An intelligent multimodal transport path planning method based on improved simulated annealing algorithm

Abstract

The present invention discloses an intelligent multimodal transport path planning method based on an improved simulated annealing algorithm, which belongs to the field of multimodal transport path planning. The method comprises the following steps: using a neural network model to predict travel time according to individual information and congestion information, constructing a road directed graph, determining an initial feasible path, an initial annealing temperature of the algorithm, a minimum annealing temperature, and the number of algorithm iterations; changing the initial feasible path as the current path to generate a new path, judging whether the path change complies with a taboo rule, and searching for a new feasible path if not; if so, detecting whether the new path is better than the current path, accepting the new path if so, and bringing the new path into the next iteration as the current path; otherwise, probabilistically accepting or discarding the new path, cooling the path, and entering the next iteration; changing the current path according to the new iteration result, judging whether to accept the new path, and then continuing the iteration until the temperature is lower than the minimum annealing temperature, and then terminating the iteration. The current path at the end is the current optimal path.

Description

Technical Field

[0001] The present invention belongs to the field of multimodal transport path planning, and in particular relates to an intelligent multimodal transport path planning method based on an improved simulated annealing algorithm. Background Art

[0002] Multimodal transport is one of the modes of transportation that the country has vigorously promoted in recent years. By organically combining two or more modes of transportation, it increases transportation speed and reduces operating costs, thereby achieving the goal of optimizing the allocation of transportation resources.

[0003] Optimizing transport speed and cost in intermodal transport requires rational route planning, dynamic calculation of transport time and cost, and real-time adjustment of transport plans. Traditional route optimization algorithms only consider the connectivity between different nodes, but fail to account for the multiple modes of transport between them. Specifically, two nodes may have more than one mode of transport, each with its own unique time and cost. Therefore, neural networks are needed to account for the impact of various transport mode (vehicle) characteristics on travel time when designing the algorithm. Furthermore, in real-world situations, due to the multiple modes of transport involved in intermodal transport, additional time and cost are incurred during transfers due to operations such as loading and unloading. Furthermore, factors such as weather and holidays can cause changes in the original transport speed and cost. Therefore, the algorithm must replan the transport plan based on the latest transport speed and cost before executing the transfer to ensure the scientific and reasonable transfer. In practice, due to the large number of nodes, the diversity of transport modes, and the influence of transfer connections, traditional simulated annealing algorithms suffer from long computation times and slow convergence rates, making them unable to promptly and effectively generate optimal transport plans. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the present invention aims to provide an intelligent multimodal transport path planning method based on an improved simulated annealing algorithm.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] An intelligent multimodal transport path planning method based on an improved simulated annealing algorithm comprises the following steps:

[0007] S1, obtain real travel time, individual information and congestion information;

[0008] S2, based on the actual travel time, individual information and congestion information, and using the Wide & Deep model to build a travel time prediction model;

[0009] S3, inputs the individual information data provided by the logistics company and the congestion information data provided by the transportation department into the travel time prediction model, predicts the travel time of each road section, and obtains the transportation cost of each road section and the transshipment cost of each city based on the quotation of the logistics company;

[0010] S4, using the travel time of each road section, the transportation cost of each road section and the transfer cost of each city obtained in S3 to construct a road directed graph, and initialize the algorithm parameters, including: initial annealing temperature, minimum annealing temperature, number of algorithm iterations, temperature drop rate, maximum taboo length, and taboo table;

[0011] S5, according to the generation strategy of the initial feasible path, determine the initial feasible path and use it as the current path;

[0012] S6, arbitrarily select two nodes in the current path as modification nodes;

[0013] S7: Determine whether the modified node selection complies with the taboo rule. If not, proceed to S6 to reduce the taboo length. If yes, randomly find a new path between the two nodes and calculate the objective function value of the new path and the current path.

