A dynamic opening decision method for hard shoulder on highway based on MPC

Abstract

The present invention belongs to the field of intelligent transportation information technology and discloses a dynamic opening decision-making method for the hard shoulder of a highway based on MPC, comprising: S1. calibrating lane-level METANET basic model parameters using historical traffic flow data for the controlled section and its upstream and downstream sections, collecting real-time traffic data, and predicting traffic flow, density, and speed for the controlled section and its upstream and downstream sections based on the calibrated lane-level METANET basic model; S2. Analyzing the real-time road conditions based on the traffic flow, density, and speed prediction results from step S1, determining the condition of the hard shoulder and judging whether the hard shoulder needs to be opened; S3. Establishing a dynamic opening control system model for the hard shoulder and optimizing the opening timing and duration of the hard shoulder; and S4. Publishing dynamic hard shoulder control decision information. This invention considers efficiency and safety constraints, balances traffic efficiency and traffic safety benefits, can flexibly control the opening and closing of the hard shoulder, and has good adaptability and real-time performance.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation, and in particular relates to a dynamic opening decision method for hard shoulders of highways based on MPC. Background Art

[0002] The hard shoulder is the portion of the road (including the curb) adjacent to the roadway that is paved with a strong pavement structure. The hard shoulder is a lane marked by yellow lines on the right side of the highway. Its primary function is to temporarily park disabled vehicles and facilitate the passage of emergency vehicles in emergencies.

[0003] Dynamic hard shoulder opening decision-making technology for highways is an active traffic management and control technology. It dynamically adjusts the use of the hard shoulder based on current traffic conditions and future traffic forecasts. This increases the effective capacity of highways, alleviates congestion, and improves highway operating efficiency. Furthermore, this technology can timely open or close the hard shoulder in response to special circumstances such as accidents, construction, and extreme weather, enhancing highway emergency response capabilities, reducing accident risks, and improving highway safety.

[0004] Existing research primarily uses genetic algorithms to optimize dynamic hard shoulder opening control decisions. For example, Ye Zhen used genetic algorithms and sliding time windows to predict and optimize the opening and closing of hard shoulders under different constraints and objective functions. Tang Jinjun et al. also used genetic algorithms to minimize total travel time and total collision exposure time in dynamic hard shoulder opening decisions, with the dual constraints of open time and open space. However, genetic algorithms have high computational complexity and require significant computational resources and time. Compared to genetic algorithms, model predictive control (MPC) is more suitable for real-time, dynamic control problems and consumes less computational resources and time.

[0005] Patent document CN113034914A only considers traffic volume as a congestion evaluation indicator. When traffic volume is less than or equal to a preset threshold, the hard shoulder lane function does not need to be activated. When traffic volume exceeds the preset threshold and the hard shoulder is not occupied, the hard shoulder lane function can be activated. However, a single variable cannot accurately describe traffic flow status. This method fails to comprehensively consider road conditions and safety factors, making it difficult to make decisions based on the balance between traffic efficiency and traffic safety benefits.

[0006] Therefore, there is an urgent need for a dynamic hard shoulder opening control decision-making method that fully considers the traffic conditions and driving safety of highways, and is highly efficient and real-time, so as to assist traffic management personnel in managing highways and thereby improve highway capacity and service levels. Summary of the Invention

[0007] Dynamic hard shoulder opening, as a proactive traffic management tool, requires fixed-cycle opening and closing, which may not be suitable for current traffic conditions. Therefore, it is necessary to collect real-time traffic flow information and predict traffic conditions over a period of time to determine whether to open the hard shoulder as a driving lane. If so, the timing and duration of the hard shoulder opening must be further determined to ensure efficient traffic operation. Therefore, this paper proposes a dynamic hard shoulder opening decision-making method for highways based on MPC. This method aims to address the high computational complexity and significant consumption of computing resources and time associated with existing methods.

