Highway traffic operation status prediction method based on deep mining of checkpoint data

Abstract

The present invention belongs to the technical field of highway traffic state prediction and specifically discloses a method for predicting highway traffic operation status based on deep mining of checkpoint data, including: processing real-time speed data to construct a current speed dataset; preprocessing historical average operating speed data of highway sections and selecting data from the hour before the prediction moment to construct a historical dataset; calculating the Euclidean distance between the speed data from the hour before the prediction moment and the data from the same period in the historical dataset based on the current speed dataset and the historical dataset, thereby constructing a similarity set; calculating the predicted speed and free flow speed based on the similarity set, thereby calculating the highway congestion index; analyzing the flow-speed relationship; and determining the highway traffic operation status based on the congestion index and flow rate. The present invention solves the problems of significant monitoring blind spots, low data reliability, poor environmental adaptability, and insufficient judgment accuracy in existing highway traffic state prediction technologies.

Description

Technical Field

[0001] The present invention belongs to the technical field of highway traffic state prediction, and in particular relates to a highway traffic operation state prediction method based on deep mining of checkpoint data. Background Art

[0002] With the continued expansion of the highway network and the significant increase in the spatiotemporal complexity of traffic flows, accurate prediction of traffic status has become a core research direction for intelligent network management and control. According to the "2023 National Highway Operation Analysis Report," existing detection technologies have an average response delay of 8.2 minutes to sudden congestion events, with a false positive rate as high as 35% during holidays. Current prediction models primarily rely on fixed detectors, GPS-enabled floating vehicles, and video surveillance. However, these models face multiple bottlenecks in practical application: insufficient density of fixed detectors leads to blind spots in network monitoring; the low sampling rate of floating vehicle data makes it difficult to characterize global traffic flow; and video surveillance is susceptible to environmental interference and has high computational complexity. As a network-wide infrastructure, bayonet gantry systems can acquire cross-section-level traffic flow parameters in real time with high spatiotemporal granularity, providing a high-quality data base for status prediction. However, existing prediction models are not fully adapted to the propagation characteristics of multi-section, long-distance, linear traffic flows on highways. Summary of the Invention

[0003] The purpose of this invention is to solve the problems of significant monitoring blind spots, low data reliability, poor environmental adaptability and insufficient judgment accuracy in existing highway traffic status prediction technologies, and propose a highway traffic operation status prediction method based on deep mining of checkpoint data.

[0004] The technical solution of the present invention is: a method for predicting highway traffic operation status based on deep mining of checkpoint data, comprising the following steps:

[0005] Acquire real-time speed monitoring data, perform data cleaning and data smoothing on the real-time speed monitoring data, and construct the current speed data set;

[0006] Collect historical average speed data of highway sections, pre-process the historical average speed data to obtain a historical database, and select data from the historical database one hour before the predicted time every day to construct a historical data set;

[0007] Based on the current speed data set and the historical data set, the Euclidean distance between the speed data one hour before the prediction moment and the data of the same period in the historical data set is calculated, and then a similarity set is constructed;

[0008] Based on the speed data in the similarity set, the predicted speed is calculated;

[0009] Calculate the free flow speed and calculate the congestion index of the highway based on the free flow speed and predicted speed;

[0010] Analyze the flow-speed relationship based on historical speed data of highway sections;

[0011] Based on the relationship between congestion index and flow speed, the traffic operation status of the highway is judged.

[0012] Preferably, the historical average speed data of the highway sections are collected, pre-processed to obtain a historical database, and the data of one hour before the predicted time of each day are selected from the historical database to construct a historical data set, specifically:

[0013] The monitoring intervals are divided based on adjacent gantries on the highway section with a minimum spacing of more than 2km, and the historical average operating speed data for each preset time step in the past year is collected;

[0014] Perform outlier sliding window variance analysis and time series alignment compensation processing on the historical average running speed data to construct a standardized data set with a confidence level of ≥98%, thus obtaining the historical database.

[0015] The predicted time is obtained, and the average speed of each preset time step one hour before the predicted time of each day is selected from the historical database as the historical data set.

