Bidirectional multi-lane expressway traffic flow dynamic control method and system based on traffic simulation

Abstract

The invention discloses a two-way multi-lane highway traffic flow dynamic control method and system based on traffic simulation, and the method comprises the steps: firstly constructing a virtual sensor network covering two-way multiple lanes through a traffic simulation platform, carrying out the total factor perception of a traffic flow, and obtaining the collection data; secondly, abnormal values in the collected data are filtered in a mode of combining a double-sliding-window mechanism and an IQR algorithm, and filtered data are obtained; according to the invention, a function of performing real-time dynamic control on the expressway traffic flow by constructing a closed-loop optimization system of perception, decision, control and evaluation is realized, and accurate perception, intelligent decision and cooperative control can be performed on the traffic flow by fusing dynamic traffic simulation and a real-time data driving technology. Meanwhile, the traffic flow evolution trend pre-judgment capability can be improved through a time-space coupled traffic state prediction model, the overall traffic capacity of the expressway is improved, the life and property safety of road users is guaranteed, and the method is suitable for being widely popularized and used.

Description

Technical Field

[0001] The present invention relates to the technical field of highway traffic flow control, and specifically relates to a dynamic control method and system for two-way multi-lane highway vehicle flow based on traffic simulation. Background Art

[0002] With the acceleration of the urbanization process and the continuous growth of the motor vehicle ownership, the highway traffic flow has shown an explosive growth, and the traffic pressure faced by two-way multi-lane highways is increasing day by day.

[0003] At present, existing traffic control methods generally rely on fixed parameter models and are difficult to adapt to the dynamically changing traffic environment, resulting in frequent traffic jams, low traffic efficiency and prominent safety hazards on highways, and lacking in-depth excavation of the spatio-temporal dynamic characteristics of traffic flow. It is not only difficult to accurately predict the evolution trend of traffic states, but also unable to respond to emergencies and random disturbances in real time. At the same time, there are also problems such as insufficient accuracy in microscopic traffic behavior modeling and difficulty in reflecting real driving behavior characteristics. Therefore, it is necessary to design a dynamic control method and system for two-way multi-lane highway vehicle flow based on traffic simulation. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art, and to better and effectively solve the problems that existing traffic control methods generally rely on fixed parameter models and are difficult to adapt to the dynamically changing traffic environment, resulting in frequent traffic jams, low traffic efficiency and prominent safety hazards on highways, and lacking in-depth excavation of the spatio-temporal dynamic characteristics of traffic flow. It is not only difficult to accurately predict the evolution trend of traffic states, but also unable to respond to emergencies and random disturbances in real time. At the same time, there are also problems such as insufficient accuracy in microscopic traffic behavior modeling and difficulty in reflecting real driving behavior characteristics. A dynamic control method and system for two-way multi-lane highway vehicle flow based on traffic simulation are provided, which perform hybrid traffic control on random vehicle flow and focused route vehicle flow for the two-way multi-lane highway scenario, and then real-time collect multi-dimensional data such as vehicle flow position, speed, and lane occupancy rate through simulation virtual sensors and evaluate the traffic state after Kalman filtering and fuzzy logic processing. Then, a time-varying Poisson process is used to dynamically adjust the generation rate of random vehicle flow, and for the focused route vehicle flow, the departure time sequence, lane allocation and speed initialization are managed through a priority queue, and combined with an improved MOBIL algorithm and PID control to achieve lane-changing decision-making and following distance optimization. It also supports hierarchical enabling of the emergency lane and conflict resolution mechanism. Subsequently, the generation parameters and control model are adjusted in real time through closed-loop feedback, improving the road traffic efficiency and safety in complex vehicle flow scenarios, and providing a simulation-driven solution for highway intelligent traffic control.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: A dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation, comprising the following steps: Step A: Use a traffic simulation platform to construct a virtual sensor network covering two-way multi-lanes to perform all-element perception on the vehicle flow and obtain the collected data; Step B: Use a combination of a double sliding window mechanism and the IQR algorithm to filter out the outliers in the collected data to obtain the filtered data; Step C: Construct a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain the enhanced data; Step D: Based on the enhanced data, establish a hierarchical generation control framework and refined control of microscopic driving behaviors to optimize the control of the dynamic vehicle flow and obtain the control optimization result; Step E: Perform full-process control logic and closed-loop feedback on the control optimization result and establish emergency scenario enhanced control rules to complete the dynamic control operation of the vehicle flow on the two-way multi-lane highway.

[0006] For the aforementioned dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation, in Step A, use a traffic simulation platform to construct a virtual sensor network covering two-way multi-lanes to perform all-element perception on the vehicle flow and obtain the collected data. Among them, the all-element perception includes traffic flow dynamic monitoring, road environment perception, and vehicle generation characteristics. The specific steps are as follows: Step A1: Traffic flow dynamic monitoring, specifically deploy virtual induction coils on each section, and then use the instantaneous vehicle number collected per second on the section , lane occupancy and headway to establish a real-time traffic flow matrix , where is the time window, and the lane occupancy λ is as shown in formula (1): (1) Among them, is the length of the th vehicle, is the lane length, is the number of lanes; Step A2: Road environment perception, specifically integrate the weather simulation module to dynamically update the road surface friction coefficient , and then monitor the occupancy rate of the emergency lane through the virtual camera of the emergency lane ; Step A3: Vehicle generation characteristics. Specifically, the vehicle generation characteristics are divided into random vehicle flow and focused vehicle flow. The random vehicle flow is specifically to set a generator at the road network boundary node to record the generation point coordinates , initial speed and vehicle type ratio. The focused vehicle flow is specifically for the target route Collect the expected lane sequence of the vehicle .

[0007] For the above-mentioned two-way multi-lane freeway vehicle dynamic control method based on traffic simulation, in step B, the double sliding window mechanism and the IQR algorithm are combined to filter the outliers in the collected data to obtain the filtered data, where the double sliding window mechanism includes a short window and a long window , the short window is used to detect sudden anomalies, and the long window is used to identify trend noise. The specific steps are as follows Step B1, initialize the short window and the long window and store the data of the most recent set number of seconds respectively. Then add each newly collected data point to the short window and the long window and remove the oldest data point in the window to keep the sizes of the short window and the long window unchanged; Step B2, calculate the statistical characteristics of the data in the short window and the long window. Specifically, determine whether the new data point in the short window causes a traffic surge exceeding the threshold. If so, determine that the data point is a sudden anomaly. Then compare the new data point in the long window with the statistical characteristics of the data in the long window to determine whether there is trend noise; Step B3, apply the IQR algorithm to the time headway . Specifically, sort the time headway data, and then calculate the interquartile range and the third quartile as shown in formula (2), ; (2); Step B4, determine the valid interval. For the data points in the time headway data that exceed the valid interval, they are regarded as invalid data; Step B5, fill in the invalid data using the linear interpolation method. The linear interpolation method specifically calculates the value of the invalid data point according to the values of adjacent valid data points.

