Large transportation mixed traffic flow load effect simulation method based on cellular automaton

Abstract

The invention discloses a large transport load effect evaluation method and system based on mixed traffic flow simulation, and belongs to the field of bridge safety monitoring. The method comprises the following steps: establishing an adaptive statistical model based on dynamic weighing data, and constructing a high-speed random departure model by using a Monte Carlo method; a cellular automaton is combined to simulate a dynamic evolution process of a mixed traffic flow in an HHT scene, and a load effect calculation system is established through a vehicle-bridge coupled vibration equation. The innovation point lies in that probability distribution fitting test and traffic flow microscopic simulation are fused, and the model precision is adaptively corrected and optimized through parameters. Actual measurement verification shows that the simulation vehicle speed error of the model is less than 1%, and the load effect extreme value median error of 61 days is less than 2.8%. Compared with a traditional method, the load effect extreme value is rapidly solved, dynamic load input is provided for bridge safety assessment, and the problem that in the prior art, the mixed traffic flow load effect assessment precision is insufficient is solved.

Description

Technical Field

[0001] The present invention belongs to the cross - field of bridge engineering and traffic engineering, and specifically relates to a rapid calculation method for bridge load effects of mixed traffic flow in the heavy - haul transportation (HHT) scenario based on dynamic weighing data and cellular automata. Background Art

[0002] In the modern transportation system, bridges, as key infrastructure connecting regions and promoting economic development, are of crucial safety importance. With the rapid economic development, the status of heavy - haul transportation (HHT) in the national economy has become increasingly prominent. However, due to their over - sized dimensions and weights, HHT vehicles pose significant challenges to the safety of bridge structures. When these vehicles pass through bridges, they generate huge load effects, which may lead to damage or even destruction of the bridge structure. Therefore, accurately evaluating the impact of HHT vehicles on bridge structures is of great significance for ensuring bridge safety.

[0003] Currently, the research on the load effects of HHT vehicles crossing bridges mainly focuses on their single - lane passing scenarios, without fully considering the influence of the mixed traffic of HHT vehicles and social vehicles in the actual traffic flow. In real traffic, the mixed traffic of HHT vehicles and social vehicles is the norm, and this mixed traffic flow will generate more complex and variable load effects. If traditional methods are still used for calculation, the results will deviate significantly from the actual situation, thus making it impossible to accurately evaluate the safety of bridge structures.

[0004] Existing traffic flow simulation methods (such as traditional cellular automata models) have problems with insufficient accuracy in vehicle position refinement, lane - changing behavior modeling, and dynamic parameter integration, and are difficult to meet the high - efficiency calculation requirements of load effects in complex mixed traffic flows. Summary of the Invention

[0005] The present invention belongs to the cross - field of bridge engineering and traffic engineering, and specifically relates to a rapid calculation method for bridge load effects of mixed traffic flow in the heavy - haul transportation scenario based on dynamic weighing data and cellular automata. The method specifically includes the following steps:

[0006] S1. Collect and pre - process highway dynamic weighing data (WIM) and statistically analyze traffic flow characteristic parameters.

[0007] S2. Construct a departure model for high - speed mixed traffic flow

[0008] S3. Construct an improved cellular automata (CA) traffic flow model

[0009] S4. Calculate bridge load effects and verify and optimize the model

[0010] S1 specifically includes the following steps:

[0011] S101. Data collection: Through a dynamic weighing system (WIM) deployed on a certain one-way three-lane highway, data such as the driving speed, lane position, total weight, axle weight, axle distance, vehicle length, and passing time of vehicles are collected in real time. These data constitute the original dataset, providing a basis for subsequent analysis and model construction. The data collection of the WIM system covers information such as vehicle driving speed, the lane where the vehicle is located, vehicle total weight, the distribution of each axle, the time when the vehicle passes the monitoring point, vehicle length, and the distribution of axle distances.

[0012] S102. Data screening: In the collected original data, there are some abnormal data, such as over-limit vehicle speeds and negative axle distances. These abnormal data will affect the accuracy of subsequent analysis, so they need to be eliminated. The specific steps are as follows:

[0013] Eliminate abnormal data: By setting reasonable thresholds, eliminate data with vehicle speeds exceeding the highway design speed limit or lower than the minimum speed limit, as well as data with negative or significantly unreasonable axle distances.

