A road queue distribution estimation method based on multi-section electric police data

Abstract

The present invention relates to a road queue distribution estimation method based on multi-section electric police data, comprising the following steps: step 1) obtaining the timestamp information of vehicles passing through the stop line in each direction of the upstream and downstream intersections through electric police equipment; step 2) setting the virtual section of the upstream intersection, reconstructing the cumulative vehicle arrival curve of the virtual section of the upstream intersection based on the data captured by the electric police equipment, and calculating the cumulative vehicle departure curve of the downstream intersection; step 3) considering the distribution of vehicle free flow travel time, respectively for the assumption of fixed travel time scenario and variable travel time scenario, calculating the vehicle virtual departure curve of the downstream intersection, and restoring the queue distribution of the road section by analyzing the relationship between the cumulative vehicle arrival curve of the virtual section of the upstream intersection, the cumulative vehicle departure curve of the downstream intersection, and the vehicle virtual departure curve of the downstream intersection. Compared with the prior art, the present invention has the advantages of more accurate estimation and the like.

Description

Technical Field

[0001] The invention relates to the field of traffic management, and in particular to a road queue distribution estimation method based on multi-section electric police data. Background Art

[0002] Signal-controlled intersections are important nodes in urban road networks and are also frequent locations of traffic congestion. The queue length can vividly and intuitively reflect the degree of congestion at signal-controlled intersections. It is one of the important indicators for evaluating the operating efficiency of intersections and an important parameter for optimizing intersection signal control. As one of the common indicators for evaluating signal control operation, the maximum queue length index of a cycle can quantitatively describe the highest level of queued vehicles within a cycle and has attracted widespread attention from scholars at home and abroad. However, this indicator is difficult to reflect the real-time changes in queue length within a cycle. Accurately and effectively estimating the distribution of intersection queue length within a cycle can not only evaluate the current queue status of each flow direction at the intersection and the rationality of the signal control scheme, but also formulate corresponding real-time signal control strategies based on the current queue distribution and optimize the intersection signal timing parameters.

[0003] At present, the main data source for research on queuing distribution estimation at home and abroad is mobile connected vehicle trajectory data. Based on the position, speed and other information provided by the mobile connected vehicle trajectory data, combined with shock wave theory, vehicle kinematics and other models, the queuing and dissipation process of vehicles can be restored; on this basis, the key start and stop points of the vehicle can be identified and different arrival modes can be distinguished, so as to obtain the queue distribution of the studied period. However, the mobile connected vehicle trajectory data is limited by its data upload frequency and sampling rate, and the detection environment is not common in reality. When no valid mobile connected vehicle trajectory is collected within the period, it is difficult to estimate the queue distribution.

[0004] In recent years, due to the needs of safe city construction and traffic law enforcement management, the electric police system at the checkpoint has been widely installed and used in various cities in my country. Compared with the traditional fixed-point detector data, the electric police equipment can accurately record the status information (license plate, timestamp, vehicle type, lane number, etc.) and violation information of all vehicles passing through the detection area in each lane; in addition, by matching the license plate of the electric police data upstream and downstream of the urban trunk road, the operation status of the vehicle on the urban trunk road can be obtained, further restoring the current traffic situation. However, there is currently no method for estimating the road queue distribution based on the information provided by the electric police equipment. Summary of the invention

[0005] The purpose of the present invention is to provide a road queue distribution estimation method based on multi-section electric police data, which does not need to be based on the "first come, first served" assumption in the traditional input-output model, allows vehicles to overtake, and can obtain more realistic queue estimation results.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A road queue distribution estimation method based on multi-section electric police data comprises the following steps:

[0008] Step 1) obtaining the time stamp information of vehicles passing through the stop line in each direction of the upstream and downstream intersections through the electric police equipment;

[0009] Step 2) setting a virtual cross section of the upstream intersection, reconstructing a cumulative vehicle arrival curve of the virtual cross section of the upstream intersection based on data captured by the electric police equipment, and calculating a cumulative vehicle departure curve of the downstream intersection;

[0010] Step 3) Considering the distribution of vehicle free flow travel time, the virtual vehicle departure curve of the downstream intersection is calculated for the assumptions of fixed travel time scenario and variable travel time scenario respectively. By analyzing the relationship between the cumulative vehicle arrival curve of the virtual section of the upstream intersection, the cumulative vehicle departure curve of the downstream intersection, and the virtual vehicle departure curve of the downstream intersection, the queue distribution of the road section is restored.

[0011] In the upstream and downstream intersection scenario, there are three sources of departing vehicles in any direction downstream: turning left from the upstream intersection, going straight from the upstream intersection, and turning right from the upstream intersection. The directions of departing vehicles downstream include: turning left downstream, going straight downstream, and turning right downstream.

[0012] The electric alarm equipment is arranged at the entrance of each entrance lane of the intersection, and is used to collect the timestamp information of each flow of vehicles passing through the stop line. The information collected by the electric alarm equipment is lane-level information.

[0013] The step 2) comprises the following steps:

[0014] Step 2-1) Setting the upstream intersection virtual section as the upstream arrival section;

[0015] Step 2-2) Based on the data captured by the upstream electric police equipment and the average time required for vehicles of different directions to pass through the intersection, the timestamp of each vehicle passing through the virtual section of the upstream intersection is calculated;

[0016] Step 2-3) based on the timestamp of each vehicle passing through the virtual section of the upstream intersection, solve the cumulative number of upstream arriving vehicles to obtain the cumulative vehicle arrival curve of the virtual section of the upstream intersection;

[0017] Step 2-4) Based on the data captured by the downstream electric police equipment, the cumulative number of vehicles leaving the downstream is calculated to obtain the cumulative vehicle departure curve of the downstream intersection.

[0018] The calculation method for the timestamp of each vehicle passing through the virtual section of the upstream intersection is:

[0019]

[0020] in, It indicates the timestamp when the i-th vehicle at the upstream intersection passes through the upstream virtual section; Indicates the timestamp when the i-th vehicle at the upstream intersection is captured by the electric police equipment in the three actual sections of the upstream intersection; Δt l ,Δt th ,Δt r Respectively represent the average time required for left-turning vehicles, straight-going vehicles, and right-turning vehicles in the upstream intersection to pass through the intersection; N l ,N th ,N r They represent the collection of left-turning vehicles, straight-going vehicles, and right-turning vehicles at the upstream intersection respectively.

[0021] The steps 2-3) include the following steps:

[0022] Step 2-3-1) Based on the timestamp of each vehicle passing through the virtual section of the upstream intersection, according to the flow direction of the downstream departing vehicles, distinguish between downstream left turn, downstream straight, and downstream right turn, and obtain the cumulative number of vehicles arriving in the three directions of the upstream intersection at any time for different downstream flow directions:

[0023]

[0024]

[0025]

[0026] Among them, A x (t),A y (t),A z (t) represents the number of vehicles arriving at time t for upstream left turn, upstream straight and upstream right turn of a certain downstream flow direction; is the length of the upstream analysis period; Δt is the analysis time granularity; N x ,N y ,N z Respectively represent the number of vehicles captured during the analysis period for upstream left turn, upstream straight and upstream right turn of a certain downstream flow direction;

[0027] Step 2-3-2) By adding up the cumulative number of vehicles arriving in the three directions, the cumulative number of vehicles arriving at the virtual cross section of the upstream intersection at any time in a certain downstream direction is obtained:

[0028] A(t)=A x (t)+A y (t)+A z (t)

[0029] Step 2-3-3) Match the cumulative number of arriving vehicles A(t) at the timestamp to obtain the cumulative vehicle arrival curve of the virtual section of the upstream intersection.

[0030] The method for calculating the cumulative number of downstream leaving vehicles based on the data captured by the downstream electric alarm equipment to obtain the cumulative vehicle leaving curve of the downstream intersection is:

[0031]

[0032] Where N represents the number of vehicles captured in the downstream flow direction within the length of the analysis period; is the length of the downstream analysis period; Δt is the analysis time granularity, which is consistent with the value of the upstream arrival curve process parameter; D(·) represents the function of the number of downstream leaving vehicles over time, and the time is offset compared with the upstream arriving vehicles, TT f It represents the free flow travel time of vehicles in the fixed travel time scenario, and d1 represents the signal control delay of the first vehicle in the team.

[0033] The step 3) is specifically as follows: for the assumptions of fixed travel time scenario and variable travel time scenario, a virtual vehicle departure curve of the downstream intersection is constructed to represent the departure timestamp of the upstream arriving vehicles theoretically passing through the downstream stop line when there is no traffic light at the downstream intersection; the difference between the cumulative vehicle arrival curve of the virtual section of the upstream intersection and the virtual vehicle departure curve of the downstream vehicles is the number of vehicles traveling on the road section; the difference between the virtual vehicle departure curve of the downstream vehicles and the cumulative vehicle departure curve of the downstream intersection is the number of vehicles currently queuing, based on which the queue distribution in the studied period is estimated.

[0034] For the fixed travel time scenario, that is, the scenario where the free travel time of vehicles in the road section is a fixed value, restoring the queue distribution of the road section specifically includes the following steps:

[0035] Step 3-1-1) Determine the virtual departure curve of the vehicle at the downstream intersection under the fixed travel time scenario, which is the horizontal offset of the cumulative arrival curve of the upstream vehicles:

[0036]

[0037] Where P(t) represents the number of vehicles that virtually pass the downstream stop line of the road section at time t, that is, the number of vehicles that can theoretically pass the downstream intersection without a traffic light; TT f represents the free flow travel time of vehicles under the assumed fixed travel time scenario; is the length of the upstream analysis period; N represents the cumulative number of vehicles arriving upstream;

[0038] Step 3-1-2) Based on the virtual departure curve of the vehicle at the downstream intersection, calculate the queue length value q(t) at the downstream intersection at any time t, that is, the queue length distribution:

[0039] q(t)=P(t)-D(t)

[0040] Where q(t) represents the number of queued vehicles at the downstream intersection at time t; P(t) and D(t) represent the number of virtual vehicles leaving the downstream intersection and the number of actual vehicles leaving the downstream intersection at time t, respectively;

[0041] Step 3-1-3) Based on the obtained queue distribution within the period, determine the maximum queue length within the period:

[0042]

[0043] TT′ n+1,1 =P -1 (q′)

[0044] Where q(t′) represents the maximum queue length of the analyzed period; t′ represents the time when the maximum queue length is obtained; r n+1 Indicates the time when the red light turns on in the next cycle; TT′ n+1,1 Indicates the time point where the virtual departure curve of the downstream intersection intersects the actual departure curve. When TT′ n+1,1 <r n+1 When TT′ is 0, it means that before the end of the current cycle, the virtual departure curve of the downstream intersection has intersected with the actual departure curve, and the supply and demand are balanced or the supply is greater than the demand until the end of the cycle, that is, the cycle is not saturated. At this time, the maximum queue length in the cycle is the number of queued vehicles corresponding to this intersection. When TT′ n+1,1 ≥r n+1 When P is greater than 0, it means that the intersection of the virtual departure curve and the actual departure curve of the downstream intersection is located after the cycle, that is, in the current cycle, the road section has been in a state where demand is greater than supply, and oversaturation has occurred; P -1 (·) represents the inverse function of the virtual departure curve P(·), that is, the time corresponding to the virtual departure curve when the number of vehicles is a certain value; q′ is the number of vehicles corresponding to the intersection of the virtual departure curve and the actual departure curve at the downstream intersection.

