the location proximity degree is used as the optimized Euclidean distance between the two traffic nodes.
[0015] Preferably, the obtaining of the DTW distance and the node influence factor between each two traffic nodes according to the traffic time series data of each two traffic nodes includes:
[0016] For any two traffic nodes, the t-th data point in the traffic time series data of the two traffic nodes is taken as two target points, where: , N represents the length of the traffic time series data; taking each of the target points as the center, respectively obtaining a time series window of length L in the corresponding traffic time series data, obtaining the Pearson correlation coefficient between the two time series windows, and setting the weight between the two target points according to the Pearson correlation coefficient;
[0017] Calculate the data variance corresponding to each of the time series windows respectively, perform normalization processing on each of the data variances to obtain two corresponding normalized values, calculate the difference between the two normalized values, and obtain the subtraction result between a constant 1 and the difference as a fluctuation similarity index between the two target points;
[0018] According to the weights between all target points in the traffic time series data of the two traffic nodes, all fluctuation similarity indicators are weighted summed and averaged, and the corresponding average is used as the node influence factor between the two traffic nodes.
[0019] Preferably, setting the weight between the two target points according to the Pearson correlation coefficient includes:
[0020] Detect whether the Pearson correlation coefficient is less than 0. If it is detected that the Pearson correlation coefficient is less than 0, set the weight between the two target points to 0. If it is detected that the Pearson correlation coefficient is not less than 0, use the Pearson correlation coefficient as the weight between the two target points.
[0021] Preferably, the method of performing secondary optimization on the optimized Euclidean distance between each two traffic nodes by using the DTW distance between each two traffic nodes and the node influence factor to obtain the final Euclidean distance comprises:
[0022] For any two traffic nodes, obtain a subtraction result between a constant 1 and the location proximity degree between the two traffic nodes, and use the product of the subtraction result, the DTW distance and the node influence factor as a distance optimization value;
[0023] The sum of the distance optimization value and the optimized Euclidean distance between the two traffic nodes is taken as the final Euclidean distance between the two traffic nodes.
[0024] Preferably, dynamically adjusting the traffic congestion threshold of each traffic node according to the clustering result to obtain the adjusted traffic congestion threshold includes:
[0025] The clustering result includes multiple clustering clusters. For any traffic node, the clustering cluster to which the traffic node belongs is determined as the target clustering cluster. According to the data type of the traffic time series data of the traffic node, the quartile is adaptively obtained from the traffic time series data of all traffic nodes included in the target clustering cluster as the adjusted traffic congestion threshold corresponding to the data type of the traffic time series data of the traffic node.
[0026] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:
[0027] The present invention obtains the traffic time series data of each traffic node monitored by the intelligent traffic control system respectively; obtains the position distance between each two traffic nodes respectively, and uses the position distance between each two traffic nodes to perform a first optimization on the Euclidean distance between the traffic time series data of each two traffic nodes in a clustering process to obtain the optimized Euclidean distance; obtains the DTW distance and the node influence factor between each two traffic nodes according to the traffic time series data of each two traffic nodes respectively, and uses the DTW distance and the node influence factor between each two traffic nodes to perform a second optimization on the optimized Euclidean distance between each two traffic nodes to obtain the final Euclidean distance; clusters all traffic nodes based on the final Euclidean distance to obtain corresponding clustering results, dynamically adjusts the traffic congestion threshold of each traffic node according to the clustering result to obtain the adjusted traffic congestion threshold, and performs traffic warning for each traffic node according to the adjusted traffic congestion threshold. Among them, when dividing the traffic modes of all traffic nodes monitored by the intelligent traffic control system, the Euclidean distance of the traffic time series data of every two traffic nodes in the clustering process is optimized twice by analyzing the location distance between the traffic nodes and the correlation between the traffic mode changes between the traffic nodes, so as to highlight the differences in traffic modes between the traffic nodes, making the result of clustering all traffic nodes according to the optimized final Euclidean distance more in line with reality, and then making real-time dynamic adjustments to the traffic congestion warning of each traffic node according to the clustering results, so as to adapt to the changes in traffic modes of each traffic node in time, thereby improving the timeliness and accuracy of the intelligent traffic control system's traffic congestion warning for the city. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0029] Figure 1This is a method flow chart of a traffic warning method for an intelligent traffic control system provided in Example 1 of the present invention. DETAILED DESCRIPTION
[0030] Embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.
[0031] It should be noted that the terms "first", "second", etc. in the specification of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.
[0032] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.
[0033] See also Figure 1 , is a method flow chart of a traffic warning method for an intelligent traffic control system provided by Embodiment 1 of the present invention, such as Figure 1 As shown, the method may include:
[0034] Step S101, respectively obtain the traffic time series data of each traffic node monitored by the intelligent traffic control system.
[0035] The intelligent traffic control system is used to monitor the traffic congestion situation of each traffic node in the city, and then update the traffic congestion information in real time and display it at the corresponding traffic node position in the intelligent traffic control system. In the intelligent traffic control system of the city, each traffic node is equipped with a vehicle flow monitoring device and a vehicle speed sensor for collecting vehicle flow and average vehicle speed. Therefore, in the embodiment of the present invention, the traffic time series data of each traffic node monitored by the intelligent traffic control system is collected, wherein the traffic time series data may include vehicle flow time series data and average vehicle speed time series data, and may also include vehicle flow time series data or average vehicle speed time series data, that is, the traffic time series data includes at least one dimension of time series data.
[0036] It should be noted that the traffic time series data is collected for one month and is collected every 10 minutes. There is no restriction here and it can be set according to the implementation scenario.
[0037] Step S102, respectively obtain the position distance between every two traffic nodes, and use the position distance between every two traffic nodes to perform a first optimization on the Euclidean distance between the traffic time series data of every two traffic nodes in the clustering process to obtain the optimized Euclidean distance.
[0038] After obtaining the traffic time series data of each traffic node, the traffic mode of the traffic nodes can be identified and divided using clustering algorithms according to the differences in traffic time series data between traffic nodes. Traditionally, when clustering algorithms are used to identify and divide traffic modes of traffic nodes, they are implemented based on the Euclidean distance between traffic time series data of traffic nodes. However, when traffic nodes are congested, they may transfer traffic to their neighboring traffic nodes, causing the neighboring traffic nodes to present similar traffic state changes. Therefore, when performing Euclidean distance analysis between traffic nodes, it is necessary to comprehensively measure the Euclidean distance between traffic time series data of traffic nodes in combination with the spatial relationship between traffic nodes, so as to achieve the purpose of optimizing the Euclidean distance in the traditional clustering process.
[0039] In the process of urban traffic, the correlation between traffic nodes is weak under normal traffic conditions. When traffic congestion occurs at any traffic node, the traffic flow is likely to be transferred to its neighboring traffic nodes. Therefore, in the embodiment of the present invention, the position distance between each two traffic nodes is obtained respectively to characterize the neighbor relationship between traffic nodes, and the distance measurement in the clustering process is optimized by the neighbor relationship. The specific optimization method is: using the position distance between each two traffic nodes, the Euclidean distance between the traffic time series data of each two traffic nodes in the clustering process is optimized for the first time to obtain the optimized Euclidean distance. It is worth noting that the position distance refers to the distance between two traffic intersections in the city.
[0040] Among them, the location distance between every two traffic nodes is used to perform the first optimization on the Euclidean distance between the traffic time series data of every two traffic nodes in the clustering process, and the optimized Euclidean distance is obtained, including:
[0041] For any two traffic nodes, the position distance between the two traffic nodes is normalized to obtain a corresponding normalized value, and the difference between a constant 1 and the normalized value is used as the position proximity degree;
[0042] The product of the Euclidean distance between the traffic time series data of the two traffic nodes and the location proximity degree is used as the optimized Euclidean distance between the two traffic nodes.
