Road network traffic flow distribution method considering influence of road-occupying construction area

Abstract

The invention discloses a road network traffic flow distribution method considering influence of a road-occupying construction area, which belongs to the technical field of traffic flow distribution and comprises the following steps of: determining basic information of a construction area road network; selecting an observation road section through a step-by-step point distribution method; investigating the traffic volume, the traffic capacity and the free driving time of the observed road section; performing distribution by using an increment distribution method according to the traffic volume of the observed road section to obtain a traffic volume distribution matrix; forming a new distribution matrix # imgabs0 # by using row data of an observation road section corresponding to the traffic distribution matrix, and obtaining a backstepping OD result by using the road section flow, the distribution matrix # imgabs1 # and a maximum entropy OD backstepping model; and based on a backstepping OD result, traffic flow distribution is carried out on the OD point by using a random user balance model of a path length correction Logit, and a traffic volume value of each path after distribution is obtained. According to the construction area road network, the traffic condition of the construction road section in the peak period can be relieved after distribution, and the vehicle passing time of the whole road network can be effectively shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic flow distribution, and in particular to a road network traffic flow distribution method taking into account the impact of road construction areas. Background Art

[0002] In recent years, the construction of urban transportation infrastructure has been accelerating, and the construction, reconstruction and expansion of transportation infrastructure will occupy limited road space. During the construction of the project, a construction area will be set up around the road, which will occupy the driving space of the road. The complex road conditions of the construction road are prone to induce traffic accidents, and it is easy to cause traffic congestion during peak hours in the morning and evening, forming a bottleneck section of the urban road. According to relevant investigations and studies, the construction area not only affects the smoothness and safety of road traffic, but also affects the travel of surrounding residents. Therefore, adopting reasonable management methods and means, giving full play to the efficiency of the road network, and regulating and guiding traffic flow will play a significant role in alleviating the contradiction between traffic supply and demand and alleviating local traffic pressure.

[0003] Traffic flow allocation is an important method to alleviate traffic congestion. Traditional traffic flow allocation is mainly for the allocation of road networks in non-construction areas, while road sections with road occupation are more likely to cause road congestion. Therefore, it is necessary to provide a road network traffic flow allocation method that takes into account the impact of road occupation areas. Summary of the invention

[0004] The technical solution of the present invention aims at the technical problem that the existing technical solutions are too single, and provides a solution that is significantly different from the existing technologies. It mainly provides a road network traffic flow distribution method that takes into account the impact of road construction areas, so as to solve the technical problem that the traditional traffic flow method proposed in the above background technology cannot alleviate the traffic pressure on the road construction section.

[0005] The technical solution adopted by the present invention to solve the above technical problems is:

[0006] A road network traffic flow allocation method considering the impact of road occupation construction areas includes the following steps:

[0007] S1. Determine the number of main nodes and sections in the road network including the construction area section and basic information of the section; determine the valid OD points and the valid path information between the OD points;

[0008] S2, based on the information data in step S1, selecting the observation section by stepwise point distribution method;

[0009] S3, investigating the traffic volume, traffic capacity and free travel time of the observation section selected in step S2;

[0010] S4, determining the initial demand of each OD point according to the traffic volume of the observed road section in step S3, and using the incremental allocation method to sequentially allocate the effective paths between the OD points to obtain a traffic volume allocation matrix;

[0011] S5, take the row data of the traffic distribution matrix corresponding to the observed section in step S4 to form a new distribution matrix

[0012] S6. Using the road flow and allocation matrix And the maximum entropy OD inverse model obtains the unknown parameter values, and the inverse OD results are obtained according to the parameters;

[0013] S7. Based on the inverse OD result of step S6, the random user equilibrium model of path length modified Logit is used to allocate traffic flow to the OD points to obtain the traffic volume value of each path after allocation.

[0014] Further, step S2 includes:

[0015] S2-1, all OD point demands are allocated to the road network using the shortest path method;

[0016] S2-2, find the road section where the OD point passes most times and use it as the observation section;

[0017] S2-3, delete all OD points passing through the observed section, and continue to distribute the remaining OD points to the road network using the shortest path method;

[0018] S2-4. Repeat S2-2 and S2-3 until all OD points are removed.

[0019] Furthermore, the allocation process in step S4 includes:

[0020] S4-1, initialize the road section impedance and divide each OD demand into a fixed number of parts;

[0021] S4-2, calculate the impedance of the road section and allocate a portion of the OD demand to the shortest path;

[0022] S4-3, determine whether all the requirements between the ODs have been allocated. If the allocation between the ODs has not been completed, transfer to the next requirement between the ODs and return to step S4-2; if the allocation between the ODs has been completed, use formula 2 to calculate the column value of the allocation matrix and transfer to S4-4;

[0023] p j =m i / T j (2)

[0024] Among them, P jrepresents the set of values ​​in the jth column of the allocation matrix, j = 1, 2, ..., N, N is the number of OD points; m i represents the traffic volume allocated to each road section by the OD point, i = 1, 2,,…,M, where M is the number of road sections; T j is the traffic demand of the jth OD point;

[0025] S4-4, determine whether all OD points have been allocated. If not, proceed to the next OD point and return to step S4-2; if all OD points have been allocated, finally output the allocation matrix.

[0026] Further, in step S6, first use the allocation matrix The OD of the maximum entropy model is inversely calculated based on the road section flow, and then processed by the Lagrange multiplier method. Finally, the particle swarm algorithm is used to solve the model and obtain the OD matrix.