[0045] For the variable travel time scenario, assuming that the free flow travel time of the moving vehicles satisfies the log-normal distribution, the following steps are specifically included to restore the queue distribution of the road section:

[0046] Step 3-2-1) Based on the cumulative arrival vehicle curve A(t) of the virtual section of the upstream intersection and the actual departure vehicle curve D(t) of the downstream intersection, it is assumed that the free flow travel time of the vehicle satisfies the log-normal distribution, that is, Get the probability density function:

[0047]

[0048] Among them, TT i represents the vehicle free flow travel time in the variable travel time scenario, represents the mean of the lognormal distribution, σ 2 represents the variance of the lognormal distribution;

[0049] Step 3-2-2) In the variable travel time scenario, for any vehicle i passing through the upstream virtual section, the probability value of whether the vehicle can reach the downstream intersection at any given time t is:

[0050]

[0051] Among them, P t,i It represents the probability value of whether any vehicle i passing through the upstream virtual section can reach the downstream intersection at any given time t; It refers to the timestamp of any vehicle i passing through the upstream virtual section;

[0052] Step 3-2-3) Calculate the probability value that the number of virtual vehicles leaving the downstream intersection is n at any time t:

[0053]

[0054] Among them, P t,n represents the probability value that the number of virtual vehicles leaving the downstream intersection is n at any time t; C represents the sample space, that is, the different divisions of queued vehicles and non-queued vehicles in the current cycle; A represents queued vehicles, and B represents non-queued vehicles;

[0055] Step 3-2-4) Consider the vehicle overtaking phenomenon, calculate the probability of various combination scenarios, and obtain the effective sample space number:

[0056]

[0057] Where N(C) represents the number of valid sample spaces, that is, the number of possible combinations of vehicles in a cycle when the number of virtual vehicles leaving the downstream intersection is n; δ represents the overtaking range of each vehicle in the current scenario; It represents the total number of vehicles passing through the upstream virtual section at any time t;

[0058] Step 3-2-5) Calculate the mean of the number of virtual vehicles leaving the downstream intersection at any time t, and on this basis, taking into account the volatility of vehicle travel time, calculate the upper and lower bounds of the number of virtual vehicles leaving the downstream intersection:

[0059]

[0060]

[0061]

[0062] in, They represent the mean, upper bound and lower bound of the number of virtual vehicles leaving the downstream intersection at any time t respectively;

[0063] Step 3-2-6) Calculate the mean, upper and lower bounds of the queue length at any time t based on the mean, upper and lower bounds of the number of virtual vehicles leaving the downstream intersection at any time t:

[0064]

[0065]

[0066]

[0067] in, They refer to the average value, upper bound and lower bound of the queue length at any time t respectively;

[0068] Step 3-2-7) takes the maximum queue length calculated in step 3-2-6) within the cycle as the maximum queue length value of the cycle.

[0069] Compared with the prior art, the present invention has the following beneficial effects:

[0070] (1) High estimation accuracy: The present invention takes into account the defects of existing queue distribution estimation methods, such as being mainly based on mobile network vehicle trajectory data, and proposes a queue distribution estimation method driven by electric police data, which realizes the efficient reuse of electric police data and does not need to be based on the "first come, first served" assumption in the traditional input-output model. It allows vehicles to overtake, which is more in line with the actual situation and improves the estimation accuracy and operability.

[0071] (2) High scalability: Although the present invention assumes that the travel time of vehicles in free flow satisfies the log-normal distribution, its model can be extended to other travel time distribution scenarios, which has high scalability. In addition, the present invention is also applicable to other data sources with the same detection mechanism, such as new microwave radar data, etc., which has high scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1is a flow chart of the method of the present invention;

[0073] Figure 2 It is a schematic diagram of the research scenario of the present invention;

[0074] Figure 3 It is a virtual cross-section schematic diagram of the mid-upstream intersection of the present invention;

[0075] Figure 4 A schematic diagram of the cumulative arrival curve of the upstream intersection of the present invention;

[0076] Figure 5 is a schematic diagram of a cumulative arrival curve of a virtual cross section of an upstream intersection of the present invention;

[0077] Figure 6 Schematic diagram of probability values ​​of virtual vehicles leaving the downstream intersection of the present invention;

[0078] Figure 7 It is a schematic diagram of the joint probability value of the number of virtual vehicles leaving the downstream intersection of the present invention;

[0079] Figure 8 is a schematic diagram of the overtaking range in the road section of the present invention;

[0080] Fig. 9 A schematic diagram of a simulation verification scenario of an embodiment of the present invention;

[0081] Fig.10 A schematic diagram of signal timing in a simulation verification of an embodiment of the present invention;

[0082] Fig.11 This is a schematic diagram of a cycle-level maximum queue length estimation result according to an embodiment of the present invention;

[0083] Fig.12 A schematic diagram of visualizing cycle-level estimation results according to an embodiment of the present invention;

[0084] Fig.13 The joint distribution of the coverage rate and the deviation value index during the red light period of the embodiment of the present invention;

[0085] Fig.14 The joint distribution of the coverage rate and the deviation value index during the green light period of the embodiment of the present invention;

[0086] Fig.15 It is the sensitivity analysis result of the mean absolute error index under the condition of full sample under the change of saturation parameter in the embodiment of the present invention;

[0087] Fig.16 It is the sensitivity analysis result of the mean absolute error index under the condition of non-saturation period under the saturation parameter change of the embodiment of the present invention;

[0088] Fig.17 It is the sensitivity analysis result of the mean absolute error index under the saturation parameter change under the saturation cycle condition of the embodiment of the present invention;

[0089] Fig.18 It is the sensitivity analysis result of the mean absolute percentage error index under the condition of full sample under the change of saturation parameter in the embodiment of the present invention;

[0090] Fig.19 It is the sensitivity analysis result of the mean absolute percentage error index under the condition of non-saturation cycle in the embodiment of the present invention under the change of saturation parameter;

[0091] Fig. 20 It is the sensitivity analysis result of the mean absolute percentage error index under the saturation parameter change under the saturation cycle condition of the embodiment of the present invention;

[0092] Fig.21 The sensitivity analysis result of the coverage index under the change of saturation parameter in the embodiment of the present invention is shown;

[0093] Fig. 22 It is the sensitivity analysis result of the deviation value indicator under the change of saturation parameter of the embodiment of the present invention. DETAILED DESCRIPTION

[0094] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0095] This embodiment provides a road queue distribution estimation method based on multi-section electric police data, such as Figure 1 As shown, the following steps are included:

[0096] Step 1) The timestamp information of vehicles passing through the stop line in each direction at the upstream and downstream intersections is obtained through the electric police equipment.

[0097] In the upstream and downstream intersection scenario, there are three sources of vehicles leaving the downstream in any direction: turning left from the upstream intersection, going straight from the upstream intersection, and turning right from the upstream intersection. The flow directions of vehicles leaving the downstream include: turning left downstream, going straight downstream, and turning right downstream. Figure 2 As shown in FIG. 1 , for the downstream straight-going vehicles, their sources include the upstream left-turning vehicle M5, the upstream straight-going vehicle M2, and the upstream right-turning vehicle M8, so there are 9 situations in total.

[0098] The electric alarm equipment is arranged at the entrance of each entrance lane of the intersection, and is used to collect the timestamp information of each flow of vehicles passing through the stop line. The information collected by the electric alarm equipment is lane-level information.

[0099] Step 2) Set the virtual cross section of the upstream intersection, reconstruct the cumulative vehicle arrival curve of the virtual cross section of the upstream intersection based on the data captured by the electric police equipment, and calculate the cumulative vehicle departure curve of the downstream intersection.

[0100] Step 2-1) Set the virtual cross section of the upstream intersection as the upstream arrival cross section. Since different flows need different times to pass through the upstream intersection, the present invention sets a virtual cross section at the entrance of the road section as the upstream arrival cross section, such as Figure 3 shown.

[0101] Step 2-2) Based on the data captured by the upstream electric police equipment and the average time required for vehicles of different directions to pass through the intersection, the timestamp of each vehicle passing through the virtual section of the upstream intersection is calculated:

[0102]

[0103] in, It indicates the timestamp when the i-th vehicle at the upstream intersection passes through the upstream virtual section; Indicates the timestamp when the i-th vehicle at the upstream intersection is captured by the electric police equipment in the three actual sections of the upstream intersection; Δt l ,Δt th ,Δt r Respectively represent the average time required for left-turning vehicles, straight-going vehicles, and right-turning vehicles in the upstream intersection to pass through the intersection; N l ,N th ,N r They represent the collection of left-turning vehicles, straight-going vehicles, and right-turning vehicles at the upstream intersection respectively.

[0104] Step 2-3) Based on the timestamp of each vehicle passing through the virtual section of the upstream intersection, the cumulative number of upstream arriving vehicles is solved to obtain the cumulative vehicle arrival curve of the virtual section of the upstream intersection.

[0105] Step 2-3-1) Based on the timestamp of each vehicle passing through the virtual section of the upstream intersection, according to the flow direction of the downstream departing vehicles, distinguish between downstream left turn, downstream straight, and downstream right turn, and obtain the cumulative number of vehicles arriving in the three directions of the upstream intersection at any time for different downstream flow directions:

[0106]

[0107]

[0108]

[0109] Among them, A x (t),A y (t),A z(t) represents the number of vehicles arriving at time t for upstream left turn, upstream straight and upstream right turn of a certain downstream flow direction; is the length of the upstream analysis period; Δt is the analysis time granularity; N x ,N y ,N z Respectively represent the number of vehicles captured during the analysis period for upstream left turn, upstream straight and upstream right turn of a certain downstream flow direction;

[0110] Step 2-3-2) By adding up the cumulative number of vehicles arriving in the three directions, the cumulative number of vehicles arriving at the virtual cross section of the upstream intersection at any time in a certain downstream direction is obtained:

[0111] A(t)=A x (t)+A y (t)+A z (t) (5)

[0112] Step 2-3-3) Match the cumulative number of arriving vehicles A(t) at the timestamp to obtain the cumulative vehicle arrival curve of the virtual section of the upstream intersection.

[0113] Taking the downstream straight flow of vehicles as an example, the calculated cumulative vehicle arrival curve of the upstream intersection virtual section is as follows: Figure 4 shown.

[0114] Step 2-4) Based on the data captured by the downstream electric police equipment, the cumulative number of vehicles leaving the downstream is calculated to obtain the cumulative vehicle departure curve of the downstream intersection:

[0115]

[0116] Where N represents the number of vehicles captured in the downstream flow direction within the length of the analysis period; The length of the downstream analysis period, usually equal to the length of the upstream analysis period ; Δt is the analysis time granularity, which is consistent with the value of the upstream arrival curve process parameter; D(·) represents the function of the number of downstream leaving vehicles over time, and there is a certain offset in time compared with the upstream arriving vehicles, TT f It represents the free flow travel time of vehicles in the fixed travel time scenario, and d1 represents the signal control delay of the first vehicle in the team.

[0117] Step 3) Considering the distribution of vehicle free flow travel time, the virtual vehicle departure curve of the downstream intersection is calculated for the assumptions of fixed travel time scenario and variable travel time scenario respectively. By analyzing the relationship between the cumulative vehicle arrival curve of the virtual section of the upstream intersection, the cumulative vehicle departure curve of the downstream intersection, and the virtual vehicle departure curve of the downstream intersection, the queue distribution of the road section is restored.

[0118] Based on the assumptions of fixed travel time scenario and variable travel time scenario, the virtual departure curve of vehicles at the downstream intersection is constructed to represent the departure timestamp of upstream arriving vehicles theoretically passing the downstream stop line when there is no signal light at the downstream intersection, such as Figure 5 As shown. During the cycle, the difference between the cumulative vehicle arrival curve of the virtual section of the upstream intersection and the virtual vehicle departure curve of the downstream vehicles is the number of vehicles traveling in the road section, and the difference between the virtual departure curve of the downstream vehicles and the cumulative vehicle departure curve of the downstream intersection is the number of vehicles currently queuing. Based on this, the queue distribution in the studied period is estimated.

[0119] For the fixed travel time scenario, that is, the scenario where the free travel time of vehicles in the road section is a fixed value, restoring the queue distribution of the road section specifically includes the following steps:

[0120] Step 3-1-1) Determine the virtual departure curve of the vehicle at the downstream intersection under the fixed travel time scenario, which is the horizontal offset of the cumulative arrival curve of the upstream vehicles:

[0121]

[0122] Where P(t) represents the number of vehicles that virtually pass the downstream stop line of the road section at time t, that is, the number of vehicles that can theoretically pass the downstream intersection without a traffic light; TT f represents the free flow travel time of vehicles under the assumed fixed travel time scenario; is the length of the upstream analysis period; N represents the cumulative number of vehicles arriving upstream;

[0123] Step 3-1-2) Based on the virtual departure curve of the vehicle at the downstream intersection, calculate the queue length value q(t) at the downstream intersection at any time t, that is, the queue length distribution:

[0124] q(t)=P(t)-D(t) (8)

[0125] Where q(t) represents the number of queued vehicles at the downstream intersection at time t; P(t) and D(t) represent the number of virtual vehicles leaving the downstream intersection and the number of actual vehicles leaving the downstream intersection at time t, respectively;

[0126] Step 3-1-3) Based on the obtained queue distribution within the period, determine the maximum queue length within the period:

[0127]

[0128] TT′ n+1,1 =P -1 (q′) (10)

[0129] Where q(t′) represents the maximum queue length of the analyzed period; t′ represents the time when the maximum queue length is obtained; r n+1 Indicates the time when the red light turns on in the next cycle; TT′ n+1,1 Indicates the time point where the virtual departure curve of the downstream intersection intersects the actual departure curve. When TT′ n+1,1 <r n+1 When TT′ is 0, it means that before the end of the current cycle, the virtual departure curve of the downstream intersection has intersected with the actual departure curve, and the supply and demand are balanced or the supply is greater than the demand until the end of the cycle, that is, the cycle is not saturated. At this time, the maximum queue length in the cycle is the number of queued vehicles corresponding to this intersection. When TT′ n+1,1 ≥r n+1 When P is greater than 0, it means that the intersection of the virtual departure curve and the actual departure curve of the downstream intersection is located after the cycle, that is, in the current cycle, the road section has been in a state where demand is greater than supply, and oversaturation has occurred; P -1 (·) represents the inverse function of the virtual departure curve P(·), that is, the time corresponding to the virtual departure curve when the number of vehicles is a certain value; q′ is the number of vehicles corresponding to the intersection of the virtual departure curve and the actual departure curve at the downstream intersection.

[0130] For variable travel time scenarios, even under free flow conditions, vehicles may have different free flow travel times due to differences in driver behavior, driving age, weather, environment and other factors. Therefore, the free flow travel time of a moving vehicle should satisfy a certain distribution, not a fixed value. The present invention assumes that the travel time distribution of vehicles under free flow conditions satisfies the log-normal distribution, but it can be extended to other free flow travel time distribution scenarios. Restoring the queue distribution of a road section specifically includes the following steps:

[0131] Step 3-2-1) Based on the cumulative arrival vehicle curve A(t) of the virtual section of the upstream intersection and the actual departure vehicle curve D(t) of the downstream intersection, it is assumed that the free flow travel time of the vehicle satisfies the log-normal distribution, that is, Get the probability density function:

[0132]

[0133] Among them, TT i represents the vehicle free flow travel time in the variable travel time scenario, represents the mean of the lognormal distribution, σ 2 represents the variance of the lognormal distribution;

[0134] Step 3-2-2) In the variable travel time scenario, for any vehicle i passing through the upstream virtual section, at any given time t, whether the vehicle can reach the downstream intersection is different from the fixed travel time scenario (0-1 variables, indicating whether it can reach or not, respectively), and should be a probability value, such as Figure 6 As shown, the corresponding probability calculation method is:

[0135]

[0136] Among them, P t,i It represents the probability value of whether any vehicle i passing through the upstream virtual section can reach the downstream intersection at any given time t; It refers to the timestamp of any vehicle i passing through the upstream virtual section. For each vehicle passing through the upstream virtual section, the probability value of whether it can pass through the downstream intersection at any time t can be calculated according to the above formula.

[0137] Step 3-2-3) Calculate the probability value that the number of virtual vehicles leaving the downstream intersection is n at any time t, such as Figure 7 As shown, the calculation method is:

[0138]

[0139] Among them, P t,n represents the probability value that the number of virtual vehicles leaving the downstream intersection is n at any time t; C represents the sample space, that is, the different divisions of queued vehicles and non-queued vehicles in the current cycle; A represents queued vehicles, and B represents non-queued vehicles;

[0140] In this step, if the first-in-first-out (FIFO) principle in the traditional input-output model is used, when the number of virtual vehicles leaving the downstream intersection is n, there is only one sample division scheme, that is, the first n vehicles passing through the upstream virtual section can pass through the downstream intersection (i.e., set A), and the subsequent vehicles passing through the upstream virtual section cannot reach the downstream intersection (i.e., set B); however, in the actual process, there is an overtaking phenomenon in the road section. When the number of virtual vehicles leaving the downstream intersection is n, there are multiple sample division schemes. Therefore, the present invention takes this phenomenon into consideration when calculating the probability;

[0141] Step 3-2-4) Consider the vehicle overtaking phenomenon, calculate the probability of various combination scenarios, and obtain the effective sample space number:

[0142]

[0143] Among them, N(C) represents the number of valid sample spaces, that is, when the number of virtual vehicles leaving the downstream intersection is n, the number of possible combinations of vehicles in the cycle; δ represents the overtaking range of each vehicle in the current scenario. Due to the speed limit of urban roads and actual driving conditions, this value has a certain range, such as Figure 8 As shown; It represents the total number of vehicles passing through the upstream virtual section at any time t;

[0144] Step 3-2-5) Calculate the mean value of the number of virtual vehicles leaving the downstream intersection at any time t, as shown in formula (15), and on this basis, taking into account the volatility of vehicle travel time, calculate the upper and lower bounds of the number of virtual vehicles leaving the downstream intersection, as shown in formulas (16)-(17):

[0145]

[0146]

[0147]

[0148] in, They represent the mean, upper bound and lower bound of the number of virtual vehicles leaving the downstream intersection at any time t respectively;

[0149] Step 3-2-6) Calculate the mean, upper and lower bounds of the queue length at any time t based on the mean, upper and lower bounds of the number of virtual vehicles leaving the downstream intersection at any time t:

[0150]

[0151]

[0152]

[0153] in, They refer to the average value, upper bound and lower bound of the queue length at any time t respectively;

[0154] Step 3-2-7) takes the maximum queue length calculated in step 3-2-6) within the cycle as the maximum queue length value of the cycle.

[0155] This embodiment is based on the micro-simulation software VISSIM, and takes the intersections of Tongjiang Middle Road-Hanjiang Road (downstream intersection) and Tongjiang Middle Road-Zhujiang Road (upstream intersection) in Changzhou as verification scenarios to build a simulation model, and calibrates the simulation parameters based on the real electric police data collected at the intersection. The Tongjiang Middle Road-Hanjiang Road South entrance through lane is selected for calculation in the verification phase; the saturation of the Tongjiang Middle Road-Hanjiang Road South entrance through lane during the study period is 0.6, and the proportion of heavy vehicles is 5%. Fig. 9 To verify the intersection diagram of the analysis, Fig.10 The signal timing diagram for the Tongjiang Middle Road-Hanjiang Road intersection for verification and analysis.

[0156] Based on the above method, this embodiment verifies the queue distribution estimation in the fixed travel time scenario and the variable travel time scenario (hereinafter referred to as fixed travel time and variable travel time), and selects the following indicators for analysis:

[0157]

[0158]

[0159]

[0160]

[0161] Among them, MAE k,1 ,MAE k,2 They refer to the mean absolute error in fixed travel time scenarios and variable travel time scenarios respectively; MAPE k,1 ,MAPE k,2 They refer to the mean absolute percentage errors in the fixed travel time scenario and the variable travel time scenario respectively; Q refers to the true value of the maximum queue length in the period; Refers to the maximum queue length estimate under a fixed travel time scenario; Refers to the upper bound estimate of the maximum queue length in a variable travel time scenario.

[0162] On this basis, this embodiment further selects the queue distribution interval coverage (hereinafter referred to as coverage) and the second-level queue length value deviation (hereinafter referred to as deviation) indicators for quantitative analysis:

[0163]

[0164]

[0165] in, Refers to the coverage rate of the queue distribution interval in period k, which is used to evaluate whether the estimated result interval (the interval composed of the queue length upper bound and the queue length lower bound) covers the actual queue length value. This indicator is a percentile; C k represents the time length of cycle k; It represents the length of time that the true value of the queue length is covered by the estimated interval within period k; Indicates the deviation of the queue length value in seconds for period k; It represents the deviation of the queue length value at the i-th second in cycle k.

[0166] This embodiment is based on the constructed simulation model and sets the simulation duration to 21800 seconds, which includes 2600 seconds of simulation warm-up time and 100 valid cycle samples. First, the mean absolute value error and mean absolute percentage error of the maximum queue length index of each cycle can be verified based on formulas (21-24), as follows: Fig.11 As shown. Fig.11 It can be seen that in terms of mean absolute error, the fixed travel time scenario is 1.42 vehicles / cycle, and the variable travel time scenario is 1.13 vehicles / cycle; in terms of mean absolute percentage error, the fixed travel time scenario is 10.43%, and the variable travel time scenario is 8.53%. The estimation result in the variable travel time scenario is significantly better than that in the fixed travel time scenario. Fig.11 It can be seen that the maximum queue length estimate in the variable travel time scenario has fewer outliers than in the fixed travel time scenario.

[0167] For each period, a graph can be drawn based on four verification indicators, such as Fig.12 As shown, there are four graphs, namely, vehicle distribution under fixed travel time, vehicle distribution under variable travel time, queue distribution under variable travel time, and second-level deviation. Combining the first two graphs with the accuracy of the above estimation results, it can be seen that the method of the present invention can estimate the maximum queue length within the cycle more accurately. On this basis, it can be seen from the queue distribution graph that the estimated upper and lower bounds of the queue length can more accurately include the true value of the queue length. In addition, it can be seen from the second-level deviation graph that the method of the present invention can obtain better coverage and deviation regardless of whether it is during the red light period or the green light period.

[0168] Furthermore, this embodiment analyzes the estimated distribution of queue lengths during red light periods and green light periods, such as Figure 13-14 As shown, the distribution relationship between the estimated deviation value and the coverage rate during the red and green light periods is represented respectively. Fig.13 The distribution during the red light period is shown to have a coverage rate of 54% and a deviation value of only 0.85 vehicles / cycle; Fig.14 From the distribution during the green light period, it can be seen that the coverage rate is 61.5% and the deviation value is 3.75 vehicles / cycle; in general, the total coverage rate of the cycle is 56.84% and the deviation value is 0.81 vehicles / cycle. From the results, it can be seen that due to the fluctuation of traffic volume, the coverage rate of the estimated interval is limited, but the deviation value is only 0.81 vehicles / cycle, which shows that the queue distribution estimated by the present invention can restore the queue status more accurately.

[0169] A simulation model is constructed based on real scenarios, saturation parameters are selected, and sensitivity analysis is performed on the constructed queuing distribution estimation method.

[0170] The sensitivity analysis parameter selected in this embodiment is saturation, with a value of 0.4 to 1.2 (interval of 0.2) to study the difference in the estimation accuracy of the method of the present invention with the change of saturation. The results are as follows: Figure 15-Figure 20 As shown. The results show that both the mean absolute error and the mean absolute percentage error indicators show that the method of the present invention is insensitive to changes in saturation. For the full sample data, when the saturation reaches 1.2, the mean absolute error under variable travel time is still less than 2 vehicles / cycle; specifically, for unsaturated cycle data, the mean absolute error under variable travel time is less than 2 vehicles / cycle; for oversaturated cycle data, the mean absolute error under variable travel time is about 2 vehicles / cycle, which is less sensitive to saturation; in terms of mean absolute percentage error, it can be stabilized at about 10%.