[0014] Retain valid data: After eliminating abnormal data, the remaining data are all valid data, which will be used for subsequent vehicle type classification and traffic flow characteristic analysis.

[0015] Vehicle type category division: According to the number of axles of the vehicle, the vehicle is divided into different categories such as two-axle vehicles, three-axle vehicles, four-axle vehicles, five-axle vehicles, and six-axle vehicles. This classification method helps to analyze the traffic flow characteristics of different vehicle types subsequently.

[0016] S103. Vehicle type classification: Classify vehicles based on the axle type combination and count the proportion of vehicle types in each lane. The specific classification method is as follows:

[0017] Axle type combination classification: According to the axle type combination of the vehicle, the vehicle is divided into different categories. For example, two-axle vehicles, three-axle vehicles, etc. This classification method can more accurately reflect the structural characteristics of the vehicle.

[0018] Count the proportion of vehicle types in each lane: Count the vehicle types in each lane and analyze the vehicle type distribution in each lane.

[0019] Traffic volume conversion: Perform traffic volume conversion according to the "Automobile Design" standard. Through traffic volume conversion, the traffic volume of different vehicle types can be converted into the traffic volume of standard vehicle types, which is convenient for subsequent traffic flow characteristic analysis and model construction.

[0020] S104. Analyze the processed traffic volume

[0021] Average daily traffic volume statistics: Statistically analyze the average daily traffic volume of each lane.

[0022] Hourly traffic volume change rule: Analyze the hourly traffic volume change rule of each lane.

[0023] Comparative design traffic volume: Compare the measured traffic volume with the design traffic volume.

[0024] S105. Perform distribution fitting on the collected vehicle speeds, normal distribution fitting: Based on the normal distribution, fit the vehicle speed parameters of vehicles with different axle numbers in each lane to determine the expectation and standard deviation of the vehicle speeds in each lane.

[0025] Vehicle speed expectation and standard deviation: The vehicle speed expectations and standard deviations of different axle numbers in the slow lane, fast lane, and overtaking lane

[0026] S106. Headway calculation: Headway back-calculation: Back-calculate the headway through vehicle speed and time interval, and calculate the rear overhang and front overhang in combination with the "Motor Vehicle Operating Safety Technical Conditions" (GB 7258-2017).

[0027] Log-normal distribution fitting: Perform log-normal distribution fitting on the headways under different service level conditions

[0028] S106. Vehicle weight and axle weight distribution fitting:

[0029] Two-axle and three-axle vehicles: Use log-normal distribution to fit the vehicle weight data;

[0030] Four-axle, five-axle, and six-axle vehicles: Use multi-peak mixed Gaussian distribution to fit the vehicle weight data;

[0031] S107. Statistically analyze the vehicle length and wheelbase, and input the vehicle length and wheelbase distribution coefficients of each vehicle type based on the fixed mean.

[0032] S2. Construct a departure model for high-speed mixed traffic flow, which specifically includes the following steps:

[0033] S201. Monte Carlo random sampling

[0034] Input parameters: Use the distribution parameters of vehicle speed, vehicle weight, wheelbase, etc. obtained by fitting in step 1 as inputs. These parameters include the vehicle speed expectations and standard deviations of vehicles with different axle numbers in each lane, vehicle weight distribution parameters, and wheelbase distribution coefficients, etc.

[0035] Randomly generate a vehicle sequence: Use the Monte Carlo method to randomly generate a vehicle sequence based on the input distribution parameters. Through random sampling, simulate the departure process of social vehicles.

[0036] S202. Embed HHT vehicle parameters

[0037] Determine HHT vehicle parameters: Based on HHT waybill data, determine the wheelbase, axle weight, and total vehicle and cargo weight of the tractor and trailer of HHT vehicles. The tractor of HHT vehicles does not participate in sharing the weight of the transported goods, and its wheelbase and axle weight data are directly obtained from the waybill.

[0038] Incorporate into the traffic flow sequence: Incorporate HHT vehicles into the traffic flow sequence so that they mix with social vehicles. In the model, HHT vehicles are regarded as special vehicle types with specific parameters and behavior rules.

[0039] Restrict lane-changing behavior: According to the "HHT Management Regulations", restrict the lane-changing behavior of HHT vehicles. In the model, HHT vehicles are specifically marked, and their lane-changing behavior is restricted separately in the model according to the regulations, so that their lane-changing behavior when crossing the bridge is strictly restricted, making it more in line with the actual scenario and ensuring the accuracy of the mixed traffic flow simulation.