[0008] For the above-mentioned two-way multi-lane freeway vehicle dynamic control method based on traffic simulation, in step C, a multi-dimensional vector containing spatio-temporal features is constructed to enhance the feature vectors in the filtered data and obtain the enhanced data. The multi-dimensional vector is used to provide richer information for subsequent traffic flow analysis and control, as shown in formula (3), (3) where is the multi-dimensional vector, is the vehicle generation time, and the vehicle generation time Used to record the specific time when a vehicle is generated in the road network and reflect the time distribution characteristics of the traffic flow, is the vehicle type, is the target route, and the target route is used to analyze the flow direction and destination distribution of the traffic flow, is the initial lane, is the target lane, is the initial density of the section where the generation point is located, and the initial density of the section where the generation point is located is used to predict the impact of traffic flow merging on the traffic conditions of the section.

[0009] The above-mentioned dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation, step D, based on the enhanced data, establish a hierarchical generation control framework and refined control of microscopic driving behavior to optimize the control of the dynamic traffic flow and obtain the control optimization result. The specific steps are as follows: Step D1, establish a hierarchical generation control framework, specifically construct background traffic flow and focused traffic flow to conduct refined management of the traffic flow on the two-way multi-lane highway; Step D2, establish refined control of microscopic driving behavior, specifically establish a lane usage cost function weighted by time and space and achieve dynamic equilibrium distribution of the traffic flow. The refined control of microscopic driving behavior includes a lane-changing decision model and an adaptive car-following model.

[0010] The above-mentioned dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation, the specific steps of step D1 are as follows: Step D11, generate background traffic flow using a random generation mode, where the background traffic flow is specifically modeled based on a time-varying Poisson process, and the generation rate dynamic adjustment process is shown in formula (4), (4) Where, is the vehicle generation rate at time t, is the peak period reference value, is the congestion sensitivity, is the smoothing coefficient, is the traffic density at time t, is the critical traffic density, and tanh is the hyperbolic tangent function; Step D12, generate focused traffic flow using a directional generation mode, specifically use a priority queue to manage the vehicle generation of the target route and ensure that the traffic flow merges in an orderly manner according to the preset lane sequence. The specific steps are as follows: Step D121, construct a departure time sequence control, specifically use a safe car-following time interval model, as shown in formula (5), (5) wherein, is the safe departure time of the following vehicle, is the time when the leading vehicle passes the same reference point, is the baseline of the safe time headway, is the relative speed difference between the following vehicle and the leading vehicle, is the magnitude of the maximum acceleration or deceleration that the vehicle can achieve; Step D122, construct an initial lane allocation algorithm, specifically, establish a lane selection cost function, as shown in formula (6): (6) wherein, is the vehicle selection cost, is the real-time density of the target lane, is the emergency lane penalty term, is the constant offset term, is the lane offset weight, is the current lane; Step D123, calculate the optimal lane, as shown in formula (7): (7) wherein, is the optimal lane, and L is the set of available lanes; Step D124, establish a speed initialization strategy, specifically, dynamically adjust the initial speed according to the real-time congestion degree, as shown in formula (8): (8) wherein, is the maximum speed of the vehicle under ideal conditions, is the maximum density when the traffic is completely blocked, is the shape index, is the vehicle type correction term.

[0011] For the above-mentioned dynamic control method of vehicle flow on a two-way multi-lane expressway based on traffic simulation, the specific steps of Step D2 are as follows: Step D21, construct a lane-changing decision model, and the specific steps are as follows: Step D211, introduce the lane-changing condition of the emergency lane yielding rule, as shown in formula (9): (9) wherein, is the expected acceleration of the vehicle after changing lanes to the target lane, is the acceleration of the vehicle in the current lane, and are the accelerations of the leading vehicle and the following vehicle in the target lane respectively; Step D212, set the emergency lane exit resistance coefficient, which is used to prevent vehicles from frequently entering and exiting the emergency lane and ensure the efficient use of the emergency lane. Specifically, when a vehicle needs to return from the emergency lane to the normal lane, it must not only meet the basic conditions for changing lanes on the normal lane but also overcome the emergency lane exit resistance coefficient; Step D213, establish an emergency lane occupancy prohibition mechanism, which is used to ensure the unobstructedness of the emergency lane in case of emergency. Specifically, when the current vehicle density is reached, vehicles are prohibited from occupying the emergency lane; Step D214, establish a lane-changing decision model and specific lane-changing decision algorithms. The specific steps are as follows: Step D2141, set the input and output of the lane-changing decision model. The input of the lane-changing decision model includes the information of the lane where the vehicle is currently located, the driving direction, the destination information, the vehicle type, the current road density, and the real-time traffic condition information. The output of the lane-changing decision model is the decision result of whether to change lanes and the target lane; Step D2142, initialize the decision result of whether to change lanes to False and the target lane to None, and then perform lane-changing demand detection. If there is a lane-changing demand, evaluate the lane change to the target lane; The specific process of the lane-changing demand detection is as follows: First, calculate the remaining distance to the target exit. If the remaining distance is less than 500 meters and the current lane is not the lane where the target exit is located, a lane-changing demand is triggered. Then, obtain the traffic flow density and the speed of the vehicle in front on the current lane. If the traffic flow density is greater than 1.2 times the critical density and the speed of the vehicle in front is less than the threshold, a lane-changing demand is triggered. Subsequently, check whether there are obstacles in front. If there are, a lane-changing demand is triggered; The specific process of evaluating the lane change to the target lane is as follows: First, screen the target lanes according to the driving direction and the destination. Then, for each target lane, obtain the distance to the vehicle in front, the speed of the vehicle in front, the acceleration of the vehicle in front, the distance to the vehicle behind, the speed of the vehicle behind, the acceleration of the vehicle behind, the vehicle type, and the current density on the target lane. Then, calculate the target lane evaluation index, which includes the expected acceleration after changing lanes on the target lane; Step D2143, judge the lane-changing conditions for the target lane. If multiple target lanes meet the lane-changing conditions, select the target lane with the greatest lane-changing benefit; The specific process of judging the lane - changing conditions for the target lane is as follows: First, calculate the lane - changing benefit, which is the difference between the expected acceleration after lane - changing and the current acceleration. Then, calculate the yielding cost and consider the accelerations of the vehicles in front of and behind in the target lane. Next, if the target lane is the emergency lane or if the vehicle type is a car and the current density is less than the critical density, then the target lane does not meet the lane - changing conditions; otherwise, add an additional resistance value to the yielding cost. Subsequently, if the lane - changing benefit is greater than the yielding cost, then the target lane meets the lane - changing conditions. Finally, if multiple target lanes meet the lane - changing conditions simultaneously, select the target lane with the largest lane - changing benefit and update the decision result of whether to change lanes to and the ID of the selected target lane; Step D22, establish an adaptive car - following model. The specific steps are as follows, Step D221, dynamically calculate the car - following distance, as shown in formula (10), (10) where, is the minimum distance that the vehicle needs to maintain from the vehicle in front at time t, is the instantaneous speed of the vehicle at time t, is the time interval maintained between the vehicle and the vehicle in front, is the maximum deceleration, is the minimum safety distance; Step D222, calculate the acceleration control, as shown in formula (11), (11) where, is the acceleration of the vehicle at time t, is the error, 、 and are all PID parameters.

[0012] For the aforementioned dynamic control method of two - way multi - lane highway vehicle flow based on traffic simulation, in step E, perform full - process control logic and closed - loop feedback on the control optimization results and establish emergency - scenario enhanced control rules to complete the dynamic control operation of two - way multi - lane highway vehicle flow. The specific steps are as follows, Step E1, perform full - process control logic and closed - loop feedback on the control optimization results. The specific steps are as follows, Step E11, data acquisition and state evaluation. Specifically, use the simulation sensor network to obtain the traffic flow, density, and vehicle - type distribution data of all lanes in real time, and then calculate the congestion index of each section and classify it into unobstructed, slightly congested, moderately congested, and severely congested states. And the section congestion index is as shown in formula (12), (12) Among them, if it is in an unobstructed state, and if it is in a slightly congested state, and if it is in a moderately congested state, and if it is in a severely congested state; Step E12, traffic flow generation mode decision. If it is in an unobstructed or slightly congested state, enable the random generation mode and generate background traffic flow at the reference rate. If it is in a moderately or severely congested state, activate the directional generation mode to generate focused traffic flow and give priority to ensuring the traffic flow on the target route to merge, and at the same time open the emergency lane; Step E13, dynamic configuration of generation parameters, specifically including parameter configuration of the random generation mode and parameter configuration of the directional generation mode. The specific steps are as follows. Step E131, configure the parameters of the random generation mode. Specifically, adjust the vehicle generation rate according to the real-time density and maintain the density balance of each lane; Step E132, configure the parameters of the directional generation mode. Specifically, generate vehicles according to the priority order of the target route and ensure that the departure time sequence meets the safe following distance. Then, allocate the right to use the emergency lane for large vehicles and preferentially select the upstream lane of the target lane for the initial lane, so as to avoid directly cutting into the inner high-flow lane; Step E14, execute micro-behavior control. The micro-behaviors include lane-changing decision and following control. The lane-changing decision is specifically to update the lane selection every 500 ms and give priority to meeting the lane-changing needs of the focused traffic flow. The following control is specifically to calculate the safe distance in real time and adjust the acceleration through a PID controller; Step E15, closed-loop feedback and parameter optimization. The specific steps are as follows. Step E151, statistically calculate the average vehicle speed and lane balance every 10 seconds, where the lane balance is as shown in formula (13). (13) Among them, is the maximum value in the lane density, is the minimum value in the lane density, is the average value of all lane densities; Step E152, if the average vehicle speed and the lane balance , then trigger the self-adjustment of PID parameters as shown in formula (14). (15); Step E2: Establish emergency scenario enhanced control rules to complete the dynamic control operation of vehicle flow on a two-way multi-lane highway, where the emergency scenario enhanced control rules include an emergency lane hierarchical activation mechanism and a conflict resolution strategy.

[0013] For the above-mentioned dynamic control method of vehicle flow on a two-way multi-lane highway based on traffic simulation, the specific steps of Step E2 are as follows: Step E21: Establish an emergency lane hierarchical activation mechanism. Specifically, when in a moderately congested state, trucks or buses are allowed to use the emergency lane with a maximum speed limit of 90 km / h, and lane changes are prohibited. When in a severely congested state, the emergency lane is open to all vehicle types with a speed limit of 80 km / h, and a forced lane change point is set every 2 kilometers to guide vehicles back to the normal lane. Priority vehicles in the focused traffic flow can use the emergency lane throughout the journey, and the priority vehicles include ambulances and buses. Step E22: Construct a conflict resolution strategy. Specifically, when a vehicle in the emergency lane needs to return to the normal lane, it turns on the turn signal in advance at the response distance threshold and sends a lane change request to surrounding vehicles. If a vehicle in the normal lane detects the request and its own safety distance permits, it will actively decelerate and give way. If the lane change is not completed within the specified time, an emergency lane forced speed limit mechanism will be triggered.

[0014] A dynamic control system for vehicle flow on a two-way multi-lane highway based on traffic simulation includes a data acquisition module, a data filtering module, a data enhancement module, a control optimization module, and a dynamic control module. The data acquisition module is used to construct a virtual sensor network covering the two-way multi-lane using a traffic simulation platform to perform all-element perception of vehicle flow and obtain acquisition data. The data filtering module is used to filter out outliers in the acquisition data using a combination of a double sliding window mechanism and the IQR algorithm to obtain filtered data. The data enhancement module is used to construct a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain enhanced data. The control optimization module is used to establish a hierarchical generation control framework and refined control of microscopic driving behaviors based on the enhanced data to optimize the control of dynamic vehicle flow and obtain a control optimization result. The dynamic control module is used to perform full-process control logic and closed-loop feedback on the control optimization result and establish emergency scenario enhanced control rules to complete the dynamic control operation of vehicle flow on a two-way multi-lane highway.

[0015] The beneficial effects of the present invention are: (1) First, the present invention uses a traffic simulation platform to construct a virtual sensor network covering two-way multi-lane roads to comprehensively perceive vehicle flows and obtain acquisition data. Then, it combines a double sliding window mechanism and an IQR algorithm to filter outlier values in the acquisition data to obtain filtered data. Subsequently, it constructs a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain enhanced data. Then, based on the enhanced data, it establishes a hierarchical generation control framework and refined control of microscopic driving behaviors to optimize the control of dynamic vehicle flows and obtain control optimization results. Finally, it conducts full-process control logic and closed-loop feedback on the control optimization results and establishes emergency scenario enhanced control rules to complete the dynamic control operation of two-way multi-lane highway vehicle flows; effectively realizing that the vehicle flow dynamic control method and system have the function of constructing a closed-loop optimization system for perception, decision-making, control, and evaluation for the complex traffic characteristics of two-way multi-lane highways to perform real-time dynamic control on highway vehicle flows. Moreover, through the integration of dynamic traffic simulation and real-time data-driven technology, it can accurately perceive vehicle flows, make intelligent decisions, and perform collaborative control. At the same time, through a spatio-temporal coupled traffic state prediction model, it can improve the ability to predict the evolution trend of traffic flows. And by designing a multi-objective collaborative optimization control strategy, it can balance efficiency, safety, and fairness. The present invention not only supports the rapid iterative optimization of control strategies but also can perform online optimization and adjustment of control parameters by constructing a virtual-real combined simulation verification platform.