[0008] The present invention provides a method for dynamic opening decision of a highway hard shoulder based on MPC, comprising the following steps:

[0009] S1. Calibrate lane-level METANET basic model parameters using historical traffic flow data for the controlled road section and its upstream and downstream sections. Collect real-time traffic data and use the calibrated lane-level METANET basic model to predict traffic flow, density, and speed for the controlled road section and its upstream and downstream sections.

[0010] S2. Based on the traffic flow, density, and speed predictions from step S1, analyze the real-time road conditions to determine whether the hard shoulder is open, whether the hard shoulder is occupied, whether there are any special circumstances where opening the hard shoulder is inappropriate, whether the hard shoulder is in a state where opening is permitted, and whether opening the hard shoulder is necessary.

[0011] S3. Establish a hard shoulder dynamic opening control system model based on the lane-level METANET basic model calibrated in step S1, and use the hard shoulder dynamic opening control system model to optimize the opening timing and opening duration of the hard shoulder;

[0012] S4. Based on the hard shoulder opening timing and duration obtained in step S3, publish the hard shoulder dynamic control decision information.

[0013] Furthermore, step S2 includes the following sub-steps:

[0014] S2.1 Determine whether the hard shoulder is in a state where it can be opened;

[0015] I. Based on video and radar data, determine whether the hard shoulder of the hard shoulder controlled section is currently occupied. If so, disable the hard shoulder as a driving lane.

[0016] II. Based on video data, radar data, and event reports, determine whether there are any special events on the hard shoulder controlled section and its downstream that prohibit the use of the hard shoulder. If so, the hard shoulder will not be used as a driving lane.

[0017] III. Based on video data, radar data, and work schedule notifications, determine whether there are special security conditions on the hard shoulder controlled section and its downstream that prohibit the use of the hard shoulder. If so, the hard shoulder will not be used as a driving lane.

[0018] IV. Based on video data, radar data, and predicted traffic flow, determine whether there is a slowdown downstream of the hard shoulder controlled section at the current moment. If so, the hard shoulder is not allowed to be used as a driving lane.

[0019] V. If none of the above four conditions exist, the hard shoulder may be opened as a driving lane;

[0020] S2.2 If the hard shoulder is permitted to be used as a driving lane, obtain traffic flow information through highway sensing equipment and, combined with the traffic flow volume, density, and speed predicted in step S1, analyze whether the current road conditions require the hard shoulder to be used as a driving lane;

[0021] I. Ensure that the hard shoulder is in a state where it can be used as a driving lane; otherwise, do not open the hard shoulder;

[0022] II. Based on road infrastructure data or highway regulations, set a speed threshold (ASL) for opening the hard shoulder on the controlled section and a slowdown threshold (SRL) for the length of traffic upstream of the controlled section. When the speed on the controlled section falls below the ASL and the slowdown length upstream exceeds the SRL, the hard shoulder is opened as a driving lane.

[0023] III. Based on road infrastructure data or highway regulations, set a one-way traffic volume threshold (OTM) for opening the hard shoulder on the controlled section. When the one-way traffic volume on the controlled section exceeds the OTM, the hard shoulder is opened as a driving lane.

[0024] IV. Based on road infrastructure data or highway regulations, a traffic growth threshold (TCG) is set for opening the hard shoulder on the controlled section. If the traffic growth exceeds the TCG for two consecutive and equal time periods on the controlled section, the hard shoulder is opened as a driving lane.

[0025] V. When construction and maintenance work on the hard shoulder controlled section occupies the inner lane of the main line, causing a large number of vehicles to slow down, and the slow-down length is greater than the SRL, the hard shoulder will be opened as a driving lane.