[0027] Further, in step S6, the maximum entropy OD inverse model expression is:

[0028]

[0029]

[0030]

[0031] Among them, Formula 7 is the objective function, Formula 8 and Formula 9 are constraints; maxE represents the matrix W(T ij ) is maximized; a represents the road section name; i and j represent the community names respectively; T ij It represents the traffic volume from community i to community j, which is the volume of OD reverse demand; Indicates T ij The sharing ratio on road section a is obtained through traffic distribution. The rows of the matrix represent the number of road sections and the columns represent the number of OD points. a Represents the traffic volume on road section a.

[0032] Furthermore, the processing by Lagrange multiplier method includes:

[0033] S6-1. Use the Lagrange multiplier method to process formula 7 and obtain the Lagrange function L as follows:

[0034]

[0035] In the formula, λ a represents the Lagrange multiplier;

[0036] S6-2, T in the above formula ij Find the partial derivative, and we get

[0037]

[0038] S6-3, after solving and sorting, we get

[0039]

[0040] S6-4, let T = exp(-λ0), and substitute the above equation 12 into the objective function and the constraint conditions respectively, and we can get the following two equations:

[0041]

[0042] Equation 13 is a nonlinear equation system containing m+1 variables and m+1 equations.

[0043] Furthermore, the method of solving the model using the particle swarm algorithm comprises the following steps:

[0044] S6-5. Determination of the objective function: The objective function is transformed into the formula 14:

[0045]

[0046] According to formula 14, the closer the objective function is to 0, the closer the values ​​on both sides of the nonlinear equation system 13 are, and the corresponding parameters λ0,λ1,…,λ m The closer the value is to the actual situation;

[0047] S6-6, obtaining the traffic volume and allocation matrix of the road section;

[0048] S6-7, randomly initialize the particle swarm;

[0049] S6-8, calculating the particle fitness value;

[0050] S6-9. Update the optimal vector P of individual particles i and the global optimal vector p g ;

[0051] S6-10, update particle speed and position;

[0052] S6-11, termination criteria: Repeat steps S6-8 to S6-10, set the number of iterations according to the convergence curve, and determine the parameters λ0, λ1, …, λ under the minimum y value m The value of and output;

[0053] S6-12, Calculate the OD matrix: The obtained parameters λ0,λ1,…,λ m Substitute the allocation matrix into equation 12 to obtain the OD matrix.

[0054] Furthermore, in step S7, the expression of the random user equilibrium model of the path length corrected Logit is:

[0055]

[0056]

[0057]

[0058]

[0059] Among them, formula 17 is the minimization objective function; formula 18 is the relationship between path flow and OD demand; formula 19 is the non-negative constraint of path flow; formula 20 is the relationship between path flow and section flow; represents the flow rate on the kth path with the starting point r and the end point s; t α () represents the actual impedance of section a; q rs It represents the traffic demand of OD point with starting point r and ending point s; The variable that represents the relationship between the road segment and the path is a 0-1 variable. If the road segment α belongs to the kth path of the OD point of the departure point r and the destination point s, then otherwise x a represents the traffic volume of road section a; represents the distance of path k between OD points rs; γ represents the path parameter; θ represents the positive dispersion coefficient.

[0060] Furthermore, in step S7, a random user equilibrium allocation model of path length corrected Logit is solved based on a genetic algorithm, wherein the length of each section of the road network, the set of all paths between each OD point, and the free travel time and traffic capacity of each section are used as relevant parameters of the objective function.

[0061] Further, in step S7, the solution step is:

[0062] S7-1, determine the objective function: substitute the road section traffic volume calculation formula into the objective function formula 17, taking the path traffic volume as a variable; obtain a new objective function;

[0063] S7-2, determine the relevant parameters of the objective function;

[0064] S7-3, coding chromosome;

[0065] S7-4, generate initial population;

[0066] S7-5, calculate fitness;

[0067] S7-6, selection, crossover and mutation operations;

[0068] S7-7, termination criteria;

[0069] S7-8, output the traffic volume of each path: Output the traffic volume of each path after allocation, use formula 20 to calculate the traffic volume of each section, and calculate the section travel time based on the section traffic volume and combined with the BPR function. The BPR function is:

[0070] t / t0=1+0.3915(Q / C) 1.1515 (31)

[0071] Among them, t is the road section travel time; t0 is the road section free travel time (road travel time without the influence of other vehicles); Q is the road section traffic volume; C is the road section traffic capacity.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] (1) The present invention proposes a 0D inverse model without a priori matrix. Based on the original 0D matrix obtained by 0D inverse, a random user equilibrium allocation model based on path length modified Logit is used to conduct allocation research. Experiments show that for the construction area road network, the allocation can alleviate the traffic conditions during the peak hours of the construction section and effectively reduce the vehicle travel time of the entire road network.

[0074] (2) The present invention is different from the existing OD matrix inversion method (all of which require a priori OD matrix). It proposes to select observation sections based on the step-by-step point distribution method. The selected observation sections can ensure that the section coverage information is wide and the investigation workload is moderate. Then, the travel OD matrix of the construction area road network is inverted based on the traffic volume of the observation section. That is, the present invention does not require a priori matrix and is more convenient to operate.

[0075] (3) The present invention is different from the traditional allocation method that uses the Logit model to determine the probability of path selection (which only considers the impact of travel time on path selection). On the basis of travel time, the impact of travel distance (i.e., path length) on travelers is also considered, which is more in line with the actual situation and is conducive to improving the rationality of allocation.

[0076] (4) The present invention uses the BPR function proposed by the US Highway Administration in traffic flow distribution and recalibrates the parameters to determine the road travel time, taking into account the impact of road traffic volume on the travel time, making it more in line with the conditions of Chinese roads.

[0077] The present invention will be explained in detail below with reference to the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 A diagram showing a construction road network in an embodiment of the present invention;

[0079] Figure 2 A diagram of a network topology structure in an embodiment of the present invention;

[0080] Figure 3 It is a statistical diagram of hourly traffic volume data during evening peak hours on a road section in an embodiment of the present invention;

[0081] Figure 4 It is a statistical diagram of the traffic capacity of each road section in the embodiment of the present invention;

[0082] Figure 5 A statistical diagram of free travel time of each road section in an embodiment of the present invention;

[0083] Figure 6 A traffic volume statistics diagram of each path after road network allocation in an embodiment of the present invention;

[0084] Figure 7 Traffic volume statistics diagram before and after allocation of each road section in an embodiment of the present invention;

[0085] Figure 8 It is a statistical diagram of average impedance before and after allocation of each road section in an embodiment of the present invention;

[0086] Fig. 9 A statistical diagram of the total travel time before and after the allocation of each road section in an embodiment of the present invention;

[0087] Fig.10 A diagram of the steps of the allocation method of the present invention. DETAILED DESCRIPTION

[0088] To facilitate understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings, but the present invention can be implemented in different forms and is not limited to the embodiments described in the text. On the contrary, these embodiments are provided to make the content disclosed in the present invention more thorough and comprehensive.

[0089] The model-related contents in the present invention include:

[0090] 1. Basic principles of OD matrix inversion

[0091] Traditional OD surveys usually require a lot of manpower and material resources. In order to reduce the disadvantage of large preliminary surveys in traffic planning, researchers began to use road section traffic volume to reverse the OD matrix. Traffic flow allocation is to allocate road section traffic volume through the OD matrix. Therefore, OD matrix reverse calculation is the inverse process of traffic flow allocation. The OD matrix reverse calculation method is to reverse the current travel OD matrix of the traffic partition based on the observed current road network traffic flow. The expression of the OD reverse calculation matrix is ​​shown in Formula 1.

[0092]

[0093] Where a represents the road section name; i and j represent the community names; T ij It represents the traffic volume from community i to community j, which is the volume of OD reverse demand; Indicates T ij The sharing ratio on road section a can be obtained through traffic allocation. The rows of the matrix represent the number of road sections and the columns represent the number of OD points. a It represents the traffic volume on road section a, which can be obtained through actual investigation.

[0094] Formula 1 is the relationship between OD demand and road section traffic volume. This formula has N (N-1) unknowns and M equations. When N (N-1)>M, the equation group has infinite solutions. However, the number of OD points N (N-1) in the actual road network is usually greater than the number of road sections M. Therefore, some constraints need to be added to determine the optimal solution.

[0095] 2. The modeling steps of the OD inverse model without a priori matrix are as follows:

[0096] A1: First, arbitrarily generate an OD matrix T0. According to the traffic volume information of the road section, certain constraints can be set to make the randomly generated OD matrix conform to the actual situation;

[0097] A2: Use the incremental allocation method in the unbalanced allocation method to allocate the initial OD matrix, thereby obtaining the allocation matrix P0;

[0098] A3: Use the allocation matrix P0 and the road section detection flow to perform OD inverse inference of the maximum entropy model, use the particle swarm algorithm to solve the model, and obtain the OD matrix T1. Matrix T1 is the OD matrix obtained by inverse inference.

[0099] According to the above process, the OD inversion without a priori matrix is ​​mainly divided into two parts: ① randomly generate an OD matrix, and then use the incremental allocation method to allocate the allocation matrix; ② perform maximum entropy OD inversion based on the allocation matrix and the road section detection flow to obtain the OD inversion matrix.

[0100] 3. Multi-OD point allocation model based on incremental allocation method

[0101] The incremental allocation method is to divide the OD demand into several parts, and each part of the OD demand is cyclically allocated to the shortest path of the road network. After each allocation, the impedance of each section needs to be updated, and then the shortest path is calculated based on the new impedance. The OD demand is allocated to the shortest path in turn, and finally all OD demands are allocated.

[0102] The allocation matrix is ​​an important condition for OD inversion. The rows of the allocation matrix represent road sections and the columns represent OD points. The incremental allocation method calculates the allocation matrix using Equation 2 for each OD point allocated. The value of each column element.

[0103] p j =m i / T j (2)

[0104] Among them, P j represents the set of values ​​in the jth column of the allocation matrix (j = 1, 2, ..., N, N is the number of OD points), m i represents the traffic volume allocated to each road section by the OD point (i=1,2,,…,M, M is the number of road sections), T j is the traffic demand of the jth OD point. j The combination is the allocation matrix.

[0105] 4. Maximum entropy OD inversion model based on particle swarm algorithm

[0106] (1) Establishment of the maximum entropy OD inversion model

[0107] The maximum entropy model assumes that the travel of vehicles is random. If each OD point T ij The trip is regarded as a random event, and the total number of events is

[0108]

[0109] According to the principle of permutation and combination, the matrix [T ij ] is

[0110]

[0111] In the formula, W(T ij ) represents the matrix generated by formula 4.

[0112] The idea of ​​maximum entropy is to find the ij ) is the largest matrix, that is, it is believed that such a matrix has the greatest possibility of appearing. For the convenience of solving, we take the logarithm of both sides of the above formula and get

[0113]

[0114] According to the Stirling approximation formula In(x!) = x·In(x)-x, we can get

[0115]

[0116] The present invention is to make the matrix W(T ij ) is maximized, let maxE≈InW(T ij). Formula 1 is used as the constraint condition of the objective function, that is, the new OD inverse model is shown in Formula 7, and Formulas 8 and 9 are both constraints.

[0117]

[0118]

[0119]

[0120] It is difficult to directly solve the above objective function, and it is usually necessary to transform the above objective function.

[0121] (2) Processing based on the Lagrange multiplier method model

[0122] Using the Lagrange multiplier method to process the above objective function (7), the Lagrange function L is obtained as follows:

[0123]

[0124] In the formula, λ a represents the Lagrange multiplier.