[0016] Preferably, the calculation formula for the average speed of each preset time step is:

[0017]

[0018] in, represents the average velocity at each preset time step, Indicates the distance between adjacent gantries on a highway section. Indicates the total number of vehicles in the statistical period. Indicates the The travel time of a vehicle.

[0019] Preferably, the Euclidean distance between the speed data one hour before the prediction moment and the data of the same period in the historical data set is calculated based on the current speed data set and the historical data set, and then a similarity set is constructed, specifically as follows:

[0020] Based on the current speed data set, obtain the speed data for each preset time step within one hour before the prediction time;

[0021] Calculate the Euclidean distance between the speed data of each preset time step within one hour before the forecast time and the data of the same period in the historical data set, and select the one with the smallest Euclidean distance. The speed data in the most similar historical data sets constitutes a similarity set ,in, Indicates the minimum Euclidean distance The most similar historical dataset Speed ​​data, ;

[0022] The calculation formula of the Euclidean distance is:

[0023]

[0024] in, represents the Euclidean distance, represents the number of time steps, Indicates the first hour before the forecast time. Velocity data of a preset time step, Indicates the historical data set j In the sample i The velocity value of each time step, i represents the time step index, Indicates the sample index of the historical dataset.

[0025] Preferably, the predicted speed is calculated based on the speed data in the similarity set, specifically as follows:

[0026] Extract speed data from similarity sets Follow-up Step speed value , get the candidate prediction value set ;

[0027] Based on the candidate prediction value set, the weighted average method is used to calculate the prediction speed. The specific calculation formula is:

[0028]

[0029] in, Represents the predicted speed value, Represents the speed value in the candidate prediction value set The weight of .

[0030] Preferably, the weight is allocated in inverse proportion to the similarity, and the specific calculation formula is:

[0031]

[0032] in, Indicates the first hour before the forecast time. The velocity data of the preset time step is compared with the velocity data of the The Euclidean distance of similar samples, Indicates the first hour before the forecast time. The velocity data of the preset time step is compared with the velocity data of the The Euclidean distance between similar samples.

[0033] Preferably, the method for calculating the free flow speed is as follows: sorting the historical speed data of the highway section from largest to smallest, selecting the average value of the top 9th of the speed data as the free flow speed of the highway section, and the specific calculation formula is:

[0034]

[0035] in, represents the free flow speed, Indicates the number of the first ninth of the historical speed data selected. Indicates the first ninth of the historical speed data. The historical speed, Indicates the total number of historical speed data for highway sections.

[0036] Preferably, the calculation formula of the highway congestion index is specifically as follows:

[0037]

[0038] in, It represents the congestion index adjusted by the holiday correction coefficient. represents the predicted congestion index before adjustment, represents the holiday correction factor, represents the free flow speed, Indicates the predicted speed value;

[0039] The holiday correction factor The expression formula is:

[0040]

[0041]

[0042] in, represents the adjustment factor, Indicates the natural base The logarithmic function with base , represents the holiday impact index, represents the baseline impact value, Indicates the number of days before and after the holiday. It represents the average traffic volume for 5 days before and after holidays. represents the annual average daily flow rate, Indicates the speed reduction.

[0043] Preferably, the traffic operation status of the expressway is judged based on the relationship between the congestion index and the flow speed, specifically:

[0044] According to the industry standard GA / T 115-2020 Road Traffic Congestion Evaluation Method, set the endpoint values ​​of the average travel speed division interval of the highway section 、 and and the endpoint values ​​of the highway congestion index interval 、 and ;in, , ;

[0045] The inflection point flow value where the unstable flow state turns to the congested flow state in the flow-speed relationship is taken as the critical flow value between the congested state and the crowded state b ;

[0046] At low speed, When the traffic status of the expressway is judged based on the dual thresholds of "speed + flow" and the congestion index of the expressway, represents the average travel speed on a highway segment;

[0047] When the traffic status of the expressway is determined, the congestion index of the expressway and the average travel speed of the expressway section are used to determine the traffic status of the expressway;

[0048] The expressway traffic operation status includes:

[0049] Congestion status: , the flow rate is less than b , ;

[0050] Crowded state: , flow rate is greater than or equal to b , ,or, , ;

[0051] Slow-moving state: , ;

[0052] Smooth status: , .

[0053] The beneficial effects of the present invention are:

[0054] The present invention eliminates the monitoring blind spots caused by the large spacing between traditional detection equipment by deeply mining the data from highway checkpoints, thus meeting the mandatory requirements of industry specifications for coverage of the entire road section; by analyzing the traffic flow parameters of highway sections and combining the "speed-flow" dual-threshold judgment system, the accuracy of traffic status recognition in low-speed and high-flow scenarios is significantly improved; the introduction of a holiday correction coefficient to dynamically adjust the congestion index solves the problem of misjudgment caused by the particularity of traffic patterns on non-working days; at the same time, the algorithm calculation efficiency is optimized, and the prediction response time of long sections is shortened to minutes, meeting real-time requirements. Overall, it is superior to traditional methods in terms of perception coverage, computing efficiency and adaptability to complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 The figure shows a flow chart of a highway traffic operation status prediction method based on deep mining of checkpoint data provided by Example 1 of the present invention.

[0056] Figure 2 Shown is a flow-speed relationship diagram provided in Example 1 of the present invention.

[0057] Figure 3 Shown is a schematic diagram of the actual operating status of a certain road section 1 provided by Example 2 of the present invention at 7:13 on July 16, 2024.

[0058] Figure 4 The figure shows a schematic diagram of the actual operating status of a certain road section 1 provided by Example 2 of the present invention at 7:38 on July 16, 2024.

[0059] Figure 5 The figure shows a schematic diagram of the actual operating status of a certain road section 2 provided by Example 2 of the present invention at 17:00 on August 4, 2024.

[0060] Figure 6 The figure shows a schematic diagram of the actual operating status of a certain road section 2 provided by Example 2 of the present invention at 17:01 on August 4, 2024. DETAILED DESCRIPTION

[0061] The exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the accompanying drawings are merely exemplary and are intended to illustrate the principles and spirit of the present invention, rather than to limit the scope of the present invention.

[0062] Example 1:

[0063] like Figure 1 As shown, a highway traffic operation status prediction method based on deep mining of checkpoint data includes the following steps:

[0064] S1. Acquire real-time speed monitoring data, perform data cleaning and data smoothing on the real-time speed monitoring data, and construct the current speed data set;

[0065] S2. Collect historical average speed data for highway sections, preprocess the historical average speed data to obtain a historical database, and select data from the historical database one hour before the predicted time each day to construct a historical data set;

[0066] S3. Based on the current speed dataset and the historical dataset, calculate the Euclidean distance between the speed data one hour before the prediction time and the data of the same period in the historical dataset, and then construct a similarity set;

[0067] S4. Calculate the predicted speed based on the speed data in the similarity set;

[0068] S5. Calculate the free flow speed, and calculate the congestion index of the highway based on the free flow speed and the predicted speed;

[0069] S6. Analyze the flow-speed relationship based on historical speed data of highway sections;

[0070] S7. Determine the traffic operation status of the expressway based on the relationship between the congestion index and the flow speed.

[0071] In this embodiment, step S2 is specifically as follows:

[0072] The monitoring interval is divided based on adjacent gantries on the highway section with a minimum spacing of more than 2km. The historical average operating speed data for each preset time step (the preset time step is set to 5 minutes) in the past year is collected;

[0073] Perform outlier sliding window variance analysis and time series alignment compensation processing on the historical average running speed data to construct a standardized data set with a confidence level of ≥98%, thus obtaining the historical database.

[0074] Get the predicted time, and select the average speed every 5 minutes in the hour before the predicted time every day from the historical database as the historical data set; the calculation formula for the average speed every 5 minutes is:

[0075]

[0076] in, represents the average velocity at each preset time step, Indicates the distance between adjacent gantries on a highway section. Indicates the total number of vehicles in the statistical period. Indicates the The travel time of a vehicle.

[0077] Analyze the characteristics and trends of historical average operating speed data, and based on the theory of traffic flow dynamics, use the Gaussian-Weibull Mixture Model (GWMM) to establish the probability density function of speed data. :

[0078]

[0079] in, Indicates the actual speed of the vehicle. represents the proportion of congestion status in historical data (0≤λ≤1, the Gaussian item weight is 1−λ), represents the natural base, represents pi, the Gaussian term ( represents the average speed in free flow conditions, represents the standard deviation) characterizes the velocity distribution law in the free stream state, and the Weibull term ( The characteristic speed representing the congestion state, The Weibull parameter (representing the decay rate of congestion speed) describes the speed distribution in congestion and characterizes the speed decay characteristics in congestion. In addition, the Weibull parameter serves as a physical constraint for analyzing the flow-speed relationship in congestion.

[0080] The EM algorithm is used to cluster historical data and separate speed subsets in different traffic states (convergence condition: log-likelihood change rate < 0.1%). It provides free-flow clustering data for free-flow speed calculation and facilitates the construction of historical data with the same cluster label for similar sets.

[0081] In this embodiment, step S3 is specifically as follows:

[0082] Based on the current speed data set, obtain the speed data for each preset time step within one hour before the prediction time;

[0083] Calculate the Euclidean distance between the speed data of each preset time step within one hour before the forecast time and the data of the same period in the historical data set, and select the one with the smallest Euclidean distance. The speed data in the most similar historical data sets constitutes a similarity set ,in, Indicates the minimum Euclidean distance The most similar historical dataset Speed ​​data, ;

[0084] The calculation formula of the Euclidean distance is:

[0085]

[0086] in, represents the Euclidean distance, represents the number of time steps, =12, Indicates the first hour before the forecast time. Velocity data of a preset time step, Indicates the historical data set j In the sample i The velocity value of each time step, i represents the time step index, Indicates the sample index of the historical dataset.

[0087] In this embodiment, by giving the current speed data, predicting the future The speed value of the next step (such as 30 minutes) is the predicted speed. In step S3, a similar sample of the current speed data is constructed based on the current speed data set. Therefore, the predicted speed is based on the speed data in the similar set. Step S4 is specifically as follows:

[0088] Extract speed data from similarity sets Follow-up Step speed value , get the candidate prediction value set ;

[0089] Based on the candidate prediction value set, the weighted average method is used to calculate the prediction speed. The specific calculation formula is:

[0090]

[0091] in, Represents the predicted speed value, Represents the speed value in the candidate prediction value set The weight is distributed in inverse proportion to the similarity. The specific calculation formula is:

[0092]

[0093] in, Indicates the first hour before the forecast time. The velocity data of the preset time step is compared with the velocity data of the The Euclidean distance of similar samples, Indicates the first hour before the forecast time. The velocity data of the preset time step is compared with the velocity data of the The Euclidean distance between similar samples.

[0094] In this embodiment, the free flow speed is the average travel speed of motor vehicles passing through a road section under low traffic volume and low density conditions. The free flow speed is not affected by the season, but is affected by the vehicle type. The road section speed of the blue-plate vehicle in each natural month is selected as the free flow speed of the road section. The free flow speed that exceeds the speed limit is selected from the speed limit. There are two methods for calculating the free flow speed, namely the standard free flow speed and the 85% percentile free flow speed. In step S5, the standard free flow speed is used to calculate the free flow speed, specifically: the historical speed data of the highway section is sorted from large to small, and the average value of the top nine percent is selected as the free flow speed of the highway section. The specific calculation formula is:

[0095]

[0096] in, represents the free flow speed, Indicates the number of the first ninth of the historical speed data selected. Indicates the first ninth of the historical speed data. The historical speed, Indicates the total number of historical speed data for highway sections.

[0097] In this embodiment, in order to reduce the misjudgment rate, the traffic state is determined in combination with the predicted congestion index. The congestion index is calculated using the following formula:

[0098]

[0099] in, represents the predicted congestion index, represents the free flow speed, Indicates the predicted speed value;

[0100] In order to eliminate the interference of holidays on traffic status judgment, the holiday correction coefficient (Spring Festival, May Day, and National Day) is added to adjust the predicted congestion index to obtain the adjusted congestion index, which is specifically:

[0101]

[0102] in, It represents the congestion index adjusted by the holiday correction coefficient. Indicates the holiday correction factor;

[0103] The holiday correction factor Using the hierarchical calculation method, The classification value rules are shown in Table 1, and the expression formula is:

[0104]

[0105]

[0106] in, It represents the adjustment factor, which is 0.18 and is calibrated by the least square method. Indicates the natural base The logarithmic function with base , Indicates the holiday impact index, with a value range of 0-100. It represents the baseline impact value determined by regression analysis of five-year data, and is taken as 30. Indicates the number of days before and after the holiday. It represents the average traffic volume for 5 days before and after holidays. represents the annual average daily flow rate, Indicates the speed reduction.