[0014] S8, judging whether to accept the new path based on the new path acceptance policy; if yes, proceed to S9; otherwise, do not change the current path and proceed to S10;

[0015] S9, the new path is taken as the current path into the subsequent iteration;

[0016] S10, determine whether the current temperature reaches the minimum annealing temperature given in S4, if yes, go to S12, otherwise go to S11;

[0017] S11, judging whether the number of iterations reaches the number of algorithm iterations given in S4, if so, reducing the current temperature according to the temperature drop rate given in S4 and proceeding to S6;

[0018] S12, ending the iteration, the current path is the current optimal path, and is used as the driving path;

[0019] S13: During the driving process, if there is a transfer behavior at the next node, enter S3 and recalculate the optimal path. If the next node is the end point, the driving process is completed.

[0020] Furthermore, the individual information in S1 includes: load capacity, driving age, mileage, and driving record; the congestion information in S1 includes: road grade, driving date, weather conditions, and historical traffic volume.

[0021] Furthermore, the steps of constructing a travel time prediction model include:

[0022] S21, the feature vector b of the congestion information obtained in S1 is b=[b1,b2,…,b d1 ] is input into the Wide model, and the mathematical formula is as follows:

[0023]

[0024] is the Wide model coefficient, φ(b) is the cross-feature of congestion information; c ki It is a Boolean variable, which is 1 if feature i belongs to the cross feature d, otherwise it is 0;

[0025] S22, input the feature vector u of the individual information into the Deep model. The Deep model is a recurrent neural network model. The hidden layer calculation formula is as follows:

[0026] a l+1 =σ(W l a l +β l )

[0027] o=tanh(W 2 a 2 +β 2 )

[0028] l∈[0,1,2] is the number of hidden layers, is the activation function, o is the output layer, a l ,β l ,W l are the activation function, bias value and weight of the hidden layer l respectively;

[0029] S23, merge the Wide model and the Deep model to establish the real travel time e i -s i The minimum loss function of the prediction time f(·) forms a logistic regression problem:

[0030]

[0031]

[0032] Where, t i For road section p i The ideal driving time, is the Deep model coefficient, and β is the prediction error.

[0033] Furthermore, in S5, the step of generating the initial feasible path specifically includes:

[0034] S51, s is the starting point, t is the end point, x ij is a feasible direct point, Ii is the set of feasible direct points in layer i, y i is the access point;

[0035] S52, starting from point s, use point s as the retrieval point y i , search for all feasible direct points x ij (can be reached directly, distance and time are both feasible values, i = 1...m represents the number of search layers, j = 1...n represents the node order), random and non-repeated (guaranteed to traverse all x ij ) Select a feasible direct point x ik ∈I i , as the next retrieval point y i+1 ;

[0036] Repeat this step until y i+1 = t, then a feasible path starting from point s and ending at point t is obtained; or, if all feasible direct points are searched, the output is that there is currently no feasible path starting from point s and ending at point t.

[0037] Furthermore, in S7, the method for finding a new path is to bring the two nodes into the feasible path generation strategy in S5.

[0038] Furthermore, the steps for solving the objective function values ​​of the new path and the current path are:

[0039] 1) Obtain the transportation cost of each node on the path based on the road directed graph constructed by S4 Transshipment cost X = {x1, x2, x3,} and transportation time m∈Road, Rail, Air};

[0040] 2) The specific objective function is as follows:

[0041]

[0042]

[0043] Where min W is the objective function, a q is the transshipment decision variable, m q For the mode of transport, w u and w c are the time weight coefficient and the cost weight coefficient respectively.

[0044] Furthermore, in S8, the acceptance strategy of the new path is: if the objective function value of the new path is better than the current path, the new path is accepted; otherwise, the probability function value i is calculated according to the current temperature and the difference between the objective function values, and compared with the randomly generated random number i′∈(0, 1). If i>i′, the new path is accepted, otherwise it is not accepted.

[0045] Furthermore, the solution formula for the probability function value i is:

[0046]

[0047] Δ=f2-f1

[0048] Where f2 is the objective function value of the new path, f1 is the objective function value of the current path, and T is the initial annealing temperature given by S4.

[0049] A computer storage medium stores a readable program, and when the program is run, the above method is executed.

[0050] A device comprising: one or more processors, a memory for storing one or more programs;

[0051] When the one or more programs are executed by the one or more processors, the one or more processors are caused to perform the above method.