[0171] Furthermore, for the coverage and deviation value verification indicators, the sensitivity analysis under the condition of saturation change is as follows: Figure 21-22 As shown in the results, the coverage index has obvious stability. When the saturation increases, the coverage rate does not decrease significantly. In addition, for the deviation index, the deviation values ​​within the cycle and during the red light period can be maintained at a certain level, and at any saturation level, the main error comes from the green light period, which may come from the dissipation process of vehicles in the simulation model, that is, the number of queued vehicles in the simulation model drops sharply when the green light turns on, while in reality, the dissipation of the queue length is a gradual process, which is one of the main sources of error.

[0172] To sum up, the present invention aims at the defects of existing queue distribution estimation methods, such as being mainly based on mobile connected vehicle trajectory data, and provides a queue distribution estimation method based on multi-section electric police data. It does not require the "first come, first served" assumption in the traditional input-output model, allows vehicles to overtake, and is more in line with the actual situation. In addition, it can output the distribution interval of the length of vehicles in the queue at the second level, and has the advantages of strong operability and high scalability.

[0173] It should be noted that the terms "first", "second", etc. in the specification, claims and drawings of the present invention are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0174] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning, or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A road queue distribution estimation method based on multi-section electric police data, characterized in that: The following steps are involved: Step 1) obtaining the time stamp information of vehicles passing through the stop line in each direction of the upstream and downstream intersections through the electric police equipment; Step 2) setting a virtual cross section of the upstream intersection, reconstructing a cumulative vehicle arrival curve of the virtual cross section of the upstream intersection based on data captured by the electric police equipment, and calculating a cumulative vehicle departure curve of the downstream intersection; Step 3) Considering the distribution of vehicle free flow travel time, the virtual vehicle departure curve of the downstream intersection is calculated for the assumptions of fixed travel time scenario and variable travel time scenario respectively, and the queue distribution of the road section is restored by analyzing the relationship between the cumulative vehicle arrival curve of the virtual section of the upstream intersection, the cumulative vehicle departure curve of the downstream intersection, and the virtual vehicle departure curve of the downstream intersection; The step 3) is specifically as follows: constructing a virtual vehicle departure curve of the downstream intersection for the assumptions of a fixed travel time scenario and a variable travel time scenario, characterizing the departure timestamp of the upstream arriving vehicle theoretically passing through the downstream stop line when there is no traffic light at the downstream intersection, the difference between the cumulative vehicle arrival curve of the virtual section of the upstream intersection and the virtual vehicle departure curve of the downstream vehicle is the number of vehicles traveling in the road section, the difference between the virtual vehicle departure curve of the downstream vehicle and the cumulative vehicle departure curve of the downstream intersection is the number of vehicles currently queuing, and based on this, estimating the queue distribution in the studied period; For the fixed travel time scenario, that is, the scenario where the free travel time of vehicles in the road section is a fixed value, restoring the queue distribution of the road section specifically includes the following steps: Step 3-1-1) Determine the virtual departure curve of the vehicle at the downstream intersection under the fixed travel time scenario, which is the horizontal offset of the cumulative arrival curve of the upstream vehicles: Where P(t) represents the number of vehicles that virtually pass the downstream stop line of the road section at time t, that is, the number of vehicles that can theoretically pass the downstream intersection without a traffic light; TT f represents the free flow travel time of vehicles under the assumed fixed travel time scenario; is the length of the upstream analysis period; N represents the cumulative number of vehicles arriving upstream; Step 3-1-2) Based on the virtual departure curve of the vehicle at the downstream intersection, calculate the queue length value q(t) at the downstream intersection at any time t, that is, the queue length distribution: q(t)=P(t)-D(t) Where q(t) represents the number of queued vehicles at the downstream intersection at time t; P(t) and D(t) represent the number of virtual vehicles leaving the downstream intersection and the number of actual vehicles leaving the downstream intersection at time t, respectively; Step 3-1-3) Based on the obtained queue distribution within the period, determine the maximum queue length within the period: TT′ n+1,1 =P -1 (q′) Where q(t′) represents the maximum queue length of the analyzed period; t′ represents the time when the maximum queue length is obtained; r n+1 Indicates the time when the red light turns on in the next cycle; TT′ n+1,1 Indicates the time point where the virtual departure curve of the downstream intersection intersects the actual departure curve. When TT′ n+1,1 <r n+1 When TT′ is 0, it means that before the end of the current cycle, the virtual departure curve of the downstream intersection has intersected with the actual departure curve, and the supply and demand are balanced or the supply is greater than the demand until the end of the cycle, that is, the cycle is not saturated. At this time, the maximum queue length in the cycle is the number of queued vehicles corresponding to this intersection. When TT′ n+1,1 ≥r n+1 When P is greater than 0, it means that the intersection of the virtual departure curve and the actual departure curve of the downstream intersection is located after the cycle, that is, in the current cycle, the road section has been in a state where demand is greater than supply, and oversaturation has occurred; P -1 (·) represents the inverse function of the virtual departure curve P(·), that is, the time corresponding to the virtual departure curve when the number of vehicles is a certain value; q′ is the number of vehicles corresponding to the intersection of the virtual departure curve and the actual departure curve at the downstream intersection; For the variable travel time scenario, assuming that the free flow travel time of the moving vehicles satisfies the log-normal distribution, the following steps are specifically included to restore the queue distribution of the road section: Step 3-2-1) Based on the cumulative arrival vehicle curve A(t) of the virtual section of the upstream intersection and the actual departure vehicle curve D(t) of the downstream intersection, it is assumed that the free flow travel time of the vehicle satisfies the log-normal distribution, that is, Get the probability density function: Among them, TT i represents the vehicle free flow travel time in the variable travel time scenario, represents the mean of the lognormal distribution, σ 2 represents the variance of the lognormal distribution; Step 3-2-2) In the variable travel time scenario, for any vehicle i passing through the upstream virtual section, the probability value of whether the vehicle can reach the downstream intersection at any given time t is: Among them, P t,i It represents the probability value of whether any vehicle i passing through the upstream virtual section can reach the downstream intersection at any given time t; It refers to the timestamp of any vehicle i passing through the upstream virtual section; Step 3-2-3) Calculate the probability value that the number of virtual vehicles leaving the downstream intersection is n at any time t: Among them, P t,n represents the probability value that the number of virtual vehicles leaving the downstream intersection is n at any time t; C represents the sample space, that is, the different divisions of queued vehicles and non-queued vehicles in the current cycle; A represents queued vehicles, and B represents non-queued vehicles; Step 3-2-4) Consider the vehicle overtaking phenomenon, calculate the probability of various combination scenarios, and obtain the effective sample space number: Where N(C) represents the number of valid sample spaces, that is, the number of possible combinations of vehicles in a cycle when the number of virtual vehicles leaving the downstream intersection is n; δ represents the overtaking range of each vehicle in the current scenario; It represents the total number of vehicles passing through the upstream virtual section at any time t; Step 3-2-5) Calculate the mean of the number of virtual vehicles leaving the downstream intersection at any time t, and on this basis, taking into account the volatility of vehicle travel time, calculate the upper and lower bounds of the number of virtual vehicles leaving the downstream intersection: in, They represent the mean, upper bound and lower bound of the number of virtual vehicles leaving the downstream intersection at any time t respectively; Step 3-2-6) Calculate the mean, upper and lower bounds of the queue length at any time t based on the mean, upper and lower bounds of the number of virtual vehicles leaving the downstream intersection at any time t: in, They refer to the average value, upper bound and lower bound of the queue length at any time t respectively; Step 3-2-7) takes the maximum queue length calculated in step 3-2-6) within the cycle as the maximum queue length value of the cycle.