[0043] In one embodiment, taking the ith traffic node and the jth traffic node as an example, the position distance between the ith traffic node and the jth traffic node is , according to the location distance between the ith traffic node and the jth traffic node, obtain the location proximity degree between the ith traffic node and the jth traffic node , where 1 represents a constant and Norm() represents a normalization function. Using the proximity degree between the i-th traffic node and the j-th traffic node, the Euclidean distance between the traffic time series data of the i-th traffic node and the traffic time series data of the j-th traffic node in the clustering process is optimized for the first time, and the calculation expression of the optimized Euclidean distance is obtained as follows: in, represents the optimized Euclidean distance between the traffic time series data of the i-th traffic node and the traffic time series data of the j-th traffic node, Indicates the proximity between the i-th traffic node and the j-th traffic node. Represents the traffic time series data of the i-th traffic node and the traffic time series data of the jth traffic node The Euclidean distance between .
[0044] It should be noted that the lower the degree of location proximity, the farther the distance between the traffic nodes, the more dissimilar the traffic modes between the two traffic nodes, and the larger the Euclidean distance optimized by the degree of location proximity; while for traffic nodes with a high degree of location proximity, the closer the distance between the traffic nodes, the more likely there is a correlation in the traffic modes, the more similar the traffic modes between the two traffic nodes, and the corresponding use of the degree of location proximity to reduce the Euclidean distance, so that the value of the optimized Euclidean distance is smaller, which can better reflect the similarity of the traffic modes between the neighboring traffic nodes.
[0045] Similarly, the Euclidean distance between the traffic time series data of every two traffic nodes in the clustering process is optimized for the first time to obtain the optimized Euclidean distance.
[0046] Step S103, according to the traffic time series data of every two traffic nodes, respectively obtain the DTW distance and the node influence factor between every two traffic nodes, and use the DTW distance and the node influence factor between every two traffic nodes to perform secondary optimization on the optimized Euclidean distance between every two traffic nodes to obtain the final Euclidean distance.
[0047] After optimizing the traditional Euclidean distance by the position distance between the traffic nodes, in the actual scenario, because the position distance between the traffic nodes is relatively close, but there is no traffic flow transfer between the two in the traffic network, that is, traffic node A and traffic node B are neighboring traffic nodes, but in the actual traffic network, the two cannot reach each other directly (there are situations such as prohibited turns). Then, if the traditional Euclidean distance is optimized only by the above-mentioned position distance, there will still be inaccurate clustering results. Therefore, in the embodiment of the present invention, it is also necessary to perform an impact analysis between the traffic nodes through the traffic mode changes presented between the neighboring traffic nodes, so that in the process of distance measurement, the Euclidean distance between the traffic nodes and the mode change difference are dynamically evaluated by the degree of influence of the traffic mode change.
[0048] First, the DTW algorithm is used to obtain the DTW distance between the traffic time series data of every two traffic nodes, which is used to measure whether there is a similar traffic change pattern between the two traffic nodes, wherein the traffic time series data refers to the traffic flow time series data and the average vehicle speed time series data obtained in step S101. The DTW algorithm belongs to the prior art and will not be described here. Then, according to the traffic time series data of every two traffic nodes, the node influence factor between every two traffic nodes is obtained, which is used to evaluate the influence factor between traffic nodes through the traffic node association presented in the traffic time series data. The specific method for obtaining the node influence factor is:
[0049] For any two traffic nodes, the t-th data point in the traffic time series data of the two traffic nodes is taken as two target points, where: , N represents the length of the traffic time series data; taking each of the target points as the center, respectively obtaining a time series window of length L in the corresponding traffic time series data, obtaining the Pearson correlation coefficient between the two time series windows, and setting the weight between the two target points according to the Pearson correlation coefficient;
[0050] Calculate the data variance corresponding to each of the time series windows respectively, perform normalization processing on each of the data variances to obtain two corresponding normalized values, calculate the difference between the two normalized values, and obtain the subtraction result between a constant 1 and the difference as a fluctuation similarity index between the two target points;
[0051] According to the weights between all target points in the traffic time series data of the two traffic nodes, all fluctuation similarity indicators are weighted summed and averaged, and the corresponding average is used as the node influence factor between the two traffic nodes.
[0052] In one embodiment, taking the i-th traffic node and the j-th traffic node as examples, for the t-th data point in the traffic time-series data of the i-th traffic node and the t-th data point in the traffic time-series data of the j-th traffic node, where , N represents the length in the traffic time-series data, L represents the length of the time window. Respectively, taking the t-th data point as the center, a time window with a length of 11 is obtained in the respective traffic time-series data, corresponding to two time windows. The Pearson correlation coefficient between the two time windows is obtained and denoted as . Then, according to the Pearson correlation coefficient, the weight between the t-th data point in the traffic time-series data of the i-th traffic node and the t-th data point in the traffic time-series data of the j-th traffic node is set.
[0053] Meanwhile, the data variances of each time window are calculated respectively and denoted as and . It should be noted that for the first 5 data and the last 5 data in the traffic time-series data of the i-th traffic node and the j-th traffic node, the analysis of the time window is not performed. That is, when analyzing the node influence factor, the first 5 data and the last 5 data in the traffic time-series data of each traffic node are not considered. Then, the target points in the traffic time-series data of each traffic node are denoted as m, 5 < m < N - 5. Thus, according to the weights between all target points between the i-th traffic node and the j-th traffic node and the data variances of the corresponding time windows, the node influence factor between the i-th traffic node and the j-th traffic node is obtained. The calculation expression of the node influence factor is: Where represents the node influence factor between the i-th traffic node and the j-th traffic node, represents the weight between the t-th data point in the traffic time-series data of the i-th traffic node and the t-th data point in the traffic time-series data of the j-th traffic node, represents the data variance of the time window of the t-th data point in the traffic time-series data of the i-th traffic node, represents the data variance of the time window of the t-th data point in the traffic time-series data of the j-th traffic node, Norm() represents the normalization function, 1 represents a constant, and m represents the number of target points in the traffic time-series data.
[0054] It should be noted that the stationarity of the corresponding timestamp is measured by the variance difference of the time windows corresponding to each target point (timestamp) in the traffic time-series data of the traffic node. The more similar the stationarity difference between the time windows corresponding to the timestamps is, the more necessary it is to analyze the similarity of the fluctuations between the time windows. Therefore, in On this basis, the similarity analysis of traffic data changes between traffic nodes is performed through the Pearson correlation coefficient between time series windows. For the correlation between two traffic nodes in the time series window with the same timestamp, the more similar the stationarity between the time series windows and the higher the Pearson correlation coefficient, the higher the correlation between the two traffic nodes in the time series window. For the traffic time series data corresponding to the two traffic nodes, the higher the average influence at all timestamps, the higher the node influence factor between the two traffic nodes.
[0055] Preferably, setting the weight between the two target points according to the Pearson correlation coefficient includes:
[0056] Detect whether the Pearson correlation coefficient is less than 0. If it is detected that the Pearson correlation coefficient is less than 0, set the weight between the two target points to 0. If it is detected that the Pearson correlation coefficient is not less than 0, use the Pearson correlation coefficient as the weight between the two target points.
[0057] In one embodiment, since the value range of the Pearson correlation coefficient is [-1, 1], when the Pearson correlation coefficient between two time series windows is When it is less than 0, the corresponding weight Set to 0, otherwise, when the Pearson correlation coefficient between two time series windows When it is greater than or equal to 0, the Pearson correlation coefficient Directly as weight .