[0016] (2) The present invention conducts mixed traffic control of random vehicle flows and focused route vehicle flows for the two-way multi-lane highway scenario. Then, it uses simulation virtual sensors to collect multi-dimensional data such as vehicle flow position, speed, and lane occupancy rate in real time and evaluates the traffic state after Kalman filtering and fuzzy logic processing. Then, it dynamically adjusts the generation rate of random vehicle flows using a time-varying Poisson process. For focused route vehicle flows, it manages the departure timing, lane allocation, and speed initialization through a priority queue and combines an improved MOBIL algorithm and PID control to achieve lane-changing decisions and following distance optimization. It also supports the hierarchical enabling and conflict resolution mechanism of the emergency lane. Subsequently, it adjusts the generation parameters and control models in real time through closed-loop feedback, improving the road traffic efficiency and safety in complex vehicle flow scenarios. Effectively realizing that the vehicle flow dynamic control method and system have the function of quickly reducing traffic congestion by precisely controlling vehicle flows and focused route vehicle flows. It not only improves the overall traffic capacity of the highway but also shortens the driving time of vehicles. Moreover, through reasonable vehicle flow allocation and control, it can reduce conflicts between vehicles, not only reducing the probability of traffic accidents but also ensuring the life and property safety of road users. At the same time, by using the simulation platform for research and testing, it avoids a large number of actual road tests, greatly reducing the cost and risk of traffic control strategy research and development. This provides new ideas and methods for the intelligent development of highway traffic control and helps to promote the entire traffic field towards a more intelligent and efficient direction. Brief Description of the Drawings

[0017] Figure 1 is the overall flowchart of a dynamic control method for vehicle flow on a two-way multi-lane expressway based on traffic simulation of the present invention. Detailed Implementation Manner

[0018] The present invention will be further described below in conjunction with the drawings of the specification.

[0019] As Figure 1 shown, a dynamic control method for vehicle flow on a two-way multi-lane expressway based on traffic simulation of the present invention includes the following steps Step A: Use a traffic simulation platform to construct a virtual sensor network covering two-way multi-lanes to perform all-element perception on vehicle flow and obtain acquisition data. The all-element perception includes traffic flow dynamic monitoring, road environment perception, and vehicle generation characteristics. The specific steps are as follows Step A1: Traffic flow dynamic monitoring. Specifically, virtual induction coils are deployed on each section, and then the instantaneous vehicle number collected per second 、lane occupancy and headway are used to establish a real-time traffic flow matrix , where is the time window, and the lane occupancy λ is as shown in formula (1) (1) where is the length of the th vehicle is the lane length is the number of lanes Step A2: Road environment perception. Specifically, a weather simulation module is integrated to dynamically update the road surface friction coefficient , and then the emergency lane occupancy rate is monitored through a virtual camera in the emergency lane ; Preferably, the dry road surface friction coefficient = 0.7, the friction coefficient in rainy and snowy days = 0.3, and when a warning is triggered Step A3: Vehicle generation characteristics. The vehicle generation characteristics are specifically divided into random vehicle flow and focused vehicle flow. The random vehicle flow is specifically to record the generation point coordinates , initial speed and vehicle type ratio by setting generators at the network boundary nodes. The focused vehicle flow is specifically to collect the expected lane sequence of vehicles for the target route .

[0020] Preferably, the initial speed Subject to a normal distribution , the vehicle type proportions are specifically 70% for cars, 20% for trucks, and 10% for buses. The target route includes a continuous path from the entrance ramp to the exit ramp, and the desired lane sequence of the vehicle includes a lane-changing demand that gradually converges from the outer lane to the inner lane; Step B: Filter the outliers in the collected data by combining the double sliding window mechanism and the IQR algorithm to obtain the filtered data. The double sliding window mechanism includes a short window and a long window . The short window is used to detect sudden anomalies, and the long window is used to identify trend noises. The specific steps are as follows Preferably seconds seconds; Step B1: Initialize the short window and the long window and store the data of the most recent set number of seconds respectively. Then add each newly collected data point to the short window and the long window and remove the oldest data point in the window to keep the sizes of the short window and the long window unchanged; Preferably, the short window of the most recent set number of seconds is seconds, and the long window is seconds; Step B2: Calculate the statistical characteristics of the data in the short window and the long window. Specifically, determine whether the new data point in the short window causes a traffic surge exceeding the threshold. If so, determine that the data point is a sudden anomaly. Then compare the new data point in the long window with the statistical characteristics of the data in the long window to determine whether there is a trend noise; Step B3: Apply the IQR algorithm to the time headway . Specifically, sort the time headway data, and then calculate the interquartile range and the third quartile as shown in formula (2) (2); Step B4: Determine the valid interval. For the data points in the time headway data that exceed the valid interval, they are regarded as invalid data; Preferably, the valid interval is ; Step B5: Fill in the invalid data using the linear interpolation method. The linear interpolation method specifically calculates the value of the invalid data point based on the values of adjacent valid data points.

[0021] ​Step C: Construct a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain enhanced data. The multi-dimensional vector is used to provide richer information for subsequent traffic flow analysis and control, as shown in Equation (3). (3) where is the multi-dimensional vector, is the vehicle generation time, and the vehicle generation time is used to record the specific moment when the vehicle is generated in the road network and reflect the time distribution characteristics of the traffic flow, is the vehicle type, is the target route, and the target route is used to analyze the flow direction and destination distribution of the traffic flow, is the initial lane, is the target lane, is the initial density of the section where the generation point is located, and the initial density of the section where the generation point is located is used to predict the impact of the traffic flow merging on the traffic conditions of the section.

[0022] Step D: Based on the enhanced data, establish a hierarchical generation control framework and refined control of microscopic driving behaviors to optimize the control of the dynamic traffic flow and obtain the control optimization result. The specific steps are as follows Step D1: Establish a hierarchical generation control framework, specifically by constructing background traffic flow and focused traffic flow to conduct refined management of the traffic flow on a two-way multi-lane highway. The specific steps are as follows Step D11: Generate the background traffic flow using a random generation mode. The background traffic flow is specifically modeled based on a time-varying Poisson process, and the generation rate dynamic adjustment process is as shown in Equation (4). (4) where is the vehicle generation rate at time t, is the peak period reference value, is the congestion sensitivity, is the smoothing coefficient, is the traffic density at time t, is the critical traffic density, and tanh is the hyperbolic tangent function; Preferably, , , when , is increased to to fill the empty driving sections; Step D12: Generate the focused traffic flow using a directional generation mode, specifically by using a priority queue to manage the target route The vehicle generates and ensures that the vehicle flow merges orderly according to the preset lane sequence. The specific steps are as follows: Step D121, construct the departure time sequence control. Specifically, adopt the safe following time distance model, as shown in formula (5): (5) where: is the safe departure time of the following vehicle; is the time when the leading vehicle passes the same reference point; is the baseline of the safe time distance; is the relative speed difference between the following vehicle and the leading vehicle; is the magnitude of the maximum acceleration or deceleration that the vehicle can achieve; Preferably, seconds, and if seconds, then force a delayed departure to avoid collision; Step D122, construct the initial lane allocation algorithm. Specifically, establish a lane selection cost function, as shown in formula (6): (6) where: is the vehicle selection cost; is the real-time density of the target lane; is the emergency lane penalty term; is the constant offset term; lane offset weight; is the current lane; Preferably, , , ; Step D123, calculate the optimal lane, as shown in formula (7): (7) where: is the optimal lane, and L is the set of available lanes; Step D124, establish a speed initialization strategy. Specifically, dynamically adjust the initial speed according to the real-time congestion degree, as shown in formula (8): (8) where: is the maximum speed of the vehicle under ideal conditions; is the maximum density when the traffic is completely blocked; is the shape index; is the vehicle type correction term; Preferably, , , 。