[0026] Furthermore, step S3 includes the following sub-steps:

[0027] S3.1 Based on the lane-level METANET basic model calibrated in step S1, i.e., the lane-level METANET macroscopic traffic flow model, a hard shoulder dynamic open control system model is established;

[0028] The hard shoulder dynamic opening control system model includes the state equation and output equation of the hard shoulder dynamic opening control. The hard shoulder dynamic control system is expressed in state space form. The general form is:

[0029]

[0030] y=Cx+Du

[0031] Where x is the state vector of the system, which describes the state of the system at any point in time; u is the input vector of the system; y is the output vector of the system; A, B, C, and D are the parameters of the system, which are used to describe the dynamic characteristics and control mechanism of the system;

[0032] Combined with the lane-level METANET macro traffic flow model, the density K, speed V and flow Q of the controlled road section are used as system state variables, that is, x = [KVQ] T The traffic status of each lane, including the hard shoulder, is used as the system input variable, i.e., u = [λ1λ2λ3], where λ1 represents the traffic status of the left lane, λ2 represents the traffic status of the right lane, and λ3 represents the traffic status of the hard shoulder lane.

[0033] S3.2 Considering traffic efficiency and traffic safety, establish the objective function J, consider the proportion of vehicle types involved in traffic, and establish corresponding constraints;

[0034] Taking traffic efficiency and traffic safety as optimization goals, the total travel time is used to represent traffic efficiency. When the total travel time is the smallest, the road traffic efficiency is the highest. The collision exposure time is used to represent traffic safety.

[0035] The calculation expression of the total travel time TTT is as follows:

[0036] TTT z (t,j)=δ z ·Δt

[0037]

[0038] Where TTT z (t,j) represents the travel time of the zth vehicle in the tth cycle and the jth time step;

[0039] Indicates whether the i-th vehicle is within the controlled section; Δt represents the time step; N represents the number of participating vehicles; T represents the total execution time;

[0040] The calculation expression of the total collision exposure time TET is as follows:

[0041]

[0042]

[0043] Where TET(t,j) represents the collision exposure time of all vehicles in the tth cycle and the jth time step;

[0044] represents the collision risk of the zth vehicle at the jth time step;

[0045] TTC stands for Time to Collision, which refers to the time it takes for two vehicles to collide when they are traveling in the same direction on the same route and maintaining their respective speeds. * The threshold value representing the collision time;

[0046] The total travel time and the total collision exposure time are combined in a weighted manner as the overall optimization objective J, which is expressed as follows:

[0047]

[0048] Where, α ttt represents the weight of traffic efficiency index; β tet Indicates the weight of traffic safety index; TTT i (t,j) represents the total travel time of segment i in the tth cycle and the jth time step; TET i (t,j): total collision exposure time of road segment i in the tth cycle and the jth time step;

[0049] S3.3 Use the hard shoulder dynamic open control system model to predict the state and behavior of the hard shoulder dynamic open control system over a period of time in the future;

[0050] I. Determine the time window and control time interval for predicting system behavior;

[0051] The sampling period of MPC is T, the order is k; the control period is T c =MT, which is an integer multiple of the sampling period, in order of k c ; Prediction time domain N p , control time domain N c ;but:

[0052] μ(k),u(k)), where k c T c / T≤k≤(k c +1)T c / T

[0053] Where, μ(k) and u(k) are the basic parameters, interference parameters, and control parameters of the kth sampling period, respectively, which can be used to iterate the traffic flow parameters at time j+1;

[0054] II. At the new sampling time, the actual output of the controlled object is detected and compared with the predicted output based on the model. The error between the two is calculated and the error information is used to correct the model-based prediction result.

[0055] III. Based on the established objective function J, considering the proportion of vehicle types participating in the traffic, establishing corresponding constraints, and using genetic algorithms to minimize the objective function J, find the optimal control signal parameter u(k c ); the optimal control signal parameter u(k c ) as input variables; repeat the above steps until the set prediction step N is reached p , the predicted time period is [kT,(k+N p -1)T], to achieve rolling optimization.

[0056] Furthermore, the optimization algorithm is one of linear quadratic programming, nonlinear optimization and genetic algorithm.

[0057] Beneficial effects:

[0058] Based on the basic conditions of highways, a suitable traffic flow model is established to characterize the evolution of traffic volume and speed. Based on this, traffic conditions over a period of time are predicted, which helps to more accurately predict and control traffic flow. In the dynamic hard shoulder opening decision-making process, MPC's ability to handle high-dimensional, multivariable control problems and its ability to quickly adapt and adjust in dynamic environments is utilized. This paper proposes an MPC-based dynamic hard shoulder opening decision-making method for highways. This method considers both efficiency and safety constraints, balancing traffic efficiency and safety benefits. It can flexibly control the opening and closing of the hard shoulder and has good adaptability and real-time performance.