[0058] Finally, the optimized Euclidean distance between each two traffic nodes is optimized twice using the DTW distance between each two traffic nodes and the node influence factor to obtain the final Euclidean distance. The method of the secondary optimization is:
[0059] For any two traffic nodes, obtain a subtraction result between a constant 1 and the location proximity degree between the two traffic nodes, and use the product of the subtraction result, the DTW distance and the node influence factor as a distance optimization value;
[0060] The sum of the distance optimization value and the optimized Euclidean distance between the two traffic nodes is taken as the final Euclidean distance between the two traffic nodes.
[0061] In one embodiment, the optimized Euclidean distance between the traffic time series data of the i-th traffic node and the traffic time series data of the j-th traffic node in the clustering process is optimized twice using the DTW distance and the node influence factor, and the calculation expression of the final Euclidean distance is obtained as follows:
[0062] in, represents the final Euclidean distance between the traffic time series data of the i-th traffic node and the traffic time series data of the j-th traffic node, Indicates the proximity between the i-th traffic node and the j-th traffic node. Represents the traffic time series data of the i-th traffic node and the traffic time series data of the jth traffic node The Euclidean distance between them, 1 represents a constant, represents the node influence factor between the i-th traffic node and the j-th traffic node, Represents the traffic time series data of the i-th traffic node and the traffic time series data of the jth traffic node The DTW distance between them.
[0063] It should be noted that the final Euclidean distance means that when the proximity of the traffic nodes is low, the distance measurement during clustering tends to be the Euclidean distance between the traffic time series data. When the proximity of the traffic nodes is high, the distance measurement during clustering needs to be based on the Euclidean distance between the traffic time series data, combined with the changes in the traffic time series data between the traffic nodes, that is, the DTW distance between the traffic time series data corresponding to the traffic nodes and the node correlation to measure the distance. When there are similar traffic change patterns between neighboring traffic nodes, the distance measurement during clustering is optimized by the DTW distance and the node influence factor to avoid reducing the Euclidean distance between neighboring traffic nodes that do not affect each other in neighboring traffic nodes, thereby causing incorrect clustering division.
[0064] Similarly, the final Euclidean distance between the traffic time series data of every two traffic nodes in the clustering process is obtained to complete the distance metric optimization in the clustering process.
[0065] Step S104, clustering all traffic nodes based on the final Euclidean distance to obtain corresponding clustering results, dynamically adjusting the traffic congestion threshold of each traffic node according to the clustering results to obtain an adjusted traffic congestion threshold, and issuing a traffic warning for each traffic node according to the adjusted traffic congestion threshold.
[0066] After obtaining the final Euclidean distance between the traffic time series data of every two traffic nodes in the clustering process, K-means clustering is performed on all traffic nodes based on the final Euclidean distance, thereby dividing all traffic nodes in the city into different clusters, one cluster corresponding to one traffic mode, and realizing the recognition of traffic modes. Among them, the number of clusters in the K-means clustering process is determined by the elbow method to obtain the corresponding clustering results.
[0067] After obtaining the clustering results of all traffic nodes in the city's smart traffic control system, the traffic congestion threshold of each traffic node can be dynamically adjusted according to the clustering results to obtain the adjusted traffic congestion threshold corresponding to each traffic intersection. The dynamic adjustment method is:
[0068] The clustering result includes multiple clustering clusters. For any traffic node, the clustering cluster to which the traffic node belongs is determined as the target clustering cluster. According to the data type of the traffic time series data of the traffic node, the quartile is adaptively obtained from the traffic time series data of all traffic nodes included in the target clustering cluster as the adjusted traffic congestion threshold corresponding to the data type of the traffic time series data of the traffic node.
[0069] In one embodiment, for any traffic node, first, based on multiple clusters in the clustering results, the cluster to which the traffic node belongs is determined as the target cluster, and based on the data type of the traffic time series data of the traffic node, the quartile is adaptively obtained from the traffic time series data of all traffic nodes included in the target cluster as the adjusted traffic congestion threshold corresponding to the data type of the traffic time series data of the traffic node. For example, when the traffic time series data includes vehicle flow time series data, the vehicle flow time series data of all traffic nodes included in the target cluster are obtained to form a set. Considering that more vehicle flow means more traffic congestion, the upper quartile is obtained based on the data in the set as the traffic congestion threshold for the traffic node. The traffic congestion threshold of the corresponding traffic flow is obtained. When the real-time traffic flow of the traffic node is higher than the traffic congestion threshold, a congestion warning is issued for the traffic node in the intelligent traffic control system. When the traffic time series data includes the average vehicle speed time series data, the average vehicle speed time series data of all traffic nodes included in the target cluster are obtained to form a set. Considering that the smaller the average vehicle speed, the more congested the traffic is, the lower quartile is obtained according to the data in the set as the traffic congestion threshold of the average vehicle speed corresponding to the traffic node. When the real-time average vehicle speed of the traffic node is lower than the traffic congestion threshold, a congestion warning is issued for the traffic node in the intelligent traffic control system.
[0070] At this point, the traffic congestion threshold of each traffic node can be adaptively obtained according to the cluster to which each traffic node belongs and the data type of the corresponding traffic time series data, so as to provide real-time congestion warning for each traffic node.
[0071] It is worth noting that in the intelligent traffic control system, all traffic nodes are continuously monitored, and the traffic modes corresponding to the traffic nodes are dynamically adjusted. The cluster changes of traffic nodes in the clustering results of traffic mode division are continuously monitored. When the clustering cluster to which the traffic node belongs changes, the traffic congestion threshold is readjusted through the updated clustering cluster. For the clustering results of the traffic mode division of the traffic node, the clustering results are updated by setting an update time (the update time is set to be updated once a month in the present invention). The update of the clustering results is to re-cluster all traffic nodes, so as to adapt to the traffic mode changes of the traffic nodes in the city.
[0072] In summary, the embodiments of the present invention respectively obtain the traffic time series data of each traffic node monitored by the intelligent traffic control system; respectively obtain the position distance between each two of the traffic nodes, and use the position distance between each two of the traffic nodes to perform a first optimization on the Euclidean distance between the traffic time series data of each two of the traffic nodes in the clustering process to obtain the optimized Euclidean distance; according to the traffic time series data of each two of the traffic nodes, respectively obtain the DTW distance and the node influence factor between each two of the traffic nodes, and use the DTW distance and the node influence factor between each two of the traffic nodes to perform a second optimization on the optimized Euclidean distance between each two of the traffic nodes to obtain the final Euclidean distance; cluster all traffic nodes based on the final Euclidean distance to obtain the corresponding clustering results, dynamically adjust the traffic congestion threshold of each of the traffic nodes according to the clustering results to obtain the adjusted traffic congestion threshold, and perform traffic warning for each of the traffic nodes according to the adjusted traffic congestion threshold. Among them, when dividing the traffic modes of all traffic nodes monitored by the intelligent traffic control system, the Euclidean distance of the traffic time series data of every two traffic nodes in the clustering process is optimized twice by analyzing the location distance between the traffic nodes and the correlation between the traffic mode changes between the traffic nodes, so as to highlight the differences in traffic modes between the traffic nodes, making the result of clustering all traffic nodes according to the optimized final Euclidean distance more in line with reality, and then making real-time dynamic adjustments to the traffic congestion warning of each traffic node according to the clustering results, so as to adapt to the changes in traffic modes of each traffic node in time, thereby improving the timeliness and accuracy of the intelligent traffic control system's traffic congestion warning for the city.
[0073] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.