[0023] Step D2, establish refined control of microscopic driving behaviors, specifically, establish a lane usage cost function with spatio-temporal weighting and achieve dynamic equilibrium distribution of traffic flow. The refined control of microscopic driving behaviors includes a lane-changing decision model and an adaptive car-following model. The specific steps are as follows: Step D21, construct a lane-changing decision model. The specific steps are as follows: Step D211, introduce the lane-changing condition of the emergency lane yielding rule, as shown in formula (9): (9) where: is the expected acceleration after the vehicle changes lanes to the target lane; is the acceleration of the vehicle in the current lane; and are the accelerations of the vehicle in front and behind in the target lane respectively; Step D212, set the emergency lane exit resistance coefficient. The emergency lane exit resistance coefficient is used to prevent vehicles from frequently entering and exiting the emergency lane and ensure the efficient use of the emergency lane. Specifically, when a vehicle needs to return from the emergency lane to the normal lane, it must not only meet the basic conditions for changing lanes in the normal lane but also overcome the emergency lane exit resistance coefficient; where the emergency lane exit resistance coefficient ; Step D213, establish an emergency lane occupancy prohibition mechanism. The emergency lane occupancy prohibition mechanism is used to ensure the unobstructedness of the emergency lane in case of emergencies. Specifically, when the current vehicle density , prohibit vehicles from occupying the emergency lane; where only when the congestion reaches a certain level, are small cars allowed to reasonably use the emergency lane according to the actual situation; for trucks and large buses, although they are allowed to use the emergency lane when necessary, in order to avoid long-term occupancy affecting emergency response, it is stipulated that their maximum occupancy time limit is t minutes; when the occupancy time reaches t minutes, the vehicle must return to the normal lane as soon as possible.

[0024] Step D214, establish a specific lane-changing decision algorithm for the lane-changing decision model. The specific steps are as follows: Step D2141, set the input and output of the lane-changing decision model. The input of the lane-changing decision model is the information of the lane where the vehicle is currently located, the driving direction, the destination information, the vehicle type, the current road density, and the real-time traffic condition information. The output of the lane-changing decision model is the decision result of whether to change lanes and the target lane; Step D2142, initialize the decision result of whether to change lanes to False and the target lane to None, and then perform a lane-changing demand detection. If there is a lane-changing demand, evaluate the lane change to the target lane; The specific process of detecting the lane change requirement is as follows: First, calculate the remaining distance to the target exit. If the remaining distance is less than 500 meters and the current lane is not the lane where the target exit is located, a lane change requirement is triggered. Then, obtain the traffic flow density of the current lane and the speed of the vehicle in front. If the traffic flow density is greater than 1.2 times the critical density and the speed of the vehicle in front is less than the threshold, a lane change requirement is triggered. Subsequently, check whether there is an obstacle ahead. If there is, a lane change requirement is triggered; The specific process of evaluating the lane change to the target lane is as follows: First, screen the target lanes according to the driving direction and destination. Then, for each target lane, obtain the distance to the vehicle in front, the speed of the vehicle in front, the acceleration of the vehicle in front, the distance to the vehicle behind, the speed of the vehicle behind, the acceleration of the vehicle behind, the type of the vehicle, and the current density on the target lane. Then, calculate the evaluation index of the target lane. The evaluation index of the target lane includes the expected acceleration after changing lanes to the target lane; Step D2143, judge the lane change conditions for the target lane. If multiple target lanes meet the lane change conditions, select the target lane with the greatest lane change benefit; Among them, the specific process of judging the lane change conditions for the target lane is as follows: First, calculate the lane change benefit. The lane change benefit is the difference between the expected acceleration after changing lanes and the current acceleration. Then, calculate the yielding cost and consider the accelerations of the vehicles in front and behind on the target lane. Then, if the target lane is an emergency lane and if the vehicle type is a car and the current density is less than the critical density, this target lane does not meet the lane change conditions. Otherwise, add an additional resistance value to the yielding cost. Subsequently, if the lane change benefit is greater than the yielding cost, this target lane meets the lane change conditions. Finally, if multiple target lanes meet the lane change conditions simultaneously, select the target lane with the greatest lane change benefit and update the decision result of whether to change lanes to and the ID of the selected target lane; Step D22, establish an adaptive car-following model. The specific steps are as follows. Step D221, dynamically calculate the car-following distance as shown in formula (10). (10) Among them, is the minimum distance that the vehicle needs to maintain from the vehicle in front at time t, is the instantaneous speed of the vehicle at time t, is the time interval maintained between the vehicle and the vehicle in front, is the maximum deceleration, is the minimum safety distance; Preferably, = 1.5 seconds, , meters; Step D222, calculate the acceleration control as shown in formula (11). (11) Among them, is the acceleration of the vehicle at time t, is the error, , and are all PID parameters; Preferably, .

[0025] Step E: Perform full-process control logic and closed-loop feedback on the control optimization result and establish an emergency scenario reinforcement control rule to complete the dynamic control operation of the two-way multi-lane highway vehicle flow. The specific steps are as follows. Step E1: Perform full-process control logic and closed-loop feedback on the control optimization result. The specific steps are as follows. Step E11: Data collection and status evaluation. Specifically, use a simulation sensor network to obtain real-time traffic flow, density, and vehicle type distribution data of all lanes, and then calculate the congestion index of each section and divide it into unobstructed, lightly congested, moderately congested, and severely congested states, and the section congestion index is as shown in formula (12). (12) Among them, if , it is in an unobstructed state. If , it is in a lightly congested state. If , it is in a moderately congested state. If , it is in a severely congested state; Step E12: Decision on the traffic flow generation mode. If it is in an unobstructed or lightly congested state, enable the random generation mode and generate background traffic flow at the reference rate. If it is in a moderately or severely congested state, activate the directional generation mode to generate focused traffic flow and give priority to ensuring the traffic flow convergence of the target route, and at the same time open the emergency lane; Step E13: Dynamic configuration of generation parameters, specifically including parameter configuration for the random generation mode and parameter configuration for the directional generation mode. The specific steps are as follows. Step E131: Configure parameters for the random generation mode. Specifically, adjust the vehicle generation rate according to the real-time density and maintain the density balance of each lane; Step E132: Configure parameters for the directional generation mode. Specifically, generate vehicles according to the priority ranking of the target route and ensure that the departure timing meets the safe following distance. Then, allocate the right of use of the emergency lane to large vehicles and preferentially select the upstream lane of the target lane as the initial lane for large vehicles, so as to avoid directly cutting into the inner high-flow lane; Step E14, execute micro-behavior control, where the micro-behaviors include lane-changing decision-making and car-following control. The lane-changing decision-making specifically updates the lane selection every 500 ms and preferentially meets the lane-changing requirements of the focused traffic flow. The car-following control specifically calculates the safe distance in real time and adjusts the acceleration through a PID controller; Step E15, closed-loop feedback and parameter optimization, the specific steps are as follows, Step E151, statistically calculate the average vehicle speed and lane balance every 10 seconds , where the lane balance is as shown in formula (13), (13) where, is the maximum value in the lane density, is the minimum value in the lane density, the average value of all lane densities; Step E152, if the average vehicle speed and the lane balance , then trigger the self-adjustment of PID parameters as shown in formula (14), (15); Step E2, establish an emergency scenario enhanced control rule to complete the dynamic control operation of the traffic flow on a two-way multi-lane expressway. The emergency scenario enhanced control rule includes an emergency lane hierarchical activation mechanism and a conflict resolution strategy. The specific steps are as follows, Step E21, establish an emergency lane hierarchical activation mechanism. Specifically, when in a moderately congested state, trucks or buses are allowed to use the emergency lane with a maximum speed limit of 90 km / h, and lane-changing is prohibited. When in a severely congested state, the emergency lane is open to all vehicle types with a speed limit of 80 km / h, and a forced lane-changing point is set every 2 km to guide vehicles back to the normal lane. Vehicles with higher priority in the focused traffic flow can use the emergency lane throughout the journey. The vehicles with higher priority include ambulances and buses; Step E22, construct a conflict resolution strategy. Specifically, when a vehicle in the emergency lane needs to return to the normal lane, it turns on the turn signal in advance at the response distance threshold and sends a lane-changing request to surrounding vehicles. If a vehicle in the normal lane detects the request and its own safe distance allows, it will actively decelerate and give way. If the lane-changing is not completed within the specified time, the emergency lane forced speed limit mechanism will be triggered.

[0026] A dynamic traffic control system for a two-way multi-lane expressway based on traffic simulation, including a data acquisition module, a data filtering module, a data enhancement module, a control optimization module, and a dynamic control module. The data acquisition module is used to construct a virtual sensor network covering the two-way multi-lane through a traffic simulation platform to perform all-element perception of the traffic flow and obtain acquisition data; The data filtering module is used to filter out the outliers in the collected data by combining the dual sliding window mechanism and the IQR algorithm to obtain the filtered data; The data enhancement module is used to construct a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain the enhanced data; The control optimization module is used to establish a hierarchical generation control framework and refined control of microscopic driving behaviors based on the enhanced data to control and optimize the dynamic traffic flow and obtain the control optimization result; The dynamic control module is used to perform full-process control logic and closed-loop feedback on the control optimization result and establish emergency scenario reinforcement control rules to complete the dynamic control operation of the two-way multi-lane highway traffic flow.

[0027] In summary, for a dynamic traffic flow control method and system for a two-way multi-lane highway according to the present invention, first, a virtual sensor network covering the two-way multi-lane is constructed using a traffic simulation platform to perform all-element perception of the traffic flow and obtain the collected data. Then, the outliers in the collected data are filtered by combining the dual sliding window mechanism and the IQR algorithm to obtain the filtered data. Subsequently, a multi-dimensional vector containing spatio-temporal features is constructed to enhance the feature vectors in the filtered data and obtain the enhanced data. Then, a hierarchical generation control framework and refined control of microscopic driving behaviors are established based on the enhanced data to control and optimize the dynamic traffic flow and obtain the control optimization result. Finally, full-process control logic and closed-loop feedback are performed on the control optimization result and emergency scenario reinforcement control rules are established to complete the dynamic control operation of the two-way multi-lane highway traffic flow; effectively realizing that the dynamic traffic flow control method and system have the function of constructing a closed-loop optimization system for perception, decision-making, control, and evaluation for the complex traffic characteristics of the two-way multi-lane highway to perform real-time dynamic control on the highway traffic flow. And through the integration of dynamic traffic simulation and real-time data-driven technology, it can accurately perceive, intelligently make decisions, and collaboratively control the traffic flow. At the same time, through the spatio-temporal coupled traffic state prediction model, it can improve the ability to predict the evolution trend of the traffic flow. And by designing a multi-objective collaborative optimization control strategy, it can balance efficiency, safety, and fairness. It also realizes that the dynamic traffic flow control method and system have the function of quickly reducing traffic congestion by accurately dynamically controlling the traffic flow and the traffic flow on the focused route, improving the overall traffic capacity of the highway. And through reasonable traffic flow allocation and control, it can reduce conflicts between vehicles and ensure the life and property safety of road users. At the same time, by using the simulation platform for research and testing, a large number of actual road tests are avoided, greatly reducing the cost and risk of traffic control strategy research and development. This provides new ideas and methods for the intelligent development of highway traffic control and helps to promote the entire traffic field towards a more intelligent and efficient direction.

[0028] The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification only illustrates the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation, characterized in that: including the following steps, Step A, constructing a virtual sensor network covering two-way multi-lanes using a traffic simulation platform to perform all-element perception on vehicle flows and obtain collected data; Step B, filtering outliers in the collected data by combining a double sliding window mechanism and an IQR algorithm to obtain filtered data; Step C, constructing a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain enhanced data; Step D, establishing a hierarchical generation control framework and refined control of microscopic driving behaviors based on the enhanced data to optimize the control of dynamic vehicle flows and obtain a control optimization result; Step E, performing full-process control logic and closed-loop feedback on the control optimization result and establishing emergency scenario enhanced control rules to complete the dynamic control operation of two-way multi-lane highway vehicle flows.

2. The dynamic control method for vehicle flow on a two-way multi-lane expressway based on traffic simulation according to claim 1, characterized in that: Step A, constructing a virtual sensor network covering two-way multi-lanes using a traffic simulation platform to perform all-element perception on vehicle flows and obtain collected data, where the all-element perception includes traffic flow dynamic monitoring, road environment perception, and vehicle generation characteristics. The specific steps are as follows. Step A1, traffic dynamic monitoring, specifically deploying virtual induction coils on each road section, and then using the instantaneous vehicle number collected per second on the road section , lane occupancy and headway to establish a real-time traffic flow matrix , where is the time window, and the lane occupancy λ is as shown in formula (1). （1） Among them, is the length of the th vehicle, is the length of the lane, is the number of lanes; Step A2, road environment perception, specifically integrating a weather simulation module to dynamically update the road surface friction coefficient , and then monitoring the occupancy rate of the emergency lane through a virtual emergency lane camera ; Step A3, the vehicle generates features. Specifically, the vehicle-generated features are divided into random traffic flow and focused traffic flow. The random traffic flow is specifically generated by setting a generator at the road network boundary node to record the coordinates of the generation point , the initial speed and the vehicle type ratio. The focused traffic flow is specifically for the target route to collect the expected lane sequence of the vehicle .