[0125] For T in the above formula ij Find the partial derivative, and we get

[0126]

[0127] After solving and sorting, we get

[0128]

[0129] Let T = exp(-λ0), and substitute the above equation 12 into the objective function and the constraint conditions respectively, and we can get the following two equations:

[0130]

[0131] According to observation, Equation 13 is a nonlinear system of equations containing m+1 variables and m+1 equations. By solving the above system of equations, we can obtain the unknowns λ0,λ1,…,λ m The value of; then substitute the solution result into formula 12, and the OD inverse matrix can be obtained.

[0132] (5) Maximum entropy OD inversion model based on particle swarm algorithm

[0133] 1) Advantages of seeking maximum entropy OD inverse model based on particle swarm algorithm

[0134] The advantages of the particle swarm algorithm are mainly: ① It does not require complex mathematical calculations, but only needs to determine the objective function and constraints, that is, the algorithm can be used to search for the optimal parameter value; ② The calculation accuracy of the algorithm does not need to depend on the initial value λ0; ③ The calculation accuracy of the particle swarm algorithm for solving the model is higher, usually higher than the solution accuracy of the Newton iteration method.

[0135] 2) Solution steps of the maximum entropy OD inverse model based on particle swarm algorithm

[0136] ① Determination of the objective function: The particle swarm algorithm cannot directly solve the nonlinear equations, and needs to preprocess Equation 13. It is known that the solution of the equation is to find the unknown parameters λ0, λ1, …, λ when the function on the left side of the equation is equal to the function value on the right side. m The value of , that is, the objective function becomes as shown in formula 14:

[0137]

[0138] According to formula (14), the closer the objective function is to 0, the closer the values ​​on both sides of the nonlinear equation system (13) are, and the corresponding parameters λ0, λ1, …, λ m The value is closer to the actual situation.

[0139] ② Obtaining the traffic volume and allocation matrix of the road section: Obtain the traffic volume of the road section through actual survey, and obtain the allocation matrix using the traffic flow allocation method

[0140] ③ Randomly initialize particle swarm: Random initial population, the position of each particle is x i =(x i1 , x i2 , x i3 , …, x iD ), the velocity of each particle is v i =(v i1 , v i2 , v i3 , …, v iD ), where i = 1, ..., n, and the parameters to be determined are λ0, λ1, ..., λ m , that is, the dimension D of the particle is m+1. When initializing the population, n m+1-dimensional vectors are generated as the initial position and initial velocity of each particle.

[0141] ④ Calculation of particle fitness value: The purpose is to find the parameters λ0, λ1, ..., λ under the minimum y value of formula 14 m The value of , that is, using formula 10 as the fitness function.

[0142] ⑤ Optimal vector P of individual particle i and the global optimal vector p gUpdate: Individual optimal vector P i and the global optimal vector p g The update is as follows: particle i will generate an x ​​after each iteration i , x i Substituting it into the fitness function will get a fitness value, and the one with the smallest fitness among all the positions that particle i has passed through is selected as the individual optimal vector p i , p i =(p i1 , p i2 ,…p iD ), i = 1, ..., n; select the one with the smallest fitness from the individual optimal vectors as the global optimal vector p of the particle g , p g =(p g1 , p g2 ,…,p gD ), where p g There is only one.

[0143] ⑥The update of particle velocity and position is as follows:

[0144]

[0145]

[0146] Among them, learning factors c1 and c2 are non-negative constants, usually 2; k is the number of iterations; ω is the inertia weight, which determines how much of a particle's previous speed is maintained. In order to prevent the speed update formula from being greatly affected by the previous speed and falling into the local optimum, ω is taken as 0.7298; j is the variable code, j = 1, 2, ..., D; is the individual optimal value of the jth variable of the i-th particle in the k-th iteration; is the global optimal value of the jth variable of the i-th particle in the k-th iteration; is the current value of the jth variable of the ith particle in the kth iteration; is the value of the jth variable of the ith particle in the k+1th iteration; is the velocity vector of the jth variable of the ith particle in the kth iteration; is the velocity vector of the jth variable of the ith particle in the k+1th iteration; r1 and r2 are random numbers in the interval [0, 1].

[0147] ⑦ Termination criteria: Repeat steps ④-⑥, set the number of iterations according to the convergence curve, and determine the parameters λ0, λ1, …, λ under the minimum y value. m The value of and output.

[0148] ⑧Calculation of OD matrix: The obtained parameters λ0, λ1, ..., λ m Substitute the allocation matrix into equation 12 to obtain the OD matrix.

[0149] 5. Optimized random user equilibrium allocation model

[0150] (1) Path length corrected Logit path selection model

[0151] The traditional deterministic allocation model assumes that road users fully understand the actual situation of the road, but usually road users cannot fully understand the actual situation of the road. At this time, the use of the deterministic allocation model has a certain deviation from the actual situation. In order to make the model more consistent with the actual road conditions, the present invention proposes a random user equilibrium model based on path length corrected Logit. Traditional random allocation uses the Logit model to determine the probability of each path being selected. The main calculation method is shown in Formula 15.

[0152]

[0153] in, represents the utility function of the Logit model, represents the probability of path k being selected between OD points rs; K represents the expected perceived travel cost of path k between OD points rs (usually replaced by path travel time); rs represents the path set between OD points rs, l represents any path between OD points rs; θ represents a positive dispersion coefficient parameter.

[0154] From the model, we can find the path travel time The smaller the path selection probability is, The larger the value, the more it conforms to the actual road conditions. However, the utility function of the traditional random allocation model only considers the impact of road travel time on path selection. In order to solve this problem, the present invention also considers path length as a factor in the utility function and proposes a new path selection probability:

[0155]

[0156] in, represents the distance of path k between OD points rs; γ represents the estimated parameter, which is the importance of path distance in the utility function. From the model, we can find that path distance The smaller the path selection probability is, The larger it is, the more it conforms to the actual road conditions.

[0157] (2) Construction of the random user equilibrium allocation model based on the path length corrected logit

[0158] The present invention constructs a random user balance model based on path length modified Logit as shown in Formula 17:

[0159]

[0160]

[0161]

[0162]

[0163] Among them, formula 17 is the minimization objective function; formula 18 is the relationship between path flow and 0D demand; formula 19 is the non-negative constraint of path flow; formula 20 is the relationship between path flow and section flow; represents the flow rate on the kth path with the starting point r and the end point s; t α () represents the actual impedance of section a; q rs It represents the traffic demand of OD point with starting point r and ending point s; The variable that represents the relationship between the road segment and the path is a 0-1 variable. If the road segment α belongs to the kth path of the OD point of the departure point r and the destination point s, then otherwise x a represents the traffic volume of road section a; represents the distance of path k between OD points rs; γ represents the path parameter; θ represents the positive dispersion coefficient.

[0164] Theorem: The random user equilibrium allocation model based on path length modified Logit is equivalent to the path selection model, that is,

[0165] Proof: The Lagrangian function constructed for equation 17 is shown in equation 21:

[0166]

[0167] Among them, λ rs is the Lagrange multiplier of the constraint condition, and Equation 21 satisfies the Kuhn-Tucker condition, that is, and Processing method, we can get equations 22 and 23:

[0168]

[0169]

[0170] For each valid path k between OD points, You can get

[0171]

[0172] According to equation 24, we can solve As shown below:

[0173]

[0174] By summing all the paths between OD points in equation 24, we can get

[0175]

[0176] According to Formula 25 and Formula 26, the path selection probability of path-corrected Logit can be obtained as shown below.

[0177]

[0178] Formula 17 and the above-mentioned proof result Formula 27 prove that the random user equilibrium flow allocation and path selection model based on path length modified Logit are equivalent.

[0179] (3) Solving the random user equilibrium allocation model based on genetic algorithm

[0180] 1) Determination of the objective function: Substitute the road section traffic volume calculation formula (Formula 20) into the objective function Formula 17, and use the path traffic volume as a variable. The changed objective function is shown in Formula 28:

[0181]

[0182] 2) Determination of relevant parameters of the objective function: Determine the length of each section of the road network, the set of all paths between each OD point, and the free travel time and traffic capacity of each section.

[0183] 3) Chromosome encoding: The variables of the present invention are the traffic volume of each path. In order to more conveniently perform operations such as selection, crossover and mutation, the present invention adopts binary encoding. The encoding accuracy τ of each number is calculated by formula 29,

[0184]

[0185] in, Indicates the encoding length of a binary number; σ max , σ min The decimal number range that represents the undetermined parameter. The decimal value represented by the encoding in the genetic algorithm can be obtained by converting the binary number into a decimal number and multiplying it by the encoding accuracy of the number.

[0186] 4) Generation of initial population: N initial populations are randomly generated, and the population size is generally between ω and 2ω.

[0187] 5) Calculation of fitness: Calculate the fitness of each individual in the group, which is usually the objective function of the genetic algorithm. The fitness function is the main basis for the future "survival of the fittest". The greater the fitness of an individual, the greater the probability that the individual will be inherited to the next generation. Conversely, the smaller the fitness of an individual, the smaller the probability that the individual will be inherited to the next generation. The purpose of the present invention is to find the traffic volume of each path corresponding to the minimum objective function. The objective function is shown in Formula 28.

[0188] 6) Selection, crossover and mutation operations

[0189] ① Selection: Generally, individuals with higher fitness will have more chances to be inherited to the next generation. In order to establish the relationship between individual fitness and individual survival probability, the present invention uses the Monte Carlo method to determine the survival probability of each individual. The Monte Carlo method is the link to establish the relationship between individual fitness and individual survival probability. The calculation formula is shown in Formula 30.

[0190]

[0191] Among them, F j represents the survival probability of the jth individual, f j Represents the fitness function value of the jth individual, and N is the number of individuals in the population.

[0192] ② Crossover: Crossover refers to the operation of replacing and recombining part of the structure of two parent individuals to generate new individuals. The crossover operator plays a core role in the genetic algorithm. In order to reduce the time complexity of the crossover part, single-point crossover is used, and the crossover probability is usually 0.4 to 0.99. The specific process of crossover is: crossover operation is performed on any two adjacent individuals. First, it is determined whether to crossover based on the probability. If not, the two individuals do not change; if crossover is performed, then the crossover point is selected and the codes after the crossover point are exchanged.

[0193] ③ Mutation: Mutation is to change the gene values ​​on certain loci of the individual strings in the population in order to maintain the diversity of the population individuals and prevent falling into the local optimum. The mutation probability pm is usually taken as 0.0001~0.1. The specific process of mutation is: select the mutant individuals according to the probability, and select the mutation point on the selected individuals. The original gene becomes 0 if it is 1, and the original gene becomes 1 if it is 0.

[0194] 7) Termination criteria: Repeatedly perform fitness calculation, selection, crossover and mutation operations until a satisfactory solution is obtained or the preset number of iterations is reached, and the path traffic volume value under the minimum objective function value is output.

[0195] 8) Output of path traffic volume: Output the traffic volume of each path after allocation, use formula 20 to calculate the traffic volume of each section, calculate the section travel time based on the section traffic volume and combined with the BPR function, and compare the traffic volume and travel time before and after allocation to verify the effectiveness of the traffic flow allocation of the present invention.

[0196] In order to make the BPR function conform to the actual road conditions, the parameters of the traditional BPR function are calibrated to obtain a new BPR function as shown below:

[0197] t / t0=1+0.3915(Q / C) 1.1515 (31)

[0198] Among them, t is the road section travel time; t0 is the road section free travel time (road travel time without the influence of other vehicles); Q is the road section traffic volume; C is the road section traffic capacity.

[0199] Please refer to the attached Fig.10 , a road network traffic flow allocation method considering the impact of road occupation construction area, comprising the following steps:

[0200] S1. Determine the number of main nodes and sections in the road network including the construction area section and basic information of the section; determine the valid OD points and the valid path information between the OD points;

[0201] S2, based on the information data in step S1, selecting the observation section by stepwise point distribution method;

[0202] The principle of OD matrix inversion is to use the observed traffic volume of the road section to invert the OD matrix. Generally, the more road section information is used, the more accurate the OD matrix will be. However, this does not conform to the original intention of OD inversion to be simple and convenient. In order to solve this problem, the present invention proposes to select the observation road section based on the step-by-step point distribution method. The main steps are as follows:

[0203] S2-1, all OD point demands are allocated to the road network using the shortest path method;

[0204] S2-2, find the road section where the OD point passes most times and use it as the observation section;

[0205] S2-3, delete all OD points passing through the observed section, and continue to distribute the remaining OD points to the road network using the shortest path method;

[0206] S2-4. Repeat S2-2 and S2-3 until all OD points are removed.

[0207] The observation road sections obtained by the above processing method can ensure that the road sections cover a wide range of information and that the investigation workload is moderate.

[0208] S3, investigating the traffic volume, capacity and free travel time of the observation section selected in step S2; and determining the section travel time according to the section impedance function in OD back-calculation and traffic flow distribution;

[0209] S4, determining the initial demand of each OD point according to the traffic volume of the observed road section in step S3, and using the incremental allocation method to sequentially allocate the effective paths between the OD points to obtain a traffic volume allocation matrix;

[0210] The allocation process in step S4 includes:

[0211] S4-1, initialize the road section impedance and divide each OD demand into a fixed number of parts;

[0212] S4-2, calculate the impedance of the road section and allocate a portion of the OD demand to the shortest path;

[0213] S4-3, determine whether all the demands between the ODs have been allocated. If the allocation between the ODs has not been completed, transfer to the next demand between the ODs and return to step S4-2; if the allocation between the ODs has been completed, use the above formula 2 to calculate the column value of the allocation matrix and transfer to S4-4;

[0214] S4-4, determine whether all OD points have been allocated. If not, proceed to the next OD point and return to step S4-2; if all OD points have been allocated, finally output the allocation matrix.

[0215] S5, take the row data of the traffic distribution matrix corresponding to the observed section in step S4 to form a new distribution matrix

[0216] S6. Using the road flow and allocation matrix And the maximum entropy OD inverse model obtains the unknown parameter values, and the inverse OD results are obtained according to the parameters;

[0217] Specifically: first use the allocation matrix The OD of the maximum entropy model is inversely calculated based on the road section flow, and then processed by the Lagrange multiplier method. Finally, the particle swarm algorithm is used to solve the model and obtain the OD matrix.

[0218] S7. Based on the inverse OD result of step S6, the random user equilibrium model of path length corrected Logit is used to allocate traffic flow to the OD points.

[0219] Embodiment: This embodiment uses the road network formed by the construction section of Huai'an Expressway to study the traffic conditions within the road network.

[0220] 1. Selection of road network including construction area sections

[0221] Construction road network Figure 1 As shown, Figure 1 In the figure, the construction areas include Huanghe Road, Ninglian Road, Xi'an Road and Yan'an Road. According to on-site investigations, among these construction sections, only Yan'an East Road (between Huaihai South Road and Chengde South Road) has traffic capacity. Other construction sections are either completely closed or partially closed. The partially closed roads are in poor condition and have basically lost their traffic capacity. Blue lines represent fully closed sections, and red lines represent partially closed sections. This embodiment does not consider sections that have lost their traffic capacity, and mainly focuses on the traffic flow distribution of the road network including the construction area. The network consists of 29 main nodes and 37 sections. Only one-way traffic is considered. Green represents main roads and yellow represents secondary roads. The basic section information is shown in Table 1, and the network topology is shown in Table 1. Figure 2 It covers the traffic volume data of 37 road sections during the night peak hours, as shown in Figure 3 shown.

[0222] Table 1 Information table of each road section in the road network

[0223]

[0224]

[0225] according to Figure 2 Combined with the actual road network situation, we can know that nodes 1, 2, 3, 4, and 5 are nodes on the Yellow River Road, the road is completely closed, and there are no sections connecting the construction points; nodes 6 and 11 are nodes on Xi'an Road, the road is semi-closed, and there are basically no motor vehicles; nodes 18 and 25 are nodes on Ninglian Highway, the road is completely closed, and there are no sections connecting the construction points; nodes 26, 27, 28, and 29 are nodes on Yan'an Road, except for nodes 27 and 28, other nodes are not connected. Taking nodes 1, 2, 3, 4, 5, 6, and 11 as starting points and nodes 18, 25, 26, 27, 28, and 29 as end points, only considering one direction, there are 32 pairs of valid OD points, and a total of 152 valid path information between each OD point. According to Figure 3 The traffic volume of each road section before allocation can be known.

[0226] The nine sections to be investigated using the stepwise point distribution method are Section 1, Section 12, Section 19, Section 21, Section 22, Section 23, Section 32, Section 36 and Section 37. The evening peak hour traffic volume of these nine sections is shown in Table 2. The traffic capacity and free travel time of each section are shown in Table 2. Figure 4 and Figure 5 shown.

[0227] Table 2 Traffic volume of road section survey

[0228]

[0229] First, the initial demand of 32 OD points is set to 100 pcu / h, and the incremental allocation method is used to allocate the 152 valid paths between the OD points in sequence to obtain the traffic volume allocation result T ij , i and j represent the row and column of the allocation result respectively. The allocation matrix is ​​calculated based on the traffic volume allocation result and formula (2). The allocation matrix 1, 12, 19, 21, 22, 23, 32, 36 and 37 obtained by the incremental allocation method constitute a new allocation matrix Using observed road flow and allocation matrix The maximum entropy OD inverse model obtains the unknown parameter values ​​as shown in Table 3, and the inverse OD results obtained according to the parameters are shown in Table 4.

[0230] Table 3 Undetermined parameter values

[0231]

[0232] Table 4 OD back-calculation results

[0233] Serial number OD point Round the result of reverse calculation Serial number OD point Round the result of reverse calculation 1 1-18 62 17 4-25 98 2 1-25 85 18 4-29 129 3 1-26 231 19 5-18 74 4 1-27 80 20 5-25 73 5 1-28 154 21 6-18 74 6 1-29 109 22 6-25 177 7 2-18 74 23 6-26 274 8 2-25 132 24 6-27 95 9 2-27 95 25 6-28 148 10 2-28 138 26 6-29 129 11 2-29 129 27 11-18 101 12 3-18 74 28 11-25 153 13 3-25 98 29 11-26 375 14 3-28 735 30 11-27 144 15 3-29 129 31 11-28 210 16 4-18 74 32 11-29 176

[0234] The random user equilibrium model of path length modified Logit is used to allocate traffic flow to the above 32 OD points. In the Logit model, the parameters θ and γ are 2 and 1, respectively. The traffic usage values ​​of each path after allocation are as follows: Figure 6 As shown, the traffic volume of each road section is calculated using Formula 20, and the travel time of the road section is calculated based on the traffic volume of the road section and combined with the BPR function (Formula 31). The traffic volume and travel time before and after the allocation are compared to verify the effectiveness of the traffic flow allocation of the present invention. The traffic volume, average impedance and total travel time of vehicles on each road section before and after the allocation are shown in Figure 31. Figure 7 , Figure 8 , Fig. 9 shown.

[0235] according to Figure 7 It can be seen that the traffic volume of each section has an obvious change relationship before and after the allocation, but the total traffic volume of each section does not change much. Among them, the traffic volume of sections 3, 4, 6, 7, 8, 9, 11, 12, 13, 15, 18, 19, 20, 22, 24, 25, 26, 27, 30, and 37 shows an obvious downward relationship before and after the allocation; the traffic volume of sections 1, 10, 21, 28, 29, 32, and 36 does not change much; the traffic volume of other sections shows an obvious upward relationship. In general, the change law of traffic volume after allocation is mainly affected by the comprehensive influence of free flow impedance. The smaller the free flow impedance and the larger the traffic capacity, the more traffic can be allocated to the section. If the free flow impedance is larger, the traffic of the section will be allocated to other sections.

[0236] according to Figure 8It can be seen that the overall average impedance of each road section before and after allocation does not change much. This is because the traffic volume of each road section in Huai'an City is less than the road section capacity. That is, when the road traffic volume is less than the road section capacity, the change in traffic volume has little effect on the travel time value. At the same time, according to formula (31), the present invention only considers the impact of road section traffic volume on travel time, that is, the final Figure 8 The average impedance change before and after distribution Figure 7 That is, under the premise of the same road section, the greater the traffic volume, the longer the travel time, and vice versa, the smaller the traffic volume, the shorter the travel time, which is closer to the free flow travel time.

[0237] Fig. 9 The total travel time of the considered section is obtained by multiplying the traffic volume of the section by the average impedance of the section. The research goal of this invention is to reduce the travel time of the entire road network. Fig. 9 It can be seen that the relationship between before and after allocation is that the total travel time of some sections is long before allocation, and the total travel time of some sections is long after allocation. However, the total travel time of sections before allocation is more significantly reduced in sections 9, 13, 15, and 18, that is, the traffic flow allocation greatly alleviates the congestion of these sections. Finally, the total travel time of the entire road network is added up. The total travel time before allocation is 4050327.517s, and the total travel time after allocation is added up to 3478967.635s. The travel time of the entire road network is reduced by 14.11%, that is, the traffic flow allocation method optimized by the present invention can alleviate the traffic conditions in the evening rush hour of the construction section.

[0238] The above is an exemplary description of the present invention in combination with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as such non-substantial improvements are made by adopting the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A method for allocating traffic flow in a road network considering the impact of road construction areas, characterized by: The steps include: S1. Determine the number of main nodes and sections in the road network including the construction area section and basic information of the section; determine the valid OD points and the valid path information between the OD points; S2, based on the information data in step S1, selecting the observation section by stepwise point distribution method; S3, investigating the traffic volume, traffic capacity and free travel time of the observation section selected in step S2; S4, determining the initial demand of each OD point according to the traffic volume of the observed road section in step S3, and using the incremental allocation method to sequentially allocate the effective paths between the OD points to obtain a traffic volume allocation matrix; S5, take the row data of the traffic distribution matrix corresponding to the observed section in step S4 to form a new distribution matrix S6. Using the road flow and allocation matrix And the maximum entropy OD inverse model obtains the unknown parameter values, and the inverse OD results are obtained according to the parameters; S7. Based on the inverse OD result of step S6, the random user equilibrium model of path length modified Logit is used to allocate traffic flow to the OD points to obtain the traffic volume value of each path after allocation.

2. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 1, characterized in that: Step S2 includes: S2-1, all demands between OD points are allocated to the road network using the shortest path method; S2-2, find the road section where the OD point passes most times and use it as the observation section; S2-3, delete all OD points passing through the observed section, and continue to distribute the remaining OD points to the road network using the shortest path method; S2-4. Repeat S2-2 and S2-3 until all OD points are removed.

3. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 1, characterized in that: The allocation process in step S4 includes: S4-1, initialize the road section impedance and divide each OD demand into a fixed number of parts; S4-2, calculate the impedance of the road section and allocate a portion of the OD demand to the shortest path; S4-3, determine whether all the requirements between the ODs have been allocated. If the allocation between the ODs has not been completed, transfer to the next requirement between the ODs and return to step S4-2; if the allocation between the ODs has been completed, use formula 2 to calculate the column value of the allocation matrix and transfer to S4-4; p j =m i / T j (2) Among them, P j represents the set of values ​​in the jth column of the allocation matrix, j = 1, 2, ..., N, N is the number of OD points; m i represents the traffic volume allocated to each road section by the OD point, i = 1, 2,,…,M, where M is the number of road sections; T j is the traffic demand of the jth OD point; S4-4, determine whether all OD points have been allocated. If not, proceed to the next OD point and return to step S4-2; if all OD points have been allocated, finally output the allocation matrix.

4. The method for allocating road network traffic flow considering the impact of road construction zones according to claim 1, characterized in that: In step S6, first use the allocation matrix The OD of the maximum entropy model is inversely calculated based on the road section flow, and then processed by the Lagrange multiplier method. Finally, the particle swarm algorithm is used to solve the model and obtain the OD matrix.

5. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 4, characterized in that: In step S6, the maximum entropy OD inverse model expression is: Among them, Formula 7 is the objective function, Formula 8 and Formula 9 are constraints; maxE represents the matrix W(T ij ) is maximized; a represents the road section name; i and j represent the community names respectively; T ij It represents the traffic volume from community i to community j, which is the volume of OD reverse demand; Indicates T ij The sharing ratio on road section a is obtained through traffic distribution. The rows of the matrix represent the number of road sections and the columns represent the number of OD points. a Represents the traffic volume on road section a.

6. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 5, characterized in that: The processing by Lagrange multiplier method includes: S6-1. Use the Lagrange multiplier method to process formula 7 and obtain the Lagrange function L as follows: In the formula, λ a represents the Lagrange multiplier; S6-2, T in the above formula ij Find the partial derivative, and we get S6-3, after solving and sorting, we get S6-4, let T = exp(-λ0), and substitute the above equation 12 into the objective function and the constraint conditions respectively, and we can get the following two equations: Equation 13 is a nonlinear equation system containing m+1 variables and m+1 equations.

7. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 6, characterized in that: The method of solving the model using particle swarm algorithm comprises the following steps: S6-5. Determination of the objective function: The objective function is transformed into the formula 14: According to formula 14, the closer the objective function is to 0, the closer the values ​​on both sides of the nonlinear equation system 13 are, and the corresponding parameters λ0,λ1,…,λ m The closer the value is to the actual situation; S6-6, obtaining the traffic volume and allocation matrix of the road section; S6-7, randomly initialize the particle swarm; S6-8, calculating the particle fitness value; S6-9. Update the optimal vector P of individual particles i and the global optimal vector p g ; S6-10, update particle speed and position; S6-11, termination criteria: Repeat steps S6-8 to S6-10, set the number of iterations according to the convergence curve, and determine the parameters λ0, λ1, …, λ under the minimum y value m The value of and output; S6-12, Calculate the OD matrix: The obtained parameters λ0,λ1,…,λ m Substitute the allocation matrix into equation 12 to obtain the OD matrix.

8. The method for allocating road network traffic flow considering the impact of road construction zones according to claim 1, characterized in that: In step S7, the expression of the random user equilibrium model of the path length corrected Logit is: Among them, formula 17 is the minimization objective function; formula 18 is the relationship between path flow and OD demand; formula 19 is the non-negative constraint of path flow; formula 20 is the relationship between path flow and section flow; represents the flow rate on the kth path with the starting point r and the end point s; t α () represents the actual impedance of section a; q rs It represents the traffic demand of OD point with starting point r and ending point s; The variable that represents the relationship between the road segment and the path is a 0-1 variable. If the road segment α belongs to the kth path of the OD point of the departure point r and the destination point s, then otherwise x a represents the traffic volume of road section a; represents the distance of path k between OD points rs; γ represents the path parameter; θ represents the positive dispersion coefficient.

9. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 8, characterized in that: In step S7, a random user equilibrium allocation model of path length modified Logit is solved based on a genetic algorithm, wherein the length of each section of the road network, the set of all paths between each OD point, and the free travel time and traffic capacity of each section are used as relevant parameters of the objective function.

10. A method for allocating road network traffic flow considering the impact of road construction zones according to claim 9, characterized in that: In step S7, the solution steps are: S7-1, determine the objective function: substitute the road section traffic volume calculation formula into the objective function formula 17, taking the path traffic volume as a variable; obtain a new objective function; S7-2, determine the relevant parameters of the objective function; S7-3, coding chromosome; S7-4, generate initial population; S7-5, calculate fitness; S7-6, selection, crossover and mutation operations; S7-7, termination criteria; S7-8, output the traffic volume of each path: output the traffic volume of each path after allocation, use formula 20 to calculate the traffic volume of each section, and calculate the section travel time based on the section traffic volume and combined with the BPR function. The BPR function is: t / t0=1+0.3915(Q / C) 1.1515 (31) Among them, t is the road section travel time; t0 is the road section free travel time (road travel time without the influence of other vehicles); Q is the road section traffic volume; C is the road section traffic capacity.

Images