Claims
1. A traffic warning method for an intelligent traffic control system, It is characterized in that The method comprises: Obtain the traffic time series data of each traffic node monitored by the intelligent traffic control system respectively; The position distance between each two of the traffic nodes is obtained respectively, and the Euclidean distance between the traffic time series data of each two of the traffic nodes in the clustering process is optimized for the first time by using the position distance between each two of the traffic nodes to obtain the optimized Euclidean distance; According to the traffic time series data of each two of the traffic nodes, the DTW distance and the node influence factor between each two of the traffic nodes are respectively obtained, and the optimized Euclidean distance between each two of the traffic nodes is twice optimized by using the DTW distance and the node influence factor between each two of the traffic nodes to obtain the final Euclidean distance; Clustering all traffic nodes based on the final Euclidean distance to obtain corresponding clustering results, dynamically adjusting the traffic congestion threshold of each traffic node according to the clustering results to obtain an adjusted traffic congestion threshold, and performing traffic warning for each traffic node according to the adjusted traffic congestion threshold; The method of obtaining the DTW distance and the node influence factor between each two traffic nodes according to the traffic time series data of each two traffic nodes includes: For any two traffic nodes, the t-th data point in the traffic time series data of the two traffic nodes is taken as two target points, where: , N represents the length of the traffic time series data; taking each of the target points as the center, respectively obtaining a time series window of length L in the corresponding traffic time series data, obtaining the Pearson correlation coefficient between the two time series windows, and setting the weight between the two target points according to the Pearson correlation coefficient; Calculate the data variance corresponding to each of the time series windows respectively, perform normalization processing on each of the data variances to obtain two corresponding normalized values, calculate the difference between the two normalized values, and obtain the subtraction result between a constant 1 and the difference as a fluctuation similarity index between the two target points; According to the weights between all target points in the traffic time series data of the two traffic nodes, all fluctuation similarity indicators are weighted summed and averaged, and the corresponding average is used as the node influence factor between the two traffic nodes.
2. A traffic warning method for an intelligent traffic control system according to claim 1, It is characterized in that The method of optimizing the Euclidean distance between the traffic time series data of each two traffic nodes in the clustering process by using the location distance between each two traffic nodes to obtain the optimized Euclidean distance includes: For any two traffic nodes, the position distance between the two traffic nodes is normalized to obtain a corresponding normalized value, and the difference between a constant 1 and the normalized value is used as the position proximity degree; The product of the Euclidean distance between the traffic time series data of the two traffic nodes and the location proximity degree is used as the optimized Euclidean distance between the two traffic nodes.
3. A traffic warning method for an intelligent traffic control system according to claim 1, It is characterized in that The step of setting the weight between the two target points according to the Pearson correlation coefficient includes: Detect whether the Pearson correlation coefficient is less than 0. If it is detected that the Pearson correlation coefficient is less than 0, set the weight between the two target points to 0. If it is detected that the Pearson correlation coefficient is not less than 0, use the Pearson correlation coefficient as the weight between the two target points.
4. A traffic warning method for an intelligent traffic control system according to claim 2, It is characterized in that The method of performing secondary optimization on the optimized Euclidean distance between each two traffic nodes by using the DTW distance between each two traffic nodes and the node influence factor to obtain the final Euclidean distance includes: For any two traffic nodes, obtain a subtraction result between a constant 1 and the location proximity degree between the two traffic nodes, and use the product of the subtraction result, the DTW distance and the node influence factor as a distance optimization value; The sum of the distance optimization value and the optimized Euclidean distance between the two traffic nodes is taken as the final Euclidean distance between the two traffic nodes.
5. A traffic warning method for an intelligent traffic control system according to claim 1, It is characterized in that The dynamically adjusting the traffic congestion threshold of each traffic node according to the clustering result to obtain the adjusted traffic congestion threshold includes: The clustering result includes multiple clustering clusters. For any traffic node, the clustering cluster to which the traffic node belongs is determined as the target clustering cluster. According to the data type of the traffic time series data of the traffic node, the quartile is adaptively obtained from the traffic time series data of all traffic nodes included in the target clustering cluster as the adjusted traffic congestion threshold corresponding to the data type of the traffic time series data of the traffic node.