3. A dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation according to claim 2, characterized in that: Step B: Filter the outliers in the collected data by combining the dual sliding window mechanism and the IQR algorithm to obtain the filtered data, where the dual sliding window mechanism includes a short window and a long window . The short window is used to detect sudden anomalies, and the long window is used to identify trend noises. The specific steps are as follows Step B1, initializing short and long windows and storing data for the recently set number of seconds respectively, then adding each newly collected data point to the short and long windows and removing the oldest data point in the window to keep the sizes of the short and long windows unchanged; Step B2, calculating the statistical characteristics of the data in the short and long windows. Specifically, it is judged whether the new data point in the short window causes a traffic flow surge exceeding the threshold. If so, the data point is determined to be a sudden anomaly. Then, the new data point in the long window is compared with the statistical characteristics of the data in the long window to judge whether there is trend noise; Step B3, for the headway Apply the IQR algorithm. Specifically, sort the headway data, and then use the first quartile and the third quartile to calculate the interquartile range , as shown in formula (2). （2）； Step B4, determining the effective interval. Among them, for the data points in the headway data that exceed the effective interval, they are regarded as invalid data; Step B5, filling in the invalid data using the linear interpolation method, and the linear interpolation method specifically calculates the value of the invalid data point based on the values of adjacent valid data points.

4. A dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation according to claim 3, characterized in that: Step C, constructing a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain enhanced data, where the multi-dimensional vector is used to provide richer information for subsequent vehicle flow analysis and control, as shown in formula (3); （3） Among them, is a multi-dimensional vector, is the vehicle generation time, and the vehicle generation time is used to record the specific moment when the vehicle is generated in the road network and reflect the time distribution characteristics of the traffic flow, is the vehicle type, is the target route, and the target route is used to analyze the flow direction and destination distribution of the traffic flow, is the initial lane, is the target lane, is the initial density of the section where the generation point is located, and the initial density of the section where the generation point is located is used to predict the impact of the traffic flow merging on the traffic conditions of the section.

5. A dynamic control method for vehicle flow on a two-way multi-lane expressway based on traffic simulation according to claim 4, characterized in that: Step D, establishing a hierarchical generation control framework and refined control of microscopic driving behaviors based on the enhanced data to optimize the control of dynamic vehicle flows and obtain a control optimization result. The specific steps are as follows. Step D1, establishing a hierarchical generation control framework, specifically constructing background vehicle flow and focused vehicle flow to perform refined management on two-way multi-lane highway vehicle flows; Step D2, establishing refined control of microscopic driving behaviors, specifically establishing a lane usage cost function weighted by time and space and realizing dynamic equilibrium distribution of vehicle flows. The refined control of microscopic driving behaviors includes a lane-changing decision model and an adaptive car-following model.

6. The dynamic control method for vehicle flow on a two-way multi-lane expressway based on traffic simulation according to claim 5, wherein: The specific steps of Step D1 are as follows. Step D11, generating background vehicle flow using a random generation mode, where the background vehicle flow is specifically modeled based on a time-varying Poisson process, and the generation rate dynamic adjustment process is as shown in formula (4); （4） wherein, is the vehicle generation rate at time t, is the peak period reference value, is the congestion sensitivity, is the smoothing coefficient, is the traffic density at time t, is the critical traffic density, and tanh is the hyperbolic tangent function; Step D12, generate a focused traffic flow using the directional generation mode. Specifically, use a priority queue to manage the vehicles on the target route to generate and ensure that the traffic flow merges orderly according to the preset lane sequence. The specific steps are as follows: Step D121, constructing a departure time sequence control, specifically using a safe car-following headway model, as shown in formula (5); （5） wherein, is the safe departure time of the following vehicle, is the time when the leading vehicle passes the same reference point, is the safety time headway baseline, is the relative speed difference between the following vehicle and the leading vehicle, is the magnitude of the maximum acceleration or deceleration that the vehicle can achieve; Step D122: Construct the initial lane allocation algorithm. Specifically, establish a lane selection cost function as shown in Equation (6). （6） Among them, is the vehicle selection cost, is the real-time density of the target lane, is the emergency lane penalty term, is the constant offset term, is the lane offset weight, is the current lane; Step D123: Calculate the optimal lane as shown in Equation (7). （7） Among them, is the optimal lane, and L is the set of available lanes; Step D124: Establish a speed initialization strategy. Specifically, dynamically adjust the initial speed according to the real-time congestion level as shown in Equation (8). （8） Among them, the maximum vehicle speed under ideal conditions, the maximum density when traffic is completely congested, is the shape index, is the vehicle type correction term.

7. A dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation according to claim 5, characterized in that: The specific steps of Step D2 are as follows. Step D21: Construct a lane-changing decision model. The specific steps are as follows. Step D211: Introduce the lane-changing condition of the emergency lane yielding rule as shown in Equation (9). （9） wherein, is the expected acceleration after the vehicle changes lanes to the target lane, is the acceleration of the vehicle in the current lane, and are the accelerations of the vehicle in front and behind in the target lane respectively; Step D212: Set the emergency lane exit resistance coefficient. The emergency lane exit resistance coefficient is used to prevent vehicles from frequently entering and exiting the emergency lane and ensure the efficient use of the emergency lane. Specifically, when a vehicle needs to return from the emergency lane to the normal lane, it must not only meet the basic conditions for changing lanes in the normal lane but also overcome the emergency lane exit resistance coefficient. Step D213, establish an emergency lane occupancy prohibition mechanism, which is used to ensure the unobstructedness of the emergency lane in case of emergency. Specifically, when the current vehicle density is at a certain level, prohibit vehicles from occupying the emergency lane; Step D214: Establish the specific lane-changing decision algorithm of the lane-changing decision model. The specific steps are as follows. Step D2141: Set the input and output of the lane-changing decision model. The input of the lane-changing decision model is the information of the lane where the vehicle is currently located, the driving direction, the destination information, the vehicle type, the current road density, and the real-time traffic condition information. The output of the lane-changing decision model is the decision result of whether to change lanes and the target lane. Step D2142: Initialize the decision result of whether to change lanes to False and the target lane to None, and then perform a lane-changing demand detection. If there is a lane-changing demand, evaluate the lane change to the target lane. The specific process of the lane-changing demand detection is as follows: First, calculate the remaining distance to the target exit. If the remaining distance is less than 500 meters and the current lane is not the lane where the target exit is located, a lane-changing demand is triggered. Then, obtain the traffic flow density and the speed of the vehicle in front in the current lane. If the traffic flow density is greater than 1.2 times the critical density and the speed of the vehicle in front is less than the threshold, a lane-changing demand is triggered. Subsequently, check whether there are obstacles ahead. If there are, a lane-changing demand is triggered. The specific process of evaluating the lane change to the target lane is as follows: First, screen the target lanes according to the driving direction and the destination. Then, for each target lane, obtain the distance to the vehicle in front, the speed of the vehicle in front, the acceleration of the vehicle in front, the distance to the vehicle behind, the speed of the vehicle behind, the acceleration of the vehicle behind, the vehicle type, and the current density in the target lane. Then, calculate the target lane evaluation index. The target lane evaluation index includes the expected acceleration after changing lanes in the target lane. Step D2143: Judge the lane-changing conditions for the target lane. If multiple target lanes meet the lane-changing conditions, select the target lane with the greatest lane-changing benefit. The specific process of judging the lane-changing conditions for the target lane is as follows: First, calculate the lane-changing benefit, which is the difference between the expected acceleration after lane-changing and the current acceleration. Then, calculate the yielding cost and consider the accelerations of the vehicles in front of and behind in the target lane. Next, if the target lane is the emergency lane, or if the vehicle type is a car and the current density is less than the critical density, then the target lane does not meet the lane-changing conditions. Otherwise, add an additional resistance value to the yielding cost. Subsequently, if the lane-changing benefit is greater than the yielding cost, then the target lane meets the lane-changing conditions. Finally, if multiple target lanes meet the lane-changing conditions simultaneously, select the target lane with the largest lane-changing benefit and update the decision result of whether to change lanes to and the ID of the selected target lane; Step D22: Establish an adaptive car-following model. The specific steps are as follows. Step D221: Dynamically calculate the car-following distance as shown in Equation (10). （10） Among them, is the minimum distance that the vehicle needs to maintain from the vehicle in front at time t, is the instantaneous speed of the vehicle at time t, is the time interval maintained between the vehicle and the vehicle in front, is the maximum deceleration, is the minimum safety distance; Step D222: Calculate the acceleration control as shown in Equation (11). （11） wherein, is the acceleration of the vehicle at time t, is the error, , and are all PID parameters.