[0059] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a flow chart of a method for dynamic opening decision of hard shoulder of highway based on MPC of the present invention;

[0061] Figure 2 Schematic diagram of the lane-level METANET model;

[0062] Figure 3 This is the operating structure of the hard shoulder MPC control. DETAILED DESCRIPTION

[0063] To make the technical solutions, advantages, and purposes of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0064] like Figure 1 As shown, the present invention provides a method for dynamic opening decision of hard shoulder of highway based on MPC, comprising the following steps:

[0065] S1. Calibrate lane-level METANET basic model parameters using historical traffic flow data for the controlled road section and its upstream and downstream sections. Collect real-time traffic data and use the calibrated lane-level METANET basic model to predict traffic flow, density, and speed for the controlled road section and its upstream and downstream sections.

[0066] The METANET model uses Taylor's formula and difference equations to discretize traffic flow parameters in time and space, establishing a second-order relationship model between basic traffic flow parameters. Based on the classic METANET model, the model is improved to a lane-level METANET basic model, taking into account the differences in traffic flow parameters between lanes. This model is suitable for the study of dynamic opening control of hard shoulders. The model is as follows:

[0067] like Figure 2 As shown, the time is divided into n time steps of length Δt, and the sampling time is j = 1, 2, 3, ..., n. The road section is divided into m small sections of length Δx, then i = 1, 2, ..., m. The number of lanes in the controlled section is λ, then l = 1, 2, ..., λ. After division, the controlled section is divided into λ·m units, each unit can be expressed as (i, l). In order to ensure numerical stability, it is necessary to satisfy Δt·v f <Δx,v f is the free flow velocity. Taking the (i,l) unit as the research object, the following equation is satisfied:

[0068] ① Basic equation of three parameters of traffic flow:

[0069]

[0070] in: represents the flow from unit (i, l) into (i+1, l) at time j; represents the density of unit (i, l) at time j; Represents the velocity of unit (i, l) at time j.

[0071] ② Traffic flow conservation equation:

[0072]

[0073] in: The flow from the hard shoulder to the unit (i, l) at time j; The flow rate of unit (i, l) merging into the hard shoulder at time j.

[0074] ③Dynamic velocity description equation:

[0075] This equation indicates that it takes a certain amount of time for the vehicle speed to adjust to the steady-state speed expected by the driver, and discretizes the speed.

[0076] Steady-state velocity, that is, in the velocity-density basic diagram, when the density is When , the corresponding speed value;

[0077] τ: driver adjustment delay coefficient;

[0078] η: velocity-density relationship coefficient;

[0079] κ: elastic modulus, used to avoid Error caused by being too small;

[0080] τ, η, and κ are all determined by the characteristics of the road section, and their values ​​are calibrated by studying the characteristics of road traffic flow.

[0081] The relaxation term reflects the driver's tendency to drive at a steady-state speed;

[0082] The convection term represents the speed change caused by the vehicle in the upstream unit entering the current unit;

[0083] The expectation term adjusts the speed of the current unit at the next moment by feeding back the changes in the density of downstream units.

[0084] ④ Velocity-density steady-state relationship equation:

[0085]

[0086] This equation describes the traffic flow characteristics of the unit (i, l) in a statistical sense. f : free stream velocity; kc: critical density; α: velocity-density basic diagram parameter.

[0087] S2. Based on the traffic flow, density, and speed predictions from step S1, and taking into account the real-time road conditions, determine whether the hard shoulder should be opened. First, based on real-time road monitoring data, determine whether the hard shoulder is occupied and whether there are any special circumstances that would preclude opening the hard shoulder. Then, determine whether the hard shoulder is in a state where it can be opened. Second, based on the traffic flow predictions, consider the evolving trends in traffic flow, density, and speed to determine whether the hard shoulder should be opened.

[0088] S2.1 When determining whether the hard shoulder is open, first determine whether the hard shoulder is in a state where it can be opened;

[0089] I. Based on video and radar data, determine whether the hard shoulder of the hard shoulder controlled section is currently occupied. If so, disable the hard shoulder as a driving lane.

[0090] II. Based on video data, radar data, and event reports, determine whether there are any special events on the hard shoulder controlled section and its downstream that prohibit the use of the hard shoulder. If so, the hard shoulder will not be used as a driving lane.

[0091] III. Based on video data, radar data, and work schedule notifications, determine whether there are special security conditions on the hard shoulder controlled section and its downstream that prohibit the use of the hard shoulder. If so, the hard shoulder will not be used as a driving lane.

[0092] IV. Based on video data, radar data, and predicted traffic flow, determine whether there is a slowdown downstream of the hard shoulder controlled section at the current moment. If so, the hard shoulder is not allowed to be used as a driving lane.

[0093] V. If none of the above four conditions exist, the hard shoulder may be opened as a driving lane;

[0094] S2.2 Where the hard shoulder is permitted to be used as a driving lane, traffic flow information is obtained through highway sensing equipment such as video, radar, and ETC gantries. Combined with traffic flow predictions, speed, and density data, a comprehensive consideration is given to whether the current road conditions warrant the use of the hard shoulder as a driving lane.

[0095] I. Ensure that the hard shoulder is in a state where it can be used as a driving lane; otherwise, do not open the hard shoulder;

[0096] II. Based on road infrastructure data or highway regulations, set a speed threshold (ASL) for opening the hard shoulder on the controlled section and a slowdown threshold (SRL) for the length of traffic upstream of the controlled section. When the speed on the controlled section falls below the ASL and the slowdown length upstream exceeds the SRL, the hard shoulder is opened as a driving lane.

[0097] III. Based on road infrastructure data or highway regulations, set a one-way traffic volume threshold (OTM) for opening the hard shoulder on the controlled section. When the one-way traffic volume on the controlled section exceeds the OTM, the hard shoulder is opened as a driving lane.

[0098] IV. Based on road infrastructure data or highway regulations, a traffic growth threshold (TCG) is set for opening the hard shoulder on the controlled section. If the traffic growth exceeds the TCG for two consecutive and equal time periods on the controlled section, the hard shoulder is opened as a driving lane.

[0099] V. When construction and maintenance work on the hard shoulder controlled section occupies the inner lane of the main line, causing a large number of vehicles to slow down, and the slow-down length is greater than the SRL, the hard shoulder will be opened as a driving lane.

[0100] S3. Establish a hard shoulder dynamic opening control system model based on the lane-level METANET basic model calibrated in step S1, and use the hard shoulder dynamic opening control system model to optimize the opening timing and opening duration of the hard shoulder;

[0101] When the hard shoulder needs to be opened, the model predictive control method is used to optimize the opening timing and opening duration of the hard shoulder. Based on the lane-level METANET macro traffic flow model calibrated in step S1, a dynamic opening control system model for the hard shoulder is established. The model contains information such as the opening status of the hard shoulder and the system output. The objective function J(x) is established considering traffic efficiency and traffic safety, and the corresponding constraints are established considering information such as the proportion of vehicle types participating in the traffic. The system model is used to predict the state and behavior of the dynamic opening control system of the hard shoulder in the future. Based on the predicted system behavior, a suitable optimization algorithm, such as a genetic algorithm, is used to calculate the control input sequence that can optimally achieve the control target in the future. The optimization problem is solved by minimizing the objective function, and the optimal opening timing and opening duration of the hard shoulder are determined based on the obtained traffic state function. In this process, it is ensured that the calculated control input sequence meets the system constraints.