[0107] Table 1 Holiday correction coefficient Grading value rules

[0108]

[0109] In this embodiment, step S6 is specifically as follows:

[0110] Statistical analysis of the data from the past year shows that the road speed is mainly in the range of ≥70km / h. When the speed is less than 30km / h, it is congested flow, and the operating status needs to be judged in combination with the flow rate. The speed distribution fluctuates around the center line by about ±15%, and the critical point is divided by the lower limit of the fluctuation, such as Figure 2 As shown. Figure 2 The flow-speed relationship diagram shown in the figure is used to find the boundary where the unstable flow state turns to the congested flow state. The inflection point flow value is used as the critical flow value between the congested state and the crowded state. b .

[0111] In this embodiment, step S7 is specifically as follows:

[0112] Traffic status is determined based on speed. At low speeds (less than 30 km / h), the traffic status is determined based on the "speed + flow" dual threshold and the congestion index. Specifically:

[0113] According to the industry standard GA / T 115-2020 Road Traffic Congestion Evaluation Method, set the endpoint values ​​of the average travel speed division interval of the highway section 、 and and the endpoint values ​​of the highway congestion index interval 、 and ;in, , ;

[0114] The inflection point flow value where the unstable flow state turns to the congested flow state in the flow-speed relationship is taken as the critical flow value between the congested state and the crowded state b In this embodiment, the critical flow value of the congestion state and the crowded state b It can also be set according to "GA / T 115-2020 Road Traffic Congestion Evaluation Method" and relevant highway engineering technical specifications. Specifically, it is set to 75% to 85% of the design capacity of the road section. The specific value is adjusted according to the number of lanes and design speed.

[0115] At low speed, When the traffic status of the expressway is judged based on the dual thresholds of "speed + flow" and the congestion index of the expressway, represents the average travel speed on a highway segment;

[0116] When the highway traffic operation status is determined, the highway congestion index and the average travel speed of the highway section are used to judge the traffic operation status. The traffic status level classification is shown in Table 2, which refers to the industry standard GA / T 115-2020 Road Traffic Congestion Evaluation Method shown in Table 3.

[0117] Table 2 Traffic status classification table

[0118]

[0119] Table 3 Industry Standard: GA / T 115-2020 Road Traffic Congestion Evaluation Method Classification

[0120]

[0121] To address the shortcomings of existing highway traffic status prediction technologies, such as significant monitoring blind spots, low data reliability, poor environmental adaptability, and insufficient judgment accuracy, this paper constructs a full-dimensional prediction model based on in-depth analysis of checkpoint data, specifically solving the following technical problems:

[0122] 1) To address the serious problem of monitoring blind spots, this paper proposes a full-section data fusion method based on a bayonet gantry system, which enables millisecond-level continuous acquisition of flow and speed parameters at any section of the highway. This eliminates the monitoring blind spots caused by traditional detection equipment spacing greater than 2 km, and meets the requirements for full road section coverage in Article 5.2.1 of JTGB01-2023 "Technical Specifications for Highway Traffic Status Monitoring."

[0123] 2) The proposed method improves the shortcoming of the traditional bayonet model in terms of weak spatiotemporal correlation, explores the speed situation between the upstream and downstream sections of the gantry, analyzes the spatiotemporal coupling relationship between flow and speed parameters, and effectively restores the propagation process of traffic flow parameters in the road network.

[0124] 3) The proposed method improves the limitations of single-metric judgment by establishing a dual-threshold judgment system based on speed and traffic volume, improving the accuracy of traffic status identification in low-speed, high-traffic scenarios. A holiday correction factor is incorporated into the calculation of the congestion index to eliminate the special impact of holidays on traffic status and improve prediction accuracy.

[0125] Example 2:

[0126] On the basis of Example 1, the embodiment of the present invention takes a certain place as an example to construct a basic data set of highways in the certain place to illustrate and verify the effect of the highway traffic operation status prediction method based on deep mining of checkpoint data proposed by the present invention.

[0127] The road section between two adjacent gantries in the same direction was divided, and data from 284 gantries were obtained. The average vehicle speed every 5 minutes between gantries in the past year (2024.06.01-2025.05.31) was calculated, and the historical traffic flow data with a time step of 5 minutes in the past year was also calculated. After data cleaning and other steps, a historical data set was obtained, and the traffic status level was divided based on the data set, as shown in Table 4.

[0128] Table 4 Traffic status classification table

[0129]

[0130] 1. Verify segment data

[0131] The points were selected from a certain road section 1 (free flow speed 100 km / h) as shown in Table 5 and a certain road section 2 (free flow speed 90 km / h) as shown in Table 6. Example verification and analysis were conducted on July 16, 2024 (Tuesday) and August 2, 2024 (Sunday).

[0132] Table 5 Data of a certain road section 1

[0133]

[0134] Table 6 Data of a certain road section 2

[0135]

[0136] The road section speed in the data is the average operating speed of the section between the two gantries. When the distance between the gantries is long, the cross-sectional video cannot fully prove the operating status of the entire section.

[0137] 2. Comparison and verification of predicted and actual traffic conditions

[0138] (1) From AB gantry to BC gantry on a certain road section 1

[0139] It is predicted that the road section will be congested from 5:00 to 9:00 on July 16, 2024 (Tuesday). In reality, vehicles will be queued from 7:00 to 8:00, and the congestion will last for a long time. Figure 3 and Figure 4 As shown in Table 7, the actual operation status of the road section is basically consistent with the prediction results. The prediction data of the road section are shown in Table 7.

[0140] Table 7 Prediction data of section 1

[0141]

[0142] (2) From DE gantry to DF gantry on a certain road section 2

[0143] It is predicted that the road section will be congested from 15:00 to 19:00 on August 4, 2024 (Sunday). The actual operation status is that vehicles will queue up from 17:00 to 18:00, which is a congested state. Figure 5 and Figure 6 As shown in Table 8, the actual operating status of the road section is basically consistent with the prediction results. The prediction data of the road section is shown in Table 8.

[0144] Table 8 Prediction data of section 2

[0145]

[0146] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the present invention.

Claims

1. A highway traffic operation status prediction method based on deep mining of checkpoint data, characterized in that: The following steps are involved: Acquire real-time speed monitoring data, perform data cleaning and data smoothing on the real-time speed monitoring data, and construct the current speed data set; Collect the historical average speed data of highway sections, pre-process the historical average speed data to obtain a historical database, and select the data one hour before the prediction time every day from the historical database to construct a historical data set. Specifically: The monitoring intervals are divided based on adjacent gantries on the highway section with a minimum spacing of more than 2km, and the historical average operating speed data for each preset time step in the past year is collected; Perform outlier sliding window variance analysis and time series alignment compensation processing on the historical average running speed data to construct a standardized data set with a confidence level of ≥98%, thus obtaining the historical database. Obtain the predicted time, and select the average speed of each preset time step one hour before the predicted time every day in the historical database as the historical data set; Based on the current speed data set and the historical data set, the Euclidean distance between the speed data one hour before the prediction moment and the data of the same period in the historical data set is calculated, and then a similarity set is constructed; Based on the speed data in the similarity set, the predicted speed is calculated; Calculate the free flow speed and calculate the congestion index of the highway based on the free flow speed and predicted speed; The calculation formula for the highway congestion index is as follows: in, It represents the congestion index adjusted by the holiday correction coefficient. represents the predicted congestion index before adjustment, represents the holiday correction factor, represents the free flow speed, Indicates the predicted speed value; The holiday correction factor The expression formula is: in, represents the adjustment factor, Indicates the natural base The logarithmic function with base , represents the holiday impact index, represents the baseline impact value, Indicates the number of days before and after the holiday. It represents the average traffic volume for 5 days before and after holidays. represents the annual average daily flow rate, Indicates the speed reduction; set the holiday correction coefficient according to the holiday type The value range is: Spring Festival The value range is 1.12-1.25, National Day The value range is 1.10-1.20, May Day / Tomb-Sweeping Day / Dragon Boat Festival The value range is 1.05-1.15, and the weekend is closed The value range is 1.03-1.12; based on the historical speed data of the highway section, the flow-speed relationship is analyzed; Based on the relationship between the congestion index and flow speed, the traffic operation status of the expressway is judged as follows: According to the industry standard GA / T 115-2020 Road Traffic Congestion Evaluation Method, set the endpoint values ​​of the average travel speed division interval of the highway section 、 and and the endpoint values ​​of the highway congestion index interval 、 and ;in, , ; The inflection point flow value where the unstable flow state turns to the congested flow state in the flow-speed relationship is taken as the critical flow value between the congested state and the crowded state b ; At low speed, When the traffic status of the expressway is judged based on the dual thresholds of "speed + flow" and the congestion index of the expressway, represents the average travel speed on a highway segment; When the traffic status of the expressway is determined, the congestion index of the expressway and the average travel speed of the expressway section are used to determine the traffic status of the expressway; The expressway traffic operation status includes: Congestion status: , the flow rate is less than b , ; Crowded state: , flow rate is greater than or equal to b , ,or, , ; Slow-moving state: , ; Smooth status: , .

2. The highway traffic operation status prediction method based on deep mining of checkpoint data according to claim 1 is characterized in that: The calculation formula for the average speed of each preset time step is: in, represents the average velocity at each preset time step, Indicates the distance between adjacent gantries on a highway section. Indicates the total number of vehicles in the statistical period. Indicates the The travel time of a vehicle.

3. The highway traffic operation status prediction method based on deep mining of checkpoint data according to claim 1 is characterized in that: The method calculates the Euclidean distance between the speed data one hour before the prediction moment and the data of the same period in the historical data set based on the current speed data set and the historical data set, and then constructs a similarity set, specifically: Based on the current speed data set, obtain the speed data for each preset time step within one hour before the prediction time; Calculate the Euclidean distance between the speed data of each preset time step within one hour before the forecast time and the data of the same period in the historical data set, and select the one with the smallest Euclidean distance. The speed data in the most similar historical data sets constitutes a similarity set ,in, Indicates the minimum Euclidean distance The most similar historical dataset Speed ​​data, ; The calculation formula of the Euclidean distance is: in, represents the Euclidean distance, represents the number of time steps, Indicates the first hour before the forecast time. Velocity data of a preset time step, Indicates the historical data set j In the sample i The velocity value of each time step, i represents the time step index, Indicates the sample index of the historical dataset.

4. The highway traffic operation status prediction method based on deep mining of checkpoint data according to claim 3 is characterized in that: The predicted speed is calculated based on the speed data in the similarity set, specifically: Extract speed data from similarity sets Follow-up Step speed value , get the candidate prediction value set ; Based on the candidate prediction value set, the weighted average method is used to calculate the prediction speed. The specific calculation formula is: in, Represents the predicted speed value, Represents the speed value in the candidate prediction value set The weight of .

5. The highway traffic operation status prediction method based on deep mining of checkpoint data according to claim 4 is characterized in that: The weight is allocated in inverse proportion to the similarity, and the specific calculation formula is: in, Indicates the first hour before the forecast time. The velocity data of the preset time step is compared with the velocity data of the The Euclidean distance of similar samples, Indicates the first hour before the forecast time. The velocity data of the preset time step is compared with the velocity data of the The Euclidean distance between similar samples.

6. The highway traffic operation status prediction method based on deep mining of checkpoint data according to claim 1 is characterized in that: The method for calculating the free flow speed is as follows: sort the historical speed data of the highway section from highest to lowest, and select the average value of the top ninth as the free flow speed of the highway section. The specific calculation formula is: in, represents the free flow speed, Indicates the number of the first ninth of the historical speed data selected. Indicates the first ninth of the historical speed data. The historical speed, Indicates the total number of historical speed data for highway sections.

Images