2. The method for estimating road queue distribution based on multi-section electric police data according to claim 1 is characterized in that: In the upstream and downstream intersection scenario, there are three sources of departing vehicles in any direction downstream: turning left from the upstream intersection, going straight from the upstream intersection, and turning right from the upstream intersection. The directions of departing vehicles downstream include: turning left downstream, going straight downstream, and turning right downstream.

3. The method for estimating road queue distribution based on multi-section electric police data according to claim 1 is characterized in that: The electric alarm equipment is arranged at the entrance of each entrance lane of the intersection, and is used to collect the timestamp information of each flow of vehicles passing through the stop line. The information collected by the electric alarm equipment is lane-level information.

4. The method for estimating road queue distribution based on multi-section electric police data according to claim 1 is characterized in that: The step 2) comprises the following steps: Step 2-1) Setting the upstream intersection virtual section as the upstream arrival section; Step 2-2) Based on the data captured by the upstream electric police equipment and the average time required for vehicles of different directions to pass through the intersection, the timestamp of each vehicle passing through the virtual section of the upstream intersection is calculated; Step 2-3) based on the timestamp of each vehicle passing through the virtual section of the upstream intersection, solve the cumulative number of upstream arriving vehicles to obtain the cumulative vehicle arrival curve of the virtual section of the upstream intersection; Step 2-4) Based on the data captured by the downstream electric police equipment, the cumulative number of vehicles leaving the downstream is calculated to obtain the cumulative vehicle departure curve of the downstream intersection.

5. The method for estimating road queue distribution based on multi-section electric police data according to claim 4 is characterized in that: The calculation method for the timestamp of each vehicle passing through the virtual section of the upstream intersection is: in, It indicates the timestamp when the i-th vehicle at the upstream intersection passes through the upstream virtual section; Indicates the timestamp when the i-th vehicle at the upstream intersection is captured by the electric police equipment in the three actual sections of the upstream intersection; Δt l ,Δt th ,Δt r Respectively represent the average time required for left-turning vehicles, straight-going vehicles, and right-turning vehicles in the upstream intersection to pass through the intersection; N l ,N th ,N r They represent the collection of left-turning vehicles, straight-going vehicles, and right-turning vehicles at the upstream intersection respectively.

6. The method for estimating road queue distribution based on multi-section electric police data according to claim 4 is characterized in that: The steps 2-3) include the following steps: Step 2-3-1) Based on the timestamp of each vehicle passing through the virtual section of the upstream intersection, according to the flow direction of the downstream departing vehicles, distinguish the downstream left turn, downstream straight, and downstream right turn, and obtain the cumulative number of vehicles arriving in the three directions of the upstream intersection at any time for different downstream flow directions: Among them, A x (t),A y (t),A z (t) represents the number of vehicles arriving at time t for upstream left turn, upstream straight and upstream right turn of a certain downstream flow direction; is the length of the upstream analysis period; Δt is the analysis time granularity; N x ,N y ,N z Respectively represent the number of vehicles captured during the analysis period for upstream left turn, upstream straight and upstream right turn of a certain downstream flow direction; Step 2-3-2) By adding up the cumulative number of vehicles arriving in the three directions, the cumulative number of vehicles arriving at the virtual cross section of the upstream intersection at any time in a certain downstream direction is obtained: A(t)=A x (t)+A y (t)+A z (t) Step 2-3-3) Match the cumulative number of arriving vehicles A(t) at the timestamp to obtain the cumulative vehicle arrival curve of the virtual section of the upstream intersection.

7. The method for estimating road queue distribution based on multi-section electric police data according to claim 4 is characterized in that: The method for calculating the cumulative number of downstream leaving vehicles based on the data captured by the downstream electric alarm equipment to obtain the cumulative vehicle leaving curve of the downstream intersection is: Where N represents the number of vehicles captured in the downstream flow direction within the length of the analysis period; is the length of the downstream analysis period; Δt is the analysis time granularity, which is consistent with the value of the upstream arrival curve process parameter; D(·) represents the function of the number of downstream leaving vehicles over time, and the time is offset compared with the upstream arriving vehicles, TT f It represents the free flow travel time of vehicles in the fixed travel time scenario, and d1 represents the signal control delay of the first vehicle in the team.

Images