8. A dynamic control method for vehicle flow on a two-way multi-lane expressway based on traffic simulation according to claim 5, characterized in that: Step E: Perform full-process control logic and closed-loop feedback on the control optimization result and establish an emergency scenario reinforcement control rule to complete the dynamic control operation of the vehicle flow on the two-way multi-lane expressway. The specific steps are as follows. Step E1, perform full-process control logic and closed-loop feedback on the control optimization result. The specific steps are as follows: Step E11, data collection and status evaluation, specifically, using a simulation sensor network to obtain real-time traffic flow, density, and vehicle type distribution data for all lanes, and then calculating the congestion index for each section and classifying them into unobstructed, mildly congested, moderately congested, and severely congested states, and the congestion index of the section as shown in formula (12). （12） Among them, if it is in an unobstructed state, if it is in a slightly congested state, if it is in a moderately congested state, if it is in a severely congested state; Step E12, traffic flow generation mode decision. If it is in a smooth or slightly congested state, enable the random generation mode and generate background traffic flow at the reference rate. If it is in a moderately or severely congested state, activate the directional generation mode to generate focused traffic flow and give priority to ensuring the traffic flow on the target route to merge, and at the same time open the emergency lane; Step E13, generate parameter dynamic configuration, specifically including randomly generated mode parameter configuration and directionally generated mode parameter configuration. The specific steps are as follows: Step E131, configure parameters for the randomly generated mode, specifically adjust the vehicle generation rate according to the real-time density and maintain the density balance of each lane; Step E132, perform parameter configuration for the directionally generated mode. Specifically, generate vehicles according to the priority ranking of the target route and ensure that the departure time sequence meets the safe following distance. Then, allocate the right of use of the emergency lane for large vehicles and preferentially select the upstream lane of the target lane for the initial lane, so as to avoid directly cutting into the inner high-flow lane; Step E14, execute microscopic behavior control. The microscopic behaviors include lane-changing decision-making and following control. The lane-changing decision-making is specifically to update the lane selection every 500 ms and preferentially meet the lane-changing requirements of the focused traffic flow. The following control is specifically to calculate the safe distance in real time and adjust the acceleration through a PID controller; Step E15, closed-loop feedback and parameter optimization. The specific steps are as follows: Step E151, statistically calculate the average vehicle speed every 10 seconds and the lane balance degree , where the lane balance degree is as shown in formula (13). （13） Among them, is the maximum value in the lane density, is the minimum value in the lane density, the average value of all lane densities; Step E152, if the average vehicle speed and the lane balance , then trigger the self-adjustment of PID parameters as shown in formula (14). （15）； Step E2, establish emergency scenario reinforcement control rules to complete the dynamic control operation of the vehicle flow on the two-way multi-lane expressway. The emergency scenario reinforcement control rules include an emergency lane hierarchical activation mechanism and a conflict resolution strategy.

9. A dynamic control method for vehicle flow on a two-way multi-lane highway based on traffic simulation according to claim 8, characterized in that: The specific steps of Step E2 are as follows: Step E21, establish an emergency lane hierarchical activation mechanism. Specifically, when in a moderately congested state, allow trucks or buses to use the emergency lane with a maximum speed limit of 90 km / h and prohibit lane-changing. When in a severely congested state, open the emergency lane to all vehicle types with a speed limit of 80 km / h, and set a forced lane-changing point every 2 km to guide vehicles back to the normal lane. The vehicles with higher priority in the focused traffic flow can use the emergency lane throughout the journey. The vehicles with higher priority include ambulances and buses; Step E22, construct a conflict resolution strategy. Specifically, when an emergency lane vehicle needs to return to the normal lane, turn on the turn signal at the response distance threshold in advance and send a lane-changing request to surrounding vehicles. If a normal lane vehicle detects the request and its own safety distance allows, it will actively decelerate and give way. If the lane change is not completed within the specified time, an emergency lane forced speed limit mechanism will be triggered.

10. A two-way multi-lane highway vehicle flow dynamic control system based on traffic simulation, the specific control process of the highway vehicle flow dynamic control system is based on the highway vehicle flow dynamic control method according to any one of claims 1-9, characterized in that: It includes a data acquisition module, a data filtering module, a data enhancement module, a control optimization module, and a dynamic control module. The data acquisition module is used to construct a virtual sensor network covering two-way multi-lanes using a traffic simulation platform to perform full-element perception of the vehicle flow and obtain acquisition data; The data filtering module is used to filter out outliers in the acquisition data using a combination of a double sliding window mechanism and an IQR algorithm to obtain filtered data; The data enhancement module is used to construct a multi-dimensional vector containing spatio-temporal features to enhance the feature vectors in the filtered data and obtain enhanced data; The control optimization module is used to establish a hierarchical generation control framework and refined control of microscopic driving behaviors based on the enhanced data to perform control optimization on the dynamic vehicle flow and obtain a control optimization result; The dynamic control module is used to perform full-process control logic and closed-loop feedback on the control optimization result and establish emergency scenario reinforcement control rules to complete the dynamic control operation of the vehicle flow on the two-way multi-lane expressway.

Images