Reversible lane effect evaluation method based on breakpoint regression model

Abstract

The invention relates to the technical field of urban traffic planning, in particular to a breakpoint regression model-based reversible lane effect evaluation method, and aims at periods before and after the reversible lane is opened, traffic state data is acquired through an online map data interface, and a breakpoint regression model weighted by a binary processing variable and a multi-kernel function is utilized to evaluate the reversible lane effect. And quantitatively evaluating the influence of the reversible lane policy on the traffic flow speed. According to the method, through combination of global polynomial regression and local linear regression, a polynomial order and a time window are dynamically adjusted, the model robustness is verified, and through kernel function and bandwidth optimization, the reliability of an estimation result is ensured. Researches show that the reversible lane policy significantly increases the traffic flow speed at the breakpoint, and the effect is more significant in a short time window. The statistical significance of the policy effect is verified through robustness test and sensitivity analysis, a scientific basis is provided for an urban traffic management department to optimize reversible lane design and traffic signal control, and the method has high popularization value and practicability.

Description

Technical Field

[0001] This application relates to the technical field of urban traffic planning, and particularly to a method for evaluating the tidal lane effect based on a regression discontinuity model. Background Art

[0002] A tidal lane is a traffic management measure for dynamically adjusting lane resources and has been applied in many cities and regions. For the evaluation of the implementation effect of tidal lanes, the existing research mainly includes three methods: traffic micro-simulation method, traffic flow statistics method, and non-parametric test method. For example, in the article "Evaluation of the Operation Status and Optimization of Control Strategies for Variable Lanes on Fuquan Road in Shanghai", the author used VISSIM simulation to compare and analyze the traffic efficiency after setting variable lanes on Fuquan Road. However, the simulation method has a large demand for data, highly depends on simulation assumptions or macro-statistical analysis, and it is difficult to obtain micro-causal effects. In the article "Evaluation of the Implementation Effect of the Tidal Lane on Chaoyang Road in Beijing", the researchers comprehensively evaluated the changes in traffic flow, speed, etc. before and after the implementation of the tidal lane. However, the traffic flow analysis method is usually limited to qualitative macro-regional effect analysis, it is difficult to capture the changes in causal relationships before and after the implementation of the policy, and it is difficult to quantify the specific effects of the policy. In the article "Evaluation Method for the Operation Efficiency of Variable Guide Lanes at Signalized Intersections", the author used the non-parametric test method to study the impact of the matching status of the functions of variable guide lanes and traffic demands on the driving efficiency at intersections.

[0003] However, when dealing with complex traffic data, the non-parametric test method may ignore some potential non-linear relationships, resulting in the accuracy of the evaluation results being affected. Therefore, a tidal lane evaluation method combining high-frequency traffic data and a regression discontinuity model is needed. By performing kernel function weighted regression on samples on both sides of the breakpoint and conducting multiple continuity verifications for control variables, it provides a scientific basis for the design of tidal lane periods and subsequent signal timing configurations. Summary of the Invention

[0004] In view of the above problems, the present invention proposes a method for obtaining values of traffic flow factors of tidal lanes based on regression discontinuity. By constructing a regression discontinuity model and combining image analysis technology, it quantitatively evaluates the impact of tidal lanes on key traffic flow factors such as vehicle flow speed, can truly reflect the implementation effect of the policy, and conducts a micro-evaluation of the implementation effect of the policy. To achieve the above object, the present application provides the following technical solutions: According to the first aspect of the present invention, the present invention claims protection for a method for evaluating the tidal lane effect based on a regression discontinuity model, including: Determine the tidal lane evaluation variables of the evaluation section and conduct a hypothesis test on the tidal lane evaluation variables; Collect traffic situation data of the tidal lane and other lanes in the evaluation section, extract traffic flow factors of the evaluation section and perform preprocessing; Draw the operation scatter plot of the evaluation section, and perform global regression fitting based on the operation scatter plot to preliminarily observe the jump of the tidal lane evaluation variable at the breakpoint and determine the validity of the breakpoint regression model; Establish a basic breakpoint regression model with the implementation time of the tidal lane as the breakpoint, add high-order terms of the time trend term and the interaction term of the disposal time for polynomial regression fitting, and obtain a global linear breakpoint regression model; Obtain the estimated value of the policy effect through weighted regression of the observed values of the kernel function, determine the optimal bandwidth through the mean square error, and construct a local linear breakpoint regression model; Verify the reliability of the global linear breakpoint regression model and the local linear breakpoint regression model through multi-dimensional verification; Integrate the breakpoint regression model into the traffic management decision-making platform, evaluate the policy effect of the tidal lane according to the LATE effect size and confidence interval obtained from historical data, and generate a decision-making recommendation report.

[0005] Furthermore, determining the tidal lane evaluation variable of the evaluation section and performing a hypothesis test on the tidal lane evaluation variable further includes: Select the opening time and closing time of the tidal lane as breakpoints. The road resource allocation changes before and after the breakpoints. Define the time \(t\) as the grouping variable and the average speed \(v\) of the section as the result variable; Introduce a binary treatment variable , when the tidal lane is in the open state, = 1, when the tidal lane is in the closed state, = 0; Under the actual traffic operation state, the arrival time of vehicles at the evaluation section is random. Drivers cannot deliberately adjust the arrival time according to the opening or closing time of the tidal lane and cannot actively avoid or choose the implementation of the tidal lane policy, meeting the first validity condition of the breakpoint regression model; The second condition for the validity of the breakpoint regression model: there is an obvious jump in the result variable before and after the breakpoint. Verify whether it holds through the jump test of the average speed \(v\) of the section; The third condition for the effectiveness of the breakpoint regression method is that the covariates are continuous before and after the breakpoint. Verify the applicability of the model by performing a stationarity test on other potential factors affecting the traffic flow of the section to ensure that these potential factors will not have a significant interference on the result variable at the breakpoint.

[0006] Furthermore, collecting traffic situation data of the tidal lane and other lanes in the evaluation section, extracting traffic flow factors of the evaluation section and performing preprocessing further includes: The traffic flow factor at least includes the average speed of the evaluated section; The preprocessing at least includes eliminating abnormal data with speeds beyond the reasonable range and complementing missing values of timestamps using cubic spline interpolation.

[0007] Furthermore, the method of plotting the running scatter plot of the evaluated section and performing global regression fitting based on the running scatter plot, and preliminarily observing the jump of the tidal lane evaluation variable at the break point to determine the validity of the break point regression model, further includes: Before performing break point regression, plot a scatter plot with speed as the vertical coordinate and time as the horizontal coordinate, and perform global regression fitting based on the scatter plot; Preliminarily observe whether there is a jump in the tidal lane evaluation variable at the break point; If there is a jump in the tidal lane evaluation variable at the break point, the prerequisite conditions of the break point regression model are met. If there is no jump, there is a possibility that the regression model is not valid.

[0008] Furthermore, the method of establishing a basic break point regression model with the implementation moment of the tidal lane as the break point, adding high-order terms of the time trend term and the interaction term of the disposal time for polynomial regression fitting to obtain a global linear break point regression model, further includes: Taking the moment of the implementation of the tidal lane as the break point, establish a break point regression model, and the formula of the basic model is as follows: ; This model represents the average running speed of vehicles on the section before and after the opening of the tidal lane; where: v t is the average speed of the target section in the t-th minute; α is the intercept term of the break point regression model, indicating the expected value of the dependent variable v when all independent variables are 0 t ; t0 is the break point processing moment; t - t0 is the time trend term; is a binary treatment variable, indicating whether it is in the period after the implementation of the tidal lane. If the break point t0 = 0 is the opening moment of the tidal lane, when t ≥ t0, the value is 1, indicating that the vehicles on the target section receive resource allocation treatment and are located after the implementation of the tidal lane. When t < t0, the value is 0, indicating that the vehicles on the target section do not receive resource allocation treatment and are located before the implementation of the tidal lane; H t is a vector composed of other control variables; is the estimated value of the policy effect at the break point; is the average speed of the target section before the break point along with the driving variable time of the overall change; is the average speed of the target section after the break point along with the driving variable time of the overall change; is the regression coefficient of the control variable, used to verify whether the control variable is related to the treatment effect; is the residual term; Based on the global linear regression, high-order terms of the time trend term t - t0 and the treatment-time interaction term are added for polynomial regression, so as to better fit the change of the average speed v with time t; the global n-order polynomial regression expression of the section average speed v is the following formula (2): .

[0009] Furthermore, obtaining the estimated value of the policy effect through weighted regression of the observed values of the kernel function, determining the optimal bandwidth through the mean square error, and constructing a local linear regression discontinuity model further includes: According to the moment t0 when the tidal lane policy is implemented on a certain date, determine the treatment variable under different grouping variables of the valuation, and only retain the observed values within the bandwidth h near the breakpoint t0; According to the treatment variable , obtain the result variable v t (1) the local linear model of the treatment group and v t (0) the control group; Use the triangular kernel function to weight the observed values within the bandwidth h, and assign higher weights to the data closer to the breakpoint t0; Solve for the local average causal effect LATE, and judge v t (1) and v t (0) difference, to measure the policy effect of the tidal lane. Furthermore, verifying the reliability of the global linear regression discontinuity model and the local linear regression discontinuity model through multiple dimensions further includes: For global polynomial regression, test the sensitivity of the estimated value by adjusting the polynomial order and dynamically shrinking the time window, and require that the treatment effect remains consistent in direction and the fluctuation range is less than 10% when the model complexity changes; For local linear regression, verify the stability of the results when the kernel function is switched or the bandwidth is enlarged / reduced by 50% by comparing different kernel function weight assignment strategies and the bandwidth parameter optimization obtained based on the mean square error minimization or cross-validation method. If the difference in the estimated values does not exceed 2 times the standard error and the confidence intervals overlap, it indicates that the policy effect has strong robustness; Joint verification ensures that the effect sizes under different settings have both statistical significance and economic significance, and visually confirms through the fitting curve that there is no abnormal jump at the breakpoint.

[0010] Further, integrating the regression discontinuity model into the traffic management decision-making platform, evaluating the effect of the tidal lane policy based on the LATE effect size and confidence interval obtained from historical data, and generating a decision-making recommendation report further includes: Lane implementation effect evaluation: LATE value, average vehicle speed before and after the breakpoint, change in congestion length, and statistical significance level; Lane opening time adjustment: Conducting time sensitivity analysis based on the LATE value and recommending the opening time of the tidal lane; Lane resource timing suggestion: Proposing a plan for increasing or decreasing the tidal lane according to the historical congestion pattern; Lane effect tracking requirement: Specifying the data monitoring granularity for the next stage; The report is pushed to the traffic manual review system through a JSON interface, providing a basis for dynamic optimization of the urban tidal lane design.

[0011] This application relates to the technical field of urban traffic planning, and particularly to a method for evaluating the effect of a tidal lane based on a regression discontinuity model. For the time periods before and after the opening of the tidal lane, traffic state data is collected through an online map data interface, and a regression discontinuity model with a binary treatment variable and multi-kernel function weighting is used to quantitatively evaluate the impact of the tidal lane policy on the vehicle flow speed. This method combines global polynomial regression and local linear regression, dynamically adjusts the polynomial order and time window to verify the robustness of the model, and ensures the reliability of the estimation results through kernel function and bandwidth optimization. Research shows that the tidal lane policy significantly improves the vehicle flow speed at the breakpoint, and the effect is more significant within a short time window. Through robustness testing and sensitivity analysis, the statistical significance of the policy effect is verified, providing a scientific basis for optimizing the design of tidal lanes and traffic signal control in urban traffic, and having high promotion value and practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a flowchart of a method for evaluating the effect of a tidal lane based on a regression discontinuity model, which is claimed and protected in an embodiment of this application; Figure 2 It is an electronic map of the evaluation section of a method for evaluating the effect of a tidal lane based on a regression discontinuity model, which is claimed and protected in an embodiment of this application; Figure 3 It is a graph showing the trend change of the average speed in the morning peak of the north-south section of the tidal lane implementation section of a method for evaluating the effect of a tidal lane based on a regression discontinuity model, which is claimed and protected in an embodiment of this application; Figure 4 It is a graph showing the trend change of the average speed in the morning peak of the south-north section of the tidal lane implementation section of another method for evaluating the effect of a tidal lane based on a regression discontinuity model, which is claimed and protected in an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts belong to the scope of protection of the present application.

[0014] The terms "first", "second", and "third" in the present application are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", and "third" may explicitly or implicitly include at least one of such features. In the description of the present application, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically defined. All directional indications (such as up, down, left, right, front, back...) in the embodiments of the present application are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products, or devices.

[0015] Referring to "embodiment" in this context means that a specific feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0016] According to the first embodiment of the present invention, the present invention claims a method for evaluating the tidal lane effect based on a regression discontinuity model, referring to Figure 1 , including: Determine the tidal lane evaluation variables of the evaluation section and perform a hypothesis test on the tidal lane evaluation variables; Collect traffic situation data of the tidal lane and other lanes in the evaluation section, extract the traffic flow factors of the evaluation section and perform preprocessing; Draw the running scatter plot of the evaluation section, and perform global regression fitting based on the running scatter plot to preliminarily observe the jump of the tidal lane evaluation variables at the break point and determine the validity of the regression discontinuity model; A basic breakpoint regression model is established with the implementation time of the tidal lane as the breakpoint. The high-order terms of the time trend item and the interaction item of the treatment time are added for polynomial regression fitting to obtain a global linear breakpoint regression model; The estimated value of the policy effect is obtained through weighted regression of the observed values of the kernel function, and the optimal bandwidth is determined through the mean square error to construct a local linear breakpoint regression model; The reliability of the global linear breakpoint regression model and the local linear breakpoint regression model is verified through multi-dimensional verification; The breakpoint regression model is integrated into the traffic management decision-making platform, and the policy effect of the tidal lane is evaluated based on the LATE effect size and confidence interval obtained from historical data, and a decision-making recommendation report is generated.

[0017] Furthermore, determining the tidal lane evaluation variables of the evaluation section and conducting a hypothesis test on the tidal lane evaluation variables further includes: Select the opening time and closing time of the tidal lane as the breakpoints. The road resource allocation changes before and after the breakpoints. Define the time 𝑡 as the grouping variable and the average speed 𝑣 of the section as the result variable; Introduce a binary treatment variable When the tidal lane is in the open state, = 1, when the tidal lane is in the closed state, = 0; Under the actual traffic operation state, the arrival time of vehicles at the evaluation section is random. Drivers cannot deliberately adjust the arrival time according to the opening or closing time of the tidal lane and cannot actively avoid or choose the implementation of the tidal lane policy, meeting the first validity condition of the breakpoint regression model; The second condition for the effectiveness of the breakpoint regression model: there is an obvious jump in the result variable before and after the breakpoint, which is confirmed by the jump test of the average speed 𝑣 of the section; The third condition for the effectiveness of the breakpoint regression method is that the covariates are continuous before and after the breakpoint. By conducting a stationarity test on other potential factors affecting the traffic flow of the section, it is ensured that the potential factors will not have a significant interference on the result variable at the breakpoint, and the applicability of the model is verified.

[0018] Furthermore, collecting the traffic situation data of the tidal lane and other lanes in the evaluation section, extracting the traffic flow factors of the evaluation section and preprocessing them further includes: The traffic flow factors at least include the average speed of the evaluation section; The preprocessing at least includes removing abnormal data with speeds exceeding the reasonable range and filling in the missing values of the time stamps using cubic spline interpolation.

[0019] Further, the method of plotting the running scatter plot of the evaluation section and performing global regression fitting based on the running scatter plot, and preliminarily observing the jump of the tidal lane evaluation variable at the breakpoint to determine the validity of the breakpoint regression model further includes: Before performing breakpoint regression, plot a scatter plot with speed as the ordinate and time as the abscissa, and perform global regression fitting based on the scatter plot; Preliminarily observe whether there is a jump in the tidal lane evaluation variable at the breakpoint; If there is a jump in the tidal lane evaluation variable at the breakpoint, the prerequisite conditions of the breakpoint regression model are met. If there is no jump, the regression model may not hold.

[0020] Further, the method of establishing a basic breakpoint regression model with the implementation time of the tidal lane as the breakpoint, adding high-order terms of the time trend term and the treatment time interaction term for polynomial regression fitting to obtain a global linear breakpoint regression model further includes: Taking the implementation time of the tidal lane as the breakpoint, establish a breakpoint regression model. The basic model formula is as follows:

[0021] This model represents the average running speed of vehicles on the section before and after the opening of the tidal lane. Where: v t is the average speed of the target section in the t-th minute; α is the intercept term of the breakpoint regression model, indicating the expected value of the dependent variable v when all independent variables are 0 t ; t0 is the breakpoint processing time; t - t0 is the time trend term; is a binary treatment variable, indicating whether it is in the period after the implementation of the tidal lane. If the breakpoint t0 = 0 is the opening time of the tidal lane, when t ≥ t0, the value is 1, indicating that the vehicles on the target section receive resource allocation treatment and are located after the implementation of the tidal lane. When t < t0, the value is 0, indicating that the vehicles on the target section do not receive resource allocation treatment and are located before the implementation of the tidal lane; H t is a vector composed of other control variables; is the estimated value of the policy effect at the breakpoint; is the average speed of the target section before the breakpoint along with the driving variable time of the overall change; is the average speed of the target section after the breakpoint along with the driving variable time of the overall change; is the regression coefficient of the control variable, used to verify whether the control variable is related to the treatment effect; is the residual term; Polynomial regression is performed by adding higher-order terms of the time trend term t - t0 and the treatment-time interaction term on the basis of global linear regression, so as to better fit the change of the average speed v with time t; the global nth-order polynomial regression expression of the section average speed v is the following formula (2): .

[0022] Among them, in this embodiment, the control variables include: whether it is in the peak period, the first-order term of the time trend , the second-order term of the time trend , the third-order term of the time trend .

[0023] Furthermore, the method for obtaining the estimated value of the policy effect by weighted regression of the observed values of the kernel function, determining the optimal bandwidth through the mean square error, and constructing a local linear regression discontinuity model further includes: Determine the estimated value of the treatment variable under different grouping variables according to the implementation time t0 of the tidal lane policy on a certain date, and only retain the observed values within the bandwidth h near the break point t0; According to the treatment variable , obtain the result variable v (1) The local linear model of the treatment group and v t (0) The control group; t (0) The local linear model of the control group; Use the triangular kernel function to weight the observed values within the bandwidth h, and give higher weights to the data closer to the break point t0; Solve for the local average causal effect LATE, and judge the difference between v t (1) and v t (0) to measure the policy effect of the tidal lane.

[0024] Among them, in this embodiment, the treatment variable is defined as: .

[0025] The local linear regression model is:

[0026] Where: represents the result variable affected by the policy at time , and represents the result variable not affected by the policy at time : the counterfactual expected result of the control group at the break point time t0, that is, the theoretical speed value at time t0 assuming that the tidal lane is not implemented.

[0027] ​α1: The expected result of the policy intervention for the treatment group at the break point t0, that is, the theoretical speed value at t0 after the implementation of the tidal lane.

[0028] β0: The time trend coefficient of the control group, representing the natural trend of the speed changing over time when the policy is not implemented.

[0029] β1: The time trend coefficient of the treatment group, reflecting the adjusted trend of the speed changing over time after the implementation of the policy. 0: The counterfactual expected result of the control group at the break point t0, that is, the theoretical speed value at t0 assuming that the tidal lane is not implemented.

[0030] The calculation process of the mean squared error (MSE) is as follows:

[0031] where : The actual observed vehicle speed at the i-th time point; : The model predicted vehicle speed at the i-th time point; is the smoothing penalty term, with a value of 0.5 times the sample variance of the speed, that is:

[0032] is the sample mean of the vehicle speed; is the bandwidth, and the bandwidth is optimized by minimizing the mean squared error , and the calculation formula is:

[0033] is the bandwidth of the local linear regression prediction value.

[0034] The weight calculation of the triangular kernel function is as follows:

[0035] The parameters are estimated using the weighted least squares method (WLS):

[0036] The local average causal effect is calculated as follows: Assume that the limit speed of the control group at is ; the limit speed of the treatment group at is , and the local average causal effect (LATE) is:

[0037] where: LATE represents the impact of the tidal lane policy on the average speed of the road section ; and respectively represent the limit values on both sides of the breakpoint .

[0038] Furthermore, the reliability verification of the global linear breakpoint regression model and the local linear breakpoint regression model through multi-dimensions also includes: For global polynomial regression, by adjusting the polynomial order and dynamically shrinking the time window to test the sensitivity of the estimated values, it is required that the treatment effect remains consistent in direction and the fluctuation amplitude is less than 10% when the model complexity changes; For local linear regression, by comparing different kernel function weight allocation strategies and optimizing the bandwidth parameters obtained based on the mean square error minimization or cross-validation method, the stability of the verification results when the kernel function switches or the bandwidth expands / shrinks by 50% is verified. If the difference in the estimated values does not exceed 2 times the standard error and the confidence intervals overlap, it indicates that the policy effect has strong robustness; The joint verification ensures that the effect sizes under different settings have both statistical significance and economic significance, and visually confirms through the fitted curve that there is no abnormal jump at the breakpoint.

[0039] Furthermore, the integration of the breakpoint regression model into the traffic management decision-making platform, evaluating the policy effect of the tidal lane based on the LATE effect size and confidence interval obtained from historical data, and generating a decision-making recommendation report also includes: Lane implementation effect evaluation: LATE value, average vehicle speed before and after the breakpoint, change in congestion length, and statistical significance level; Lane opening time adjustment: Based on the time sensitivity analysis of the LATE value, recommend the opening time of the tidal lane; Lane resource timing suggestion: Propose a plan for increasing or decreasing the tidal lane according to the historical congestion pattern; Lane effect tracking requirement: Specify the data monitoring granularity for the next stage; The report is pushed to the traffic manual review system through the JSON interface, providing a basis for dynamic optimization of the urban tidal lane design.

[0040] Among them, in this embodiment, the decision-making recommendation report is in JSON format and contains the following fields: ① "policy_effect": LATE value and confidence interval (floating-point array); ② "time_adjust": recommended adjustment amount of the tidal lane opening time (integer, unit: minute); ③ "lane_suggestion": lane increase or decrease suggestion (string type).

[0041] In a specific embodiment of the present invention, refer to Figure 2, select the section of Beijing Road in the north-south direction at the intersection of Beijing Road and Renmin Road in Panlong District, Kunming City, Yunnan Province as the target section. During the morning rush hour, there is a large amount of inbound traffic from north to south on this section, and there are few outbound vehicles from south to north. The traffic flow in the evening rush hour shows the opposite phenomenon, and the traffic tidal phenomenon is obvious. The tidal lane is set in the leftmost lane from south to north on Beijing Road, and the opening time is from 9:00 am to 20:00 pm every day. During this period, there are two left-turn lanes in the north-south direction to relieve the traffic pressure on the south-north section of Beijing Road. However, the opening time will be randomly adjusted according to the real-time road conditions and the traffic flow size.

[0042] Use the Baidu Map Traffic Situation API (interface version v3.0) to obtain the tidal lane data at the intersection of Beijing Road and Renmin Road on November 21, 2023 (Tuesday) through the following steps: GIS Spatial Matching: Take the intersection as the center point (latitude and longitude: 25.0586°N, 102.7234°E) and set a 500-meter buffer zone; Call the Baidu Map POI retrieval interface to filter the south-north (section ID: BJ_0012) and north-south (section ID: BJ_0013) lanes of Beijing Road within the buffer zone; Data Crawling: Request traffic situation data at 5-minute intervals, with a total of 288 calls from 00:00 to 23:55 every day; the request parameters include: "road_id" (section ID), "timestamp" (Unix timestamp), "speed" (average speed, km / h); the data response format is JSON, and the key fields are as follows: ```json { "status": 0, "data": { "road_id": "BJ_0012", "timestamp": 1700535600, "speed": 28.5, "congestion_level": 2 } } ``` Data Storage and Verification: The data is stored in separate files according to the road segment ID (south to north: "southbound.csv", north to south: "northbound.csv"); invalid data is removed (speed <10 km / h or >100 km / h, or abnormal response with status ≠ 0); the timestamp is aligned to the nearest 5 minutes (e.g. 08:03→08:05).

[0043] Data preprocessing: The collected raw data is processed as follows to ensure the effectiveness of the analysis: ① Outlier filtering: remove data points with speed <10km / h (extreme congestion) or >100km / h (speeding abnormality); mark and exclude API responses with abnormal status codes (status≠0); ② Timestamp alignment: use 5 minutes as the time window to align the original timestamp to the nearest hour (such as 08:03→08:05, 23:57→00:00 the next day); generate a continuous time series (example: 2023-11-21 00:00, 00:05, ..., 23:55); ③ Breakpoint regression data preparation: extract the data of the south-to-north section of Beijing Road as the processing group (the direction of tidal lane opening); take the tidal lane opening time (9:00) as the breakpoint ( ), select the time window as the scope of analysis.

[0044] Regression discontinuity analysis Please refer to Figure 3 and 4 The average speed change trend of the south-to-north and north-to-south sections shows that both directions experienced congestion development, peak congestion, and slow recovery during the morning peak. Figure 3 ), from 7:00 to 8:30, the vehicle speed continued to decrease, from about 30 km / h to a minimum of about 23 km / h, indicating that the traffic flow increased rapidly and formed significant congestion; from 8:30 to 9:00, the speed slowly rebounded to below 25 km / h, and the congestion was relieved, but the traffic efficiency was not fully restored; until 9:00-9:30, the speed gradually increased to 25-28 km / h, indicating that the congestion began to subside from 9:00 and basically returned to a smooth state (28-30 km / h) at 9:30. On the north-south section ( Figure 4), from 7:00 to 8:20, the speed drops rapidly and reaches the lowest value (close to 10 km / h) from 8:20 to 8:30, reflecting the most serious congestion state; after 8:30, the speed gradually rises above 20 km / h, remains basically stable after 9:00, and the speed returns to 25 - 30 km / h after 9:15, indicating that the traffic conditions have improved and returned to the normal level. Generally speaking, the congestion characteristics of the morning rush hour reach the peak at 8:30 in the south - to - north direction, and the morning rush hour ends after 9:30. While in the north - to - south direction, it reaches the lowest point at around 8:20 and the morning rush hour ends at around 9:00. At this time, the direction of the tidal lane can be adjusted to make full use of the idle lanes in the north - to - south direction to serve the south - to - north traffic flow and relieve the traffic pressure.

[0045] The present invention conducts a regression discontinuity analysis on the vehicle speed in the south - to - north direction at the opening time of the tidal lane through a regression discontinuity model; the break point t0 is the time node of the policy implementation; to judge the effectiveness of the regression discontinuity model, it is necessary to initially observe whether there is a jump in the result variable v at the break point t0. If the result variable v shows a jump at t0, then this model is effective for this study; if there is no jump, then the results of using this model for research are invalid.

[0046] Taking the opening time of the tidal lane as the break point for regression discontinuity analysis, the global data width is 5 hours before and after the break point. 9:00 is the break point, and the data range is from 6:30 to 11:30. The comparison of the regression data before and after the break point for the south - to - north section is shown in Table 1 below. Table 1 compares the average speed and its statistical distribution of the south - to - north section before and after the opening of the tidal lane (break point 9:00 AM). The data shows that the average speed after the break point drops from 29.36 km / h to 27.92 km / h, the standard deviation shrinks from 3.08 to 0.87, and the minimum value increases to 25.90 km / h, indicating that the speed fluctuation after the break point decreases, the extremely congested situations reduce, and the traffic flow tends to be stable.

[0047] Table 1 Comparison of the regression data before and after the break point for the south - to - north section (taking 9:00 as the break point)

[0048] Furthermore, Table 2 analyzes the sensitivity of the time window (60 / 90 / 120 minutes) and the polynomial degree (1 - 2 times) to the regression results. Under the 60 - minute window, the treatment effect of the first - degree polynomial is 3.367 km / h ( <0.05), while it drops to 3.102 km / h under the 120 - minute window ( (<0.05), the effect size decays by approximately 17.3%, indicating the locality of the policy effect. Expanding the time range will dilute the policy effect, and the long window dilutes the effect size. The sensitivity near the breakpoint needs to be verified through local regression (Table 3). The coefficient of the time trend term is -0.133 in the 60-minute window and turns to 0.055 in the 120-minute window. The sign change reflects the differences in traffic flow dynamics at different times, which also indicates that in practical applications, the bandwidth selection needs to balance the locality and stability of the effect size estimation. After the bandwidth is expanded, the absolute value of the coefficient of the time trend term (e.g., =0.055) decreases, reflecting that the natural traffic trend may be smoother under the long window. The coefficient of the interaction term increases from 0.215 (60 minutes) to 0.327 (120 minutes), further verifying the locality of the policy effect. The positive change in the coefficient of the interaction term also reflects the trend that the marginal effect of the vehicle speed increase gradually decreases after the implementation of the policy, which may be related to the fact that the diversion effect of the tidal lane tends to be stable. The R-square value is stable between 0.73 and 0.76, but the effect size fluctuates by more than 10%. The estimation accuracy needs to be improved through the bandwidth limitation of local regression.

[0049] Table 2 Comparison of regression results under different bandwidth range changes (with 9:00 AM as the breakpoint)

[0050] Furthermore, Table 3 shows the local regression treatment effect estimates of different kernel functions (uniform kernel, triangular kernel, Epanechnikov kernel) at different bandwidths (from 150 minutes to 45 minutes). We found that the Epanechnikov kernel had the highest and most stable effect size at most bandwidths (e.g., 4.217 km / h at a 90-minute bandwidth), followed by the uniform kernel (4.169 km / h), and the triangular kernel had a slightly lower effect size at smaller bandwidths (60 - 50 minutes) (3.824 - 3.555 km / h). The estimated effect sizes of the uniform kernel, triangular kernel, and Epanechnikov kernel were similar at a 60-minute bandwidth, 4.351 km / h (SE = 0.900, p < 0.001), 3.824 km / h (SE = 1.141, p < 0.001), and 4.098 km / h (SE = 1.054, p < 0.001) respectively, and the difference in effect sizes ≤ 0.53 km / h (< 2 times the standard error), verifying the consistency and robustness of the regression results. As the bandwidth decreased (e.g., 45 minutes), the effect size increased to 4.445 km / h, but the standard error expanded to 1.023, conforming to the "variance-bias trade-off" rule, indicating that although more sensitive local effects were captured when the bandwidth was smaller, the estimation accuracy decreased and it might not be robust enough due to insufficient sample size. On the other hand, when the bandwidth increased (e.g., 150 minutes), the effect size decreased to between 3.440 km / h and 3.899 km / h, indicating that a larger bandwidth might smooth out the differences in local policy effects, conforming to the typical regression discontinuity feature of "the smaller the bandwidth, the stronger the local effect". In summary, the Epanechnikov kernel performed best at a medium bandwidth (90 minutes), with an effect size of 4.217 km / h (standard error 0.916), balancing estimation efficiency and stability. The estimated effect sizes of all kernel functions were significantly higher than 3.4 km / h (p < 0.001), indicating that the tidal lane policy significantly improved vehicle speeds, and the confidence intervals did not overlap, supporting the conclusion of the regression discontinuity assessment.

[0051] Table 3 Treatment Effect Estimates of Local Regression under Different Kernel Function Switches (with 9:00 AM as the breakpoint)

[0052] In the specific policy evaluation and analysis of optimization suggestions, including: (1) Quantitative evaluation of policy effects: The local average causal effect (LATE) of the tidal lane policy on the traffic flow speed of the south-north section was evaluated through fitting analysis as follows: Core effect: Based on the analysis of the regression discontinuity model, the local average causal effect (LATE) of the tidal lane policy on the traffic flow speed of the south-north section is significant, and the effect size ranges from 3.0 to 3.8 km / h (p<0.001). After the implementation of the policy, the speed increase ranges from 12% to 18%. Among them, the Epanechnikov kernel function has the optimal effect size (4.217 km / h, SE = 0.916) under the 90-minute bandwidth, taking into account both estimation efficiency and stability, and can be used as the core evaluation basis.

[0053] Heterogeneity analysis: After the breakpoint is shifted forward, the effect size decreases, showing a downward trend, verifying that the accuracy of the opening time point of the tidal lane is crucial for the policy effect. After the bandwidth is expanded to 150 minutes, the effect size decreases to 3.440 km / h, indicating that the effect of the tidal lane policy is significantly local.

[0054] (2)Suggestions for dynamic adjustment of tidal lanes: Opening / closing timing: Combining the traffic flow pattern, it is recommended that the south-north tidal lane during the morning peak can be gradually opened after 8:30 (congestion peak), using the idle lanes during the congestion relief period from north to south for diversion; and gradually closed at 9:30 (after congestion is relieved) to avoid resource idleness.

[0055] Bandwidth selection: It is preferred to use a 90 - 120-minute bandwidth for real-time monitoring to capture the stable interval of the policy effect, avoiding the accuracy fluctuation of a small bandwidth (such as 45 minutes) or the effect dilution of a large bandwidth (such as 150 minutes).

[0056] (3)Model calibration and data monitoring Conventional analysis: The Epanechnikov kernel function is used for policy effect evaluation, and its standard error is the lowest (such as SE = 0.813 under the 120-minute bandwidth), and the overlap degree of the confidence interval is less than 5%.

[0057] Sensitivity test: Conduct cross-validation monthly in combination with the Triangular kernel function (focusing on local sensitivity). If the difference in effect size > 1.0 km / h, start the bandwidth re-optimization process.

[0058] Data closed-loop: Embed the regression discontinuity model into the traffic management platform, collect speed and congestion index data in real time, and recalibrate the breakpoint position (±5-minute dynamic offset tolerance) every quarter.

[0059] Exception handling: If the policy effect is continuously lower than the threshold (<3.0 km / h) for 5 days, automatically trigger the "Lane Configuration Review" process, and check potential problems such as lane markings and signal coordination in combination with historical data.

[0060] In several embodiments provided by this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections between each other can be through some interfaces, indirect couplings or communication connections of devices or units, and can be in electrical, mechanical, or other forms.

[0061] In addition, each functional unit in various embodiments of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. The above is only the implementation manner of this application, and does not limit the patent scope of this application. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of this application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of this application.

[0062] The specific implementation manners of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific implementation manners described above. For those skilled in the art, any equivalent modification or substitution to the invention is also within the scope of this application. Therefore, all equal transformations, modifications, improvements, etc. made without departing from the spirit and principles of this application should be covered by the scope of this application.

Claims

1. A method for evaluating the tidal lane effect based on a regression discontinuity model, characterized in that Including: Determine the tidal lane evaluation variables of the evaluation section, and conduct a hypothesis test on the tidal lane evaluation variables; Collect the traffic situation data of the tidal lane and other lanes in the evaluation section, extract the traffic flow factors of the evaluation section and conduct preprocessing; Draw the operation scatter plot of the evaluation section, and conduct global regression fitting based on the operation scatter plot, and preliminarily observe the jump of the tidal lane evaluation variable at the break point to determine the establishment of the break point regression model; Establish a basic break point regression model with the implementation time of the tidal lane as the break point, add the high-order term of the time trend term and the treatment time interaction term for polynomial regression fitting to obtain a global linear break point regression model; Obtain the estimated value of the policy effect through weighted regression of the observed values of the kernel function, determine the optimal bandwidth through the mean square error, and construct a local linear break point regression model; Verify the reliability of the global linear break point regression model and the local linear break point regression model through multi-dimensional verification; Integrate the break point regression model into the traffic management decision-making platform, evaluate the tidal lane policy effect according to the LATE effect size and confidence interval obtained from historical data, and generate a decision-making recommendation report.

2. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, wherein The determination of the tidal lane evaluation variables of the evaluation section and the hypothesis test on the tidal lane evaluation variables further include: Select the opening time and closing time of the tidal lane as the break points. The road resource allocation changes before and after the break points. Define the time \(t\) as the grouping variable and the average speed \(v\) of the section as the result variable; Introduce a binary treatment variable , when the tidal lane is in the open state, = 1, when the tidal lane is in the closed state, = 0; Under the actual traffic operation state, the arrival time of vehicles at the evaluation section is random. Drivers cannot deliberately adjust the arrival time according to the opening or closing time of the tidal lane, and cannot actively avoid or choose the implementation of the tidal lane policy, meeting the first validity condition of the break point regression model; The second condition for the validity of the break point regression model: there is an obvious jump in the result variable before and after the break point, and it is confirmed whether it holds through the jump test of the average speed \(v\) of the section; The third condition for the effectiveness of the break point regression method is that the covariates are continuous before and after the break point. By conducting a stationarity test on other potential factors affecting the traffic flow of the section, it is ensured that the potential factors will not have a significant interference on the result variable at the break point, and the applicability of the model is verified.

3. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, wherein, The collection of the traffic situation data of the tidal lane and other lanes in the evaluation section, the extraction of the traffic flow factors of the evaluation section and the preprocessing further include: The traffic flow factors at least include the average speed of the evaluation section; The preprocessing at least includes eliminating abnormal data with speeds exceeding the reasonable range and filling in the missing values of the time stamps using cubic spline interpolation.

4. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, wherein The drawing of the operation scatter plot of the evaluation section and the global regression fitting based on the operation scatter plot to preliminarily observe the jump of the tidal lane evaluation variable at the break point to determine the establishment of the break point regression model further include: Before conducting the break point regression, draw a scatter plot with speed as the vertical coordinate and time as the horizontal coordinate, and conduct global regression fitting based on the scatter plot; Preliminarily observe whether there is a jump in the tidal lane evaluation variable at the break point; If there is a jump in the tidal lane evaluation variable at the breakpoint, the prerequisite conditions of the regression discontinuity model are met. If there is no jump, the regression model may not hold.

5. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, wherein Based on the implementation time of the tidal lane as the breakpoint, a basic regression discontinuity model is established, and high-order terms of the time trend term and treatment-time interaction terms are added for polynomial regression fitting to obtain a global linear regression discontinuity model. It also includes: Taking the implementation time of the tidal lane as the breakpoint, a regression discontinuity model is established. The formula of the basic model is as follows: ； This model represents the average running speed of vehicles on the road section before and after the opening of the tidal lane; where: v t is the average speed of the target road section within the t-th minute; α is the intercept term of the regression discontinuity model, indicating the expected value of the dependent variable v when all independent variables are 0 t ; t0 is the breakpoint processing moment; t - t0 is the time trend term; is a binary treatment variable, indicating whether it is in the period after the implementation of the tidal lane. If the breakpoint t0 = 0 is the opening moment of the tidal lane, when t ≥ t0, the value is 1, indicating that the vehicles on the target road section receive resource allocation treatment and are located after the implementation of the tidal lane. When t < t0, the value is 0, indicating that the vehicles on the target road section do not receive resource allocation treatment and are located before the implementation of the tidal lane; H t is a vector composed of other control variables; is the estimated value of the policy effect at the breakpoint; is the average speed of the target road section before the breakpoint along with the driving variable time of the overall change; is the average speed of the target road section after the breakpoint along with the driving variable time of the overall change; is the regression coefficient of the control variable, used to verify whether the control variable is related to the treatment effect; is the residual term; Based on the global linear regression, high-order terms of the time trend term \(t - t_0\) and treatment-time interaction terms are added for polynomial regression to better fit the change of the average speed \(v\) with time \(t\). The global \(n\)-th order polynomial regression expression of the section average speed \(v\) is the following formula (2): 。 6. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, wherein The weighted regression is performed through the observed values of the kernel function to obtain the estimated value of the policy effect, and the optimal bandwidth is determined through the mean square error to construct a local linear regression discontinuity model. It also includes: Determine the estimated values of the treatment variables under different grouping variables according to the moment t0 when the tidal lane policy is implemented on a certain date, and only retain the observed values within the bandwidth h near the breakpoint t0; ​ According to the processing variable , the result variable v is obtained t (1) The local linear model of the treatment group and v t (0) The control group; Using a triangular kernel function Weight the observations within the bandwidth h, giving higher weights to data closer to the break point t0; Solve for the Local Average Treatment Effect (LATE) and determine the difference between v t (1) and v t (0) to measure the policy effect of the tidal lane.

7. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, characterized in that The reliability of the global linear regression discontinuity model and the local linear regression discontinuity model is verified through multiple dimensions. It also includes: For the global polynomial regression, the sensitivity of the estimated value is tested by adjusting the polynomial order and dynamically shrinking the time window, and it is required that the treatment effect remains consistent in direction and the fluctuation range is less than 10% when the model complexity changes. For the local linear regression, the stability of the verification results when the kernel function is switched or the bandwidth is expanded / shrunk by 50% is verified by comparing different kernel function weight allocation strategies and the bandwidth parameter optimization obtained based on the mean square error minimization or cross-validation method. If the difference in the estimated values does not exceed 2 times the standard error and the confidence intervals overlap, it indicates that the policy effect has strong robustness. The joint verification ensures that the effect sizes under different settings have both statistical significance and economic significance, and it is visually confirmed through the fitting curve that there is no abnormal jump at the breakpoint.

8. The tidal lane effect evaluation method based on the regression discontinuity model according to claim 1, characterized in that The regression discontinuity model is integrated into the traffic management decision-making platform, and the policy effect of the tidal lane is evaluated based on the LATE effect size and confidence interval obtained from historical data to generate a decision-making recommendation report. It also includes: Lane implementation effect evaluation: LATE value, average vehicle speed before and after the breakpoint, change in congestion length, and statistical significance level; Lane opening time adjustment: Time sensitivity analysis based on the LATE value to recommend the opening time of the tidal lane; Lane resource timing suggestion: Propose a plan for increasing or decreasing the tidal lane according to the historical congestion pattern; Lane effect tracking requirement: Specify the data monitoring granularity in the next stage; The report is pushed to the traffic manual review system through the JSON interface to provide a basis for the dynamic optimization of the urban tidal lane design.

Images