[0040] S3. Construct an improved cellular automaton (CA) traffic flow model, which specifically includes the following steps:

[0041] S301. Refine the cell length: Refine the cell length to 0.01 meters, significantly improving the continuity of vehicle positions and making the simulation results closer to the actual situation. The refined cell length can more accurately describe the position changes of vehicles on the bridge, thereby improving the accuracy of load effect calculation.

[0042] S302. Dynamic acceleration rule: Introduce the assumption of constant acceleration and update the vehicle position in real time through the motion equation. This method can more accurately simulate the acceleration and deceleration processes of vehicles, avoiding the accumulation of position errors caused by discontinuous acceleration changes in traditional models. The dynamic acceleration rule makes the motion state of vehicles more in line with the actual situation and improves the reliability of the simulation results.

[0043] Based on the speed change rule of the traditional NS model for improvement, using the change in acceleration as the initial value, and then accurately determining the vehicle position change at each moment. The improved action rule is as follows:

[0044] ① Model acceleration rule: s i (t + Δt) = min{s i (t) + a n Δt, S max}} & and 0 < a n < a max

[0045] ② Model deceleration rule: s1(t + Δt) = min(s i (t) + a n Δ t , S min ) & and a min < a n < 0

[0046] ③ Vehicle position determination:

[0047] In the above rules, wi , s i , a i are the cell position, vehicle speed, and acceleration of vehicle i, v max and v min are the upper and lower limits of the driving speed set according to vehicle performance and road section regulations, a min and a max are the upper and lower boundaries of the acceleration value considering vehicle performance. Δt is the time step, usually 1 second.

[0048] S303, Refinement of lane-changing behavior

[0049] Based on the lane width and lane-changing angle (4° ± 1°), the states of "lane change not completed" and "lane change completed" are divided, and the lane-changing probability between lanes is set (such as the lane-changing probability of the overtaking lane is 0.5). This refined modeling of lane-changing behavior can more accurately simulate the lane-changing process of vehicles in a multi-lane scenario and improve the accuracy of traffic flow simulation. By considering the details of lane-changing behavior, the model can better reflect the complexity of the actual traffic flow.

[0050] Lane-changing rules

[0051] The simulation of the driver's lane-changing behavior in the STCA model is implemented by the following formula:

[0052] (1) Generation of lane-changing behavior: d i <min(S i + 1, S max ) & d i，l >d i

[0053] (2) Judgment of lane-changing safety: d i,b >d a

[0054] In the above rules, d i , d i,l and d i,b respectively refer to the distance judgment values between vehicle i and the vehicle in front, the vehicle in front and behind in the adjacent lane. d a is the lower limit of the vehicle distance that meets the traffic safety conditions and is used as the judgment basis for whether to execute the lane-changing behavior. On this basis, regarding the probability of vehicle lane-changing behavior, the following lane-changing rules are formulated for the cellular automaton traffic flow model.

[0055] (1) Overtaking lane

[0056] When the safety distances of the vehicles in front and behind in the adjacent lane meet the requirements, the vehicle will randomly generate a lane-changing motivation, and the probability of generating this motivation is set to 0.5. The judgment formula is as follows:

[0057] Judgment of lane-changing safety: d i,b >da

[0058] (2) Fast Lane

[0059] In the scenario where lane changes can be made to both sides simultaneously, the probability of changing lanes to the left is 0.56, the probability of changing lanes to the right is 0.14, and the probability of decelerating is 0.3; in the scenario where lane changes can only be made to one side individually, the probability of changing lanes is 0.7. The determination formula is expressed as follows:

[0060] Lane change behavior occurs: d i <min(S i +1, S max ) & d i,l >d i

[0061] Lane change safety determination: d i,b >d a

[0062] (3) Slow Lane

[0063] In the scenario that meets the conditions for lane change behavior and the basis for lane change safety determination, the vehicle will enter the fast lane with a certain probability, and this value is set to 0.7. The determination formula is expressed as follows:

[0064] Lane change safety determination: d i,b >d i

[0065] S4. Calculate the bridge load effect and verify and optimize the model, which specifically includes the following steps:

[0066] S401. Calculate the load effect through the influence line loading method. The specific steps are as follows:

[0067] Influence line generation: Based on the bridge structure, calculate the influence lines of bending moment and shear force. The bending moment influence line is used to represent the bending moment distribution generated by the vehicle load at the mid-span of the bridge, while the shear force influence line is used to represent the shear force distribution generated by the vehicle load at the supports of the bridge. These influence lines are the basis for subsequent load effect calculations and can accurately reflect the influence of vehicle loads on the bridge structure.

[0068] Load effect calculation: In the cellular automaton model, record the axle weight position of each vehicle in real time, and calculate the bending moment and shear force effect values within each time step according to the bending moment and shear force influence lines.

[0069] S402. Extreme value extraction:

[0070] During the iteration of the cellular automaton, record the maximum values of the mid-span bending moment and the support shear force daily.

[0071] S403. Verify the accuracy of the model, which specifically includes the following steps:

[0072] Traffic flow simulation verification:

[0073] Compare the errors of simulated and measured vehicle weights, vehicle speeds, and headways.

[0074] Load effect accuracy verification:

[0075] Verify the extreme values of load effects under different cell lengths through finite element software (such as Midas Civil), and select the optimal parameters.

[0076] Mixed scenario verification:

[0077] Analyze the situation where the load effect in the mixed traffic scenario exceeds that of HHT traveling alone, and verify the practicality of the model. Description of the drawings

[0078] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments. The drawings are only provided for reference and illustration, and are not intended to limit the present invention.

[0079] Figure 1 It is the traffic composition diagram of the highway section in the vehicle type classification of the first stage of the present invention;

[0080] Figure 2 It is the traffic composition diagram of the slow lane, fast lane, and overtaking lane in the vehicle type proportion statistics of each lane in the first stage of the present invention;

[0081] Figure 3 It is the average hourly traffic volume diagram of the slow lane, fast lane, and overtaking lane in the traffic volume analysis of the first stage of the present invention;

[0082] Figure 4 It is the ratio diagram of the measured traffic volume to the design value in the traffic volume analysis of the first stage of the present invention;

[0083] Figure 5 It is the departure flow chart of simulated social vehicles in the Monte Carlo random sampling of the second stage of the present invention;

[0084] Figure 6 It is the bending moment calculation result diagram in the extreme value extraction of the fourth stage of the present invention; Specific embodiments

[0085] The present invention will be further described in detail below in conjunction with the drawings and specific embodiments. The drawings are only provided for reference and illustration, and are not intended to limit the present invention.

[0086] The present invention belongs to the cross - field of bridge engineering and traffic engineering, and specifically relates to a rapid calculation method for the bridge load effect of mixed traffic flow in the large - piece transportation scenario based on dynamic weighing data and cellular automata. It specifically includes the following steps:

[0087] S1. Collect and preprocess the highway dynamic weighing data (WIM) and statistically analyze the traffic flow characteristic parameters.

[0088] S2. Build a departure model for the high-speed mixed traffic flow

[0089] S3. Build an improved cellular automaton (CA) traffic flow model

[0090] S4. Calculate the bridge load effect and verify and optimize the model

[0091] S1 specifically includes the following steps:

[0092] S101. Data collection: Through the dynamic weighing system (WIM) deployed on a one-way three-lane highway, collect data such as the driving speed, lane position, total weight, axle weight, axle distance, vehicle length, and passing time of vehicles in real time. These data constitute the original data set, providing a basis for subsequent analysis and model construction. The data collection of the WIM system covers information such as vehicle driving speed, the lane where the vehicle is located, vehicle total weight, the distribution of each axle, the time when the vehicle passes the monitoring point, vehicle length, and the distribution of each axle distance. Taking the real-time monitoring points of the WIM system arranged on a one-way three-lane highway section as the parameter source, select the data sets of 61 days in October and November 2023 to carry out research work. A total of 1,738,400 dynamic weighing data were collected.

[0093] S102. Data screening: Among the collected original data, there are some abnormal data, such as over-limit vehicle speed, negative axle distance, etc. These abnormal data will affect the accuracy of subsequent analysis, so they need to be eliminated. The specific steps are as follows:

[0094] Eliminate abnormal data: By setting reasonable thresholds, eliminate data with vehicle speeds exceeding the highway design speed limit or lower than the minimum speed limit, and data with negative or significantly unreasonable axle distances.

[0095] Retain valid data: After eliminating abnormal data, the remaining data are all valid data, which will be used for subsequent vehicle type classification and traffic flow characteristic analysis.

[0096] Vehicle type category division: According to the number of axles of the vehicle, divide the vehicle into different categories such as two-axle vehicles, three-axle vehicles, four-axle vehicles, five-axle vehicles, and six-axle vehicles. This classification method helps to analyze the traffic flow characteristics of different vehicle types subsequently.

[0097] S103. Vehicle type classification: Classify vehicles based on the axle type combination and count the proportion of vehicle types in each lane. As Figure 1 shown, the analysis shows that the vehicle type composition of this section has obvious characteristics: two-axle vehicles account for 78%, and six-axle vehicles are the second, accounting for about 12% of the total. The specific classification method is as follows:

[0098] Axle type combination classification: Vehicles are classified into different categories according to the axle type combination of the vehicle. For example, two-axle vehicles, three-axle vehicles, etc. This classification method can more accurately reflect the structural characteristics of the vehicle.

[0099] Statistical analysis of the proportion of vehicle types in each lane: The vehicle types in each lane are counted to analyze the distribution of vehicle types in each lane. As Figure 2 shown, the proportion of vehicle types in the slow lane, fast lane, and overtaking lane. Through statistical analysis, it can be found that the proportion of vehicle types in each lane basically coincides with the actual distribution of the highway.

[0100] Traffic volume conversion: Traffic volume conversion is carried out according to the "Automobile Design" standard. The specific conversion coefficients are shown in Table 1. Through traffic volume conversion, the traffic volume of different vehicle types can be converted into the traffic volume of standard vehicle types, which is convenient for subsequent analysis of traffic flow characteristics and model construction.

[0101] Table 1 Vehicle type classification standards and corresponding conversion values

[0102]

[0103] S104. Analyze the processed traffic volume

[0104] Daily average traffic volume statistics: The daily average traffic volume of each lane is counted. The results show that the daily average traffic volume of the slow lane is 19,232 vehicles / day, and the daily average traffic volumes of the fast lane and overtaking lane are 14,962 vehicles / day and 9,646 vehicles / day respectively.

[0105] Hourly traffic volume variation pattern: Analyze the hourly traffic volume variation pattern of each lane. It is found that the traffic volume peaks around 9:00 and 18:00 and drops to the lowest level of the day around 1:00 am. The specific data are shown in Figure 3 .

[0106] Compare with the designed traffic volume: Compare the measured traffic volume with the designed traffic volume. The results are as Figure 4 shown. The analysis shows that the daily traffic volume of the slow lane is generally large, and there is almost no case where it is less than the designed traffic volume; the traffic volume of the fast lane fluctuates around the designed value; while the traffic volume of the overtaking lane is generally low, and at the lowest it is less than 25% of the designed value.

[0107] S105. Fit the distribution of the collected vehicle speeds, normal distribution fitting: Based on the normal distribution, fit the vehicle speed parameters of vehicles with different axle numbers in each lane to determine the expectation and standard deviation of the vehicle speeds in each lane. The specific parameters are shown in Table 2.

[0108] Table 2 Normal distribution parameters of vehicle speeds of different vehicle types in each lane

[0109]

[0110] Vehicle speed expectation and standard deviation: Vehicle speed expectations and standard deviations for different axle numbers in the slow lane, fast lane, and overtaking lane

[0111] S106, Headway calculation: Headway back-calculation: Calculate the headway by back-calculating from vehicle speed and time interval, and calculate the rear overhang and front overhang in combination with the "Motor Vehicle Operation Safety Technical Conditions" (GB 7258-2017).

[0112] Log-normal distribution fitting: Fit the headway under different service level conditions to a log-normal distribution, and the specific parameters are shown in Table 3.

[0113] Table 3 Headway distribution parameters

[0114]

[0115]

[0116] S106, Vehicle weight and axle weight distribution fitting:

[0117] Two-axle and three-axle vehicles: Fit the vehicle weight data using a log-normal distribution, and the specific parameters are as follows:

[0118] Two-axle vehicle: Expectation 19.8t, standard deviation 0.27t.

[0119] Three-axle vehicle: Expectation 28.5t, standard deviation 0.29t.

[0120] Four-axle, five-axle, and six-axle vehicles: Fit the vehicle weight data using a multi-modal mixture Gaussian distribution, and the specific parameters are shown in Table 4:

[0121] Table 4 Vehicle weight distribution parameters for four-, five-, and six-axle vehicles

[0122]

[0123] S107, Statistic vehicle length and wheelbase, and input the vehicle length (see Table 5) and wheelbase distribution coefficient (see Table 6) of each vehicle type based on the fixed mean.

[0124] Table 5 Statistical vehicle lengths of each vehicle type

[0125]

[0126] Table 6 Wheelbase calculation coefficients of each vehicle type

[0127]

[0128] S2, Construct a departure model for high-speed mixed traffic flow, which specifically includes the following steps:

[0129] S201, Monte Carlo random sampling

[0130] Input parameters: Use the distribution parameters such as vehicle speed, vehicle weight, and wheelbase obtained from fitting in Step 1 as inputs. These parameters include the expected values and standard deviations of vehicle speeds for different axle numbers in each lane (Table 2), vehicle weight distribution parameters (Table 4), and wheelbase distribution coefficients (Table 6), etc.

[0131] Randomly generate vehicle sequences: As Figure 5 shown, use the Monte Carlo method to randomly generate vehicle sequences based on the input distribution parameters. By random sampling, simulate the departure process of social vehicles. Specifically, for each vehicle, according to the distribution of parameters such as vehicle speed, vehicle weight, and wheelbase in each lane, use a random number generator to generate corresponding values. For example, the vehicle speed parameter follows a normal distribution. Use the random number generator of the normal distribution, input the corresponding expected value and standard deviation, and generate the vehicle speed value; for the vehicle weight parameter, use the lognormal distribution for two-axle and three-axle vehicles. Use the random number generator of the lognormal distribution, input the corresponding expected value and standard deviation, and generate the vehicle weight value. For four-axle, five-axle, and six-axle vehicles, use the multi-modal mixed Gaussian distribution. Use the random number generator of the multi-modal mixed Gaussian distribution, input the corresponding distribution parameters, and generate the vehicle weight value.

[0132] S202. Embed HHT vehicle parameters

[0133] Determine HHT vehicle parameters: Based on the HHT waybill data, determine the wheelbase, axle weight, and total vehicle and cargo weight of the tractor and trailer of the HHT vehicle. The tractor of the HHT vehicle does not participate in sharing the weight of the transported goods, and its wheelbase and axle weight data are directly obtained from the waybill. The axle weight of the trailer mainly concentrates in the range of 10 - 18t, and the distance between the axles of each trailer module mainly distributes in the range of 1.2 - 1.8m (mostly 1.55m and 1.22m). The total vehicle and cargo weight mainly concentrates in the range of 95t - 155t.

[0134] Incorporate into the traffic flow sequence: Incorporate the HHT vehicle into the traffic flow sequence to make it mix with social vehicles. In the model, the HHT vehicle is a special vehicle type with specific parameters and behavior rules. For example, the parameters such as the vehicle length, wheelbase, and vehicle weight of the HHT vehicle are determined according to the actual waybill information, which is relatively independent of the random sampling process of other social vehicles.

[0135] Restrict lane-changing behavior: According to the "HHT Management Regulations", restrict the lane-changing behavior of the HHT vehicle. In the model, specifically mark the HHT vehicle and separately restrict the lane-changing behavior of the HHT vehicle in the model according to the regulations, so that its lane-changing behavior when crossing the bridge is strictly restricted, thus making it more in line with the actual scenario and ensuring the accuracy of the mixed traffic flow simulation.

[0136] S3. Build an improved cellular automaton (CA) traffic flow model, which specifically includes the following steps:

[0137] S301. Refine the cell length to 0.01 meters, which significantly improves the continuity of vehicle positions, making the simulation

[0138] results closer to the actual situation. The refined cell length can more accurately describe the position changes of vehicles on the bridge, thereby improving the accuracy of load effect calculations.

[0139] S302. Dynamic acceleration rule. Introduce the assumption of constant acceleration and update the vehicle position in real time through the motion equation. This method can more accurately simulate the acceleration and deceleration processes of vehicles, avoiding the accumulation of position errors caused by discontinuous acceleration changes in traditional models. The dynamic acceleration rule makes the motion state of vehicles more in line with the actual situation and improves the reliability of simulation results.

[0140] Based on the improvement of the traditional NS model speed change rule, with the change of acceleration as the initial value, and then accurately determine the vehicle position change at each moment. The improved action rules are as follows:

[0141] ① Model acceleration rule: s i (t + Δt) = min{s i (t) + a n Δt, S max}} & 0 < a n < a max

[0142] ② Model deceleration rule: s s i (t + Δt) = min(s i (t) + a n Δt, S min ) & a min < a n < 0

[0143] ③ Vehicle position determination:

[0144] In the above rules, w i , s i , a i are the cell position, vehicle speed, and acceleration of vehicle i, v max and v min are the upper and lower limits of the driving speed set according to vehicle performance and road section regulations, a min and a max are the upper and lower boundaries of the acceleration value considering vehicle performance, and Δt is the time step, usually 1 second.

[0145] S303. Refine the lane-changing behavior. Based on the lane width and lane-changing angle (4°±1°), divide the states of "lane change incomplete" and "lane change completed", and set the lane-changing probability between lanes (e.g., the lane-changing probability from the overtaking lane is 0.5). This refined lane-changing behavior modeling can more accurately simulate the lane-changing process of vehicles in multi-lane scenarios, improving the accuracy of traffic flow simulation. By considering the details of lane-changing behavior, the model can better reflect the complexity of the actual traffic flow.

[0146] S4. Calculate the bridge load effect and verify and optimize the model, which specifically includes the following steps:

[0147] S401. Calculate the load effect by the influence line loading method, and the specific steps are as follows:

[0148] Influence line generation: Based on the bridge structure (such as a 20m simply supported beam), calculate the influence lines of bending moment and shear force. The influence line of bending moment is used to represent the bending moment distribution generated by vehicle loads at the mid-span of the bridge, while the influence line of shear force is used to represent the shear force distribution generated by vehicle loads at the supports of the bridge. These influence lines are the basis for subsequent load effect calculations and can accurately reflect the influence of vehicle loads on the bridge structure.

[0149] Load effect calculation: In the cellular automaton model, record the axle weight position of each vehicle in real time, and calculate the bending moment and shear force effect values at each time step according to the influence lines of bending moment and shear force. Specifically, by tracking the position change of vehicle axles on the bridge and combining with the influence lines, dynamically calculate the bending moment and shear force effects generated by vehicle loads on the bridge structure, so as to realize the real-time monitoring and analysis of the bridge load effect.

[0150] S402. Extreme value extraction:

[0151] During the cellular automaton iteration process, record the maximum values of the mid-span bending moment and the support shear force every day. Taking the 61-day scenario simulated by the model as an example, calculate the extreme values of the load effect in the scenarios of the HHT vehicle traveling alone and mixed with random traffic flow. Taking the bending moment as an example, compare the differences in the load effect between the scenarios of the HHT vehicle traveling alone and the mixed traffic flow, as Figure 6 shown.

[0152] S403. Verify the accuracy of the model, which specifically includes the following steps:

[0153] Traffic flow simulation verification: Compare the simulation with the measured vehicle weight, vehicle speed, and headway error (Table 7-9). The average values and 90% percentile values of the vehicle weight simulated for each vehicle type have little difference compared with the measured values, and the maximum errors are 3.91% and 2.83% respectively.

[0154] Table 7 Vehicle weight simulation test

[0155]

[0156] The comparative analysis of the speed simulation results and the measured data by vehicle type is shown in Table 8. According to the comparison results in the table, it can be seen that the error between the speed simulation results and the measured values is not large, and the absolute value of the error of each type of index is less than 1.00%.

[0157] Table 8 Vehicle speed simulation test

[0158]

[0159] The comparative analysis of the simulated mean value and the measured mean value of the time headway by vehicle type is shown in Table 9. According to the comparison results in the table, it can be seen that the error between the time headway simulation results and the measured values is not large, and the absolute value of the error of each type of index is less than 1.00%.

[0160] Table 9 Time headway verification

[0161]

[0162] Accuracy verification of load effect:

[0163] Verify the extreme values of the load effect under different cell lengths (Table 10) through finite element software (such as Midas Civil), and select 0.01m as the optimal parameter;

[0164] Table 10 Extreme values of load effect

[0165]

[0166] Analyze the data of the bending moment and shear force measured and simulated respectively. It is found that the calculation results using the load effect model based on CA are slightly higher than the measured data as a whole. The median error is less than 3%, and the relative errors at the 25% and 75% quantiles are both less than 4.5%. Analyze the maximum values of the 61 load effect calculations in four groups respectively, and the maximum error is 3.6%.

[0167] Mixed scenario verification:

[0168] Among the maximum daily load effect values in the 61-day simulation calculation, the data of 47 days exceed the calculated value under the scenario of the single passage of HHT vehicles, and the maximum value is 1.36 times that of the load effect of the single passage of HHT vehicles.

Claims

1. A simulation method for load effects of mixed traffic flow in large-piece transportation based on cellular automata, characterized in that, It includes the following steps: S1. Collect and preprocess the dynamic weighing data of expressways, including: S101. Real-time collect the vehicle driving parameters through the dynamic weighing system; S102. Eliminate abnormal data and classify vehicle types; S103. Classify vehicle types based on axle type combinations and count the lane distribution; S104. Analyze the spatio-temporal distribution characteristics of traffic volume; S105. Fit the normal distribution parameters of vehicle speed; S106. Calculate the headway and fit the distribution, and fit the distribution parameters of vehicle weight; S107. Statistic the vehicle length and wheelbase parameters; S2. Construct a departure model for high-speed mixed traffic flow: S201. Randomly generate a sequence of social vehicles based on the Monte Carlo method; S202. Embed HHT vehicle parameters and restrict their lane-changing behaviors; S3. Construct an improved cellular automaton traffic flow model: S301. Refine the cell length to 0.01 m; S302. Update the vehicle positions using the dynamic acceleration rule; S303. Establish a lane-changing probability model for each lane; S4. Calculate and verify the bridge load effect: S401. Calculate the load effect in real time based on the influence line loading method; S402. Extract the extreme values of the mid-span bending moment and the support shear force; S403. Verify the model accuracy through finite element method.

2. The method according to claim 1, characterized in that, The specific steps of S102 include: Set the vehicle speed threshold range [60, 120] km / h and eliminate the over-limit data; Classify vehicle types with two to six axles according to the number of axles; Calculate the parameters of the rear overhang and the front overhang according to GB7258-2017.

3. The method according to claim 1, characterized in that In the step S103: Use the multi-peak mixed Gaussian distribution to fit the vehicle weight of vehicles with four or more axles; When counting the proportion of vehicle types in each lane, 78% of two-axle vehicles are reserved in the slow lane, and 12% of six-axle vehicles are reserved in the overtaking lane; Conduct traffic volume equivalent conversion according to the "Automobile Design" standard.

4. The method according to claim 1, wherein In the step S201: The input parameters include the vehicle speed expectations μ∈[70.05, 93.94] km / h and the standard deviations σ∈[7.63, 25.93] for each lane; The vehicle weight distribution parameters include the log-normal distribution of two-axle vehicle pairs (μ = 19.8 t, σ = 0.27); Use the Latin hypercube sampling to generate the vehicle sequence.

5. The method according to claim 1, characterized in that, In the step S202: The HHT vehicle wheelbase adopts a modular distribution of 1.2 - 1.8 m; The total vehicle and cargo weight is controlled in the range of 95 - 155 tons; Set the lane-changing prohibition mark according to the "HHT Management Regulations".

6. The method according to claim 1, wherein The dynamic acceleration rule in the step S302 includes: The acceleration range is set as a ∈ [-3, 2] m / s 2 ; Position update formula: The time step Δt = 1 second.

7. The method according to claim 1, wherein The lane-changing rules in the step S303 include: The lane-changing probability of the overtaking lane is set to 0.5; The two-way lane-changing probabilities of the fast lane are 0.56 (left) and 0.14 (right) respectively; The lane-changing probability of the slow lane is set to 0.7; The lane-changing angle is controlled in the range of 3° - 5°.

8. The method according to claim 1, characterized in that, The verification criteria in the step S403 include: The simulation error of vehicle weight ≤%3.91; The simulation error of vehicle speed ≤%0.95; The error of headway ≤%0.92; The error of the extreme value of load effect ≤%3.

6.

9. The method according to claim 1, wherein It also includes: Establish a comparison mechanism for the load effects between the mixed traffic scenario and the single-lane traffic scenario; Set the optimization parameter of the cell length to 0.01 m.

Images