[0102] S3.1 Based on the lane-level METANET basic model calibrated in step S1, i.e., the lane-level METANET macroscopic traffic flow model, a hard shoulder dynamic open control system model is established;

[0103] The hard shoulder dynamic opening control system model includes the state equation and output equation of the hard shoulder dynamic opening control. The hard shoulder dynamic control system is expressed in state space form. The general form is:

[0104]

[0105] y=Cx+Du

[0106] Where x is the state vector of the system, which describes the state of the system at any point in time; u is the input vector of the system; y is the output vector of the system; A, B, C, and D are the parameters of the system, which are used to describe the dynamic characteristics and control mechanism of the system;

[0107] Combined with the lane-level METANET macro traffic flow model, the density K, speed V and flow Q of the controlled road section are used as system state variables, that is, x = [KVQ] T The traffic status of each lane, including the hard shoulder, is used as the system input variable, i.e., u = [λ1λ2λ3], where λ1 represents the traffic status of the left lane, λ2 represents the traffic status of the right lane, and λ3 represents the traffic status of the hard shoulder lane.

[0108] S3.2 Considering traffic efficiency and traffic safety, establish the objective function J, consider the proportion of vehicle types involved in traffic, and establish corresponding constraints;

[0109] Taking traffic efficiency and traffic safety as optimization goals, the total travel time is used to represent traffic efficiency, which describes the travel time of all vehicles participating in this traffic section through this road section. When the total travel time is minimized, the road traffic efficiency is the highest. The total collision exposure time is used to represent traffic safety, which describes the total time that all vehicles participating in this traffic section are in dangerous situations while passing through this road section, reflecting the safety status of the vehicles.

[0110] The calculation expression of the total travel time TTT is as follows:

[0111] TTT z (t,j)=δ z ·Δt

[0112]

[0113] Where TTT z (t,j) represents the travel time of the zth vehicle in the tth cycle and the jth time step;

[0114] Indicates whether the i-th vehicle is within the controlled section; Δt represents the time step; N represents the number of participating vehicles; T represents the total execution time;

[0115] The calculation expression of the total collision exposure time TET is as follows:

[0116]

[0117]

[0118] Where TET(t,j) represents the collision exposure time of all vehicles in the tth cycle and the jth time step;

[0119] represents the collision risk of the zth vehicle at the jth time step;

[0120] TTC stands for Time to Collision, which refers to the time it takes for two vehicles to collide when they are traveling in the same direction on the same route and maintaining their respective speeds. * represents the threshold value of collision time;

[0121] The total travel time and the total collision exposure time are combined in a weighted manner as the overall optimization objective J, which is expressed as follows:

[0122]

[0123] Where, α ttt represents the weight of traffic efficiency index; β tet Indicates the weight of traffic safety index; TTT i (t,j) represents the total travel time of segment i in the tth cycle and the jth time step; TET i (t,j): total collision exposure time of road segment i in the tth cycle and the jth time step;

[0124] S3.3 Determine the time window and control time interval for predicting system behavior, and predict the state and behavior of the hard shoulder dynamic open control system in the future based on the system model. Model Predictive Control (MPC) has three steps: multi-step prediction, feedback correction and rolling optimization. The operating structure of the hard shoulder MPC control is as follows: Figure 3 shown.

[0125] I. Determine the time window and control time interval for predicting system behavior;

[0126] Based on the lane-level METANET model, the density K, speed V, and flow Q of the controlled road section are system state variables, and the traffic status of each lane including the hard shoulder is the system input variable. The MPC sampling period is T, the order is k; the control period is T c =MT, which is an integer multiple of the sampling period, in order of k c ; Prediction time domain N p , control time domain N c .but:

[0127] μ(k),u(k)), where k c T c / T≤k≤(k c +1)T c / T

[0128] Where, μ(k) and u(k) are the basic parameters, interference parameters, and control parameters of the kth sampling period, respectively. This allows us to iterate the traffic flow parameters at time k+1 and achieve prediction.

[0129] II. At the new sampling time, the actual output of the controlled object is detected and compared with the model-based predicted output. The error between the two is calculated and the model-based predicted output is corrected using this error information to more accurately reflect the actual state of the system.

[0130] III. Based on the established objective function J, considering the proportion of vehicle types participating in the traffic, establishing corresponding constraints, and using genetic algorithms to minimize the objective function J, find the optimal control signal parameter u(k c ); the optimal control signal parameter u(k c ) as input variables; repeat the above steps until the set prediction step N is reached p , the predicted time period is [kT,(k+N p -1)T], to achieve rolling optimization.

[0131] S4. Based on the hard shoulder opening timing and duration obtained in step S3, information on dynamic hard shoulder control decisions is disseminated through variable message boards, broadcasts, and other channels. Variable message boards are installed at the starting, midpoint, and end points of the controlled road section, displaying the message "Small passenger vehicles in the XX direction may use the emergency lane" to inform drivers of the dynamic hard shoulder control decision. Broadcasts can include information such as the hard shoulder opening duration in addition to the information displayed on the variable message boards.

[0132] It is hereby stated that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions, which should all be included in the scope of the claims of the present invention.

Claims

1. A dynamic opening decision method for hard shoulder of highway based on MPC, characterized by: The following steps are involved: S1. Calibrate lane-level METANET basic model parameters using historical traffic flow data for the controlled road section and its upstream and downstream sections. Collect real-time traffic data and use the calibrated lane-level METANET basic model to predict traffic flow, density, and speed for the controlled road section and its upstream and downstream sections. S2. Based on the traffic flow, density, and speed predictions from step S1, analyze the real-time road conditions to determine whether the hard shoulder is open, whether the hard shoulder is occupied, whether there are any special circumstances where opening the hard shoulder is inappropriate, whether the hard shoulder is in a state where opening is permitted, and whether opening the hard shoulder is necessary. S3. Establish a hard shoulder dynamic opening control system model based on the lane-level METANET basic model calibrated in step S1, and use the hard shoulder dynamic opening control system model to optimize the opening timing and opening duration of the hard shoulder; The step S3 includes the following sub-steps: S3.1 Based on the lane-level METANET basic model calibrated in step S1, i.e., the lane-level METANET macroscopic traffic flow model, a hard shoulder dynamic open control system model is established; The hard shoulder dynamic opening control system model includes the state equation and output equation of the hard shoulder dynamic opening control. The hard shoulder dynamic control system is expressed in state space form. The general form is: y=Cx+Du Where x is the state vector of the system, which describes the state of the system at any point in time; u is the input vector of the system; y is the output vector of the system; A, B, C, and D are the parameters of the system, which are used to describe the dynamic characteristics and control mechanism of the system; Combined with the lane-level METANET macro traffic flow model, the density K, speed V and flow Q of the controlled road section are used as system state variables, that is, x = [KVQ] T The traffic status of each lane, including the hard shoulder, is used as the system input variable, i.e., u = [λ1λ2λ3], where λ1 represents the traffic status of the left lane, λ2 represents the traffic status of the right lane, and λ3 represents the traffic status of the hard shoulder lane. S3.2 Considering traffic efficiency and traffic safety, establish the objective function J, consider the proportion of vehicle types involved in traffic, and establish corresponding constraints; Taking traffic efficiency and traffic safety as optimization goals, the total travel time is used to represent traffic efficiency. When the total travel time is the smallest, the road traffic efficiency is the highest. The collision exposure time is used to represent traffic safety. The calculation expression of the total travel time TTT is as follows: Where TTT z (t,j) represents the travel time of the zth vehicle in the tth cycle and the jth time step; Indicates whether the i-th vehicle is within the controlled section; Δt represents the time step; N represents the number of participating vehicles; T represents the total execution time; The calculation expression of the total collision exposure time TET is as follows: Where TET(t,j) represents the collision exposure time of all vehicles in the tth cycle and the jth time step; represents the collision risk of the zth vehicle at the jth time step; TTC z (j) represents the time to collision, which refers to the time it takes for two vehicles to collide when they are traveling in the same direction on the same route and maintaining their respective speeds. * represents the threshold value of collision time; The total travel time and the total collision exposure time are combined in a weighted manner as the overall optimization objective J, which is expressed as follows: Where, α ttt represents the weight of traffic efficiency index; β tet Indicates the weight of traffic safety index; TTT i (t,j) represents the total travel time of segment i in the tth cycle and the jth time step; TET i (t,j): total collision exposure time of road segment t in the tth cycle and the jth time step; S3.3 Use the hard shoulder dynamic open control system model to predict the state and behavior of the hard shoulder dynamic open control system over a period of time in the future; I. Determine the time window and control time interval for predicting system behavior; The sampling period of MPC is T, the order is k; the control period is T c =MT, which is an integer multiple of the sampling period, in order of k c ; Prediction time domain N p , control time domain N c ;but: Among them, k c T c / T≤k≤(k c +1)T c / T Where, are the basic parameters, interference parameters and control parameters of the kth sampling period, respectively, so that the traffic flow parameters at time k+1 can be iterated; II. At the new sampling time, the actual output of the controlled object is detected and compared with the predicted output based on the model. The error between the two is calculated and the error information is used to correct the model-based prediction result. III. Based on the established objective function J, considering the proportion of vehicle types participating in the traffic, establishing corresponding constraints, and using genetic algorithms to minimize the objective function J, find the optimal control signal parameter u(k c ); the optimal control signal parameter u(k c ) as input variables; repeat the above steps until the set prediction step N is reached p , the predicted time period is [kT,(k+N p -1)T], to achieve rolling optimization; S4. Based on the hard shoulder opening timing and opening duration obtained in step S3, publish the hard shoulder dynamic control decision information.

2. The MPC-based dynamic highway hard shoulder opening decision method according to claim 1 is characterized by: The step S2 includes the following sub-steps: S2.1 Determine whether the hard shoulder is in a state where it can be opened; I. Based on video and radar data, determine whether the hard shoulder of the hard shoulder controlled section is currently occupied. If so, disable the hard shoulder as a driving lane. II. Based on video data, radar data, and event reports, determine whether there are any special events on the hard shoulder controlled section and its downstream that prohibit the use of the hard shoulder. If so, the hard shoulder will not be used as a driving lane. III. Based on video data, radar data, and work schedule notifications, determine whether there are special security conditions on the hard shoulder controlled section and its downstream that prohibit the use of the hard shoulder. If so, the hard shoulder will not be used as a driving lane. IV. Based on video data, radar data, and predicted traffic flow, determine whether there is a slowdown downstream of the hard shoulder controlled section at the current moment. If so, the hard shoulder is not allowed to be used as a driving lane. V. If none of the above four conditions exist, the hard shoulder may be opened as a driving lane; S2.2 If the hard shoulder is permitted to be used as a driving lane, obtain traffic flow information through highway sensing equipment and, combined with the traffic flow volume, density, and speed predicted in step S1, analyze whether the current road conditions require the hard shoulder to be used as a driving lane; I. Ensure that the hard shoulder is in a state where it can be used as a driving lane; otherwise, do not open the hard shoulder; II. Based on road infrastructure data or highway regulations, set a speed threshold (ASL) for opening the hard shoulder on the controlled section and a slowdown threshold (SRL) for the length of traffic upstream of the controlled section. When the speed on the controlled section falls below the ASL and the slowdown length upstream exceeds the SRL, the hard shoulder is opened as a driving lane. III. Based on road infrastructure data or highway regulations, set a one-way traffic volume threshold (OTM) for opening the hard shoulder on the controlled section. When the one-way traffic volume on the controlled section exceeds the OTM, the hard shoulder is opened as a driving lane. IV. Based on road infrastructure data or highway regulations, a traffic growth threshold (TCG) is set for opening the hard shoulder on the controlled section. If the traffic growth exceeds the TCG for two consecutive and equal time periods on the controlled section, the hard shoulder is opened as a driving lane. V. When construction and maintenance work on the hard shoulder controlled section occupies the inner lane of the main line, causing a large number of vehicles to slow down, and the slow-down length is greater than the SRL, the hard shoulder will be opened as a driving lane.

Images