[0052] Beneficial effects of the present invention:

[0053] 1. Starting from the underlying data collection, targeting the complexity and real-time nature of real-world traffic conditions, a neural network prediction model is established to ensure the effectiveness and accuracy of the traffic data constructed on the road directed graph;

[0054] 2. This invention takes into account the additional costs incurred by the transfer process. Before arriving at the transfer node, the current planned path is calibrated according to the current traffic data to ensure the scientificity and rationality of the transfer task;

[0055] 3. The present invention uses tabu search to improve the convergence speed of the heuristic algorithm, so that the path planning task can be completed in a short time. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0057] Figure 1 is a flow chart of multimodal transport route planning of the present invention;

[0058] Figure 2 It is the flow chart of the Wide & Deep model of the present invention;

[0059] Figure 3 It is a flow chart of the strategy for generating the initial feasible path of the present invention. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0061] like Figure 1 As shown, an intelligent multimodal transport path planning method based on an improved simulated annealing algorithm includes the following steps:

[0062] S1, obtain individual information and congestion information;

[0063] Get historical travel x based on vehicle positioning i On road section p i Departure time i and arrival time e i , from which the real travel time e can be obtained i -s i , i∈N, N represents the number of trip samples;

[0064] Obtain vehicle and driver information based on registration information, including load capacity, driver age, mileage, driving record, etc. as individual information Represents individual information as a d1-dimensional feature vector; the road section p i Road (waterway) grade, travel date (whether it is a holiday), weather conditions (whether it is bad) and historical traffic volume as congestion information The congestion information is represented by a d2-dimensional feature vector.

[0065] S2, using the Wide & Deep model to build a travel time prediction model;

[0066] like Figure 2 As shown in the figure, the Wide&Deep model includes the Wide model and the Deep model. The Wide model is a generalized linear model, and the Deep model is a multi-layer perceptron.

[0067] The steps to build a travel time prediction model are:

[0068] S21, the feature vector of the congestion information obtained in S1 Input to the Wide model, the mathematical formula is as follows:

[0069]

[0070] is the Wide model coefficient, φ(b) is the cross-feature of congestion information; c ki It is a Boolean variable, which is 1 if feature i belongs to the cross feature d, otherwise it is 0;

[0071] S22, input the feature vector u of the individual information into the Deep model. The Deep model is a recurrent neural network model. The hidden layer calculation formula is as follows:

[0072] a l+1 =σ(W l a l +β l

[0073] o=tanh(W 2 a 2 +β 2 )

[0074] l∈[0, 1, 2] is the number of hidden layers, is ReLU (activation function), o is the output layer, a l ,β l ,W l are the activation function, bias value and weight of the hidden layer l respectively;

[0075] S23, merge the Wide model and the Deep model to establish the real travel time e i -s i The minimum loss function of the prediction time f(·) forms a logistic regression problem:

[0076]

[0077]

[0078] Where, t i For road section p i The ideal driving time (e.g. a road section with a total length of 120 km and a speed limit of 120 km / h, then t = 1 hour), is the Deep model coefficient, and β is the prediction error; thus, the travel time prediction model is obtained, and the travel time is then predicted.

[0079] S3, input the individual information data provided by the logistics company and the congestion information data released by the railway, shipping, and aviation departments into the travel time prediction model to predict the travel time of each section; according to the quotation of the logistics company, the transportation price (cost) of each section and the transshipment price (cost) of each city are obtained.

[0080] S4, using the time and cost data obtained in S3 to construct a road directed graph, and initialize the algorithm parameters: including the initial annealing temperature, the minimum annealing temperature, the number of algorithm iterations at each temperature, the temperature drop rate, the maximum taboo length, and the taboo table;

[0081] Use traffic data (i.e., time and cost data obtained from S3) to construct a directed road graph. This is done by creating a transport matrix with the same number of transport modes as the number of modes. The rows and columns of the matrix represent the nodes corresponding to the map. Whether the nodes are connected, the mode of transport, the length of the path, and the cost are represented by the values ​​of the corresponding rows and columns in the matrix. If the value of the transport matrix corresponding to two nodes is positive infinity, it means that the two nodes are not connected under this mode of transport.

[0082] It is worth pointing out that due to the dynamic nature of the actual situation, the connectivity, distance and transportation costs between nodes are also changing in real time. Therefore, the algorithm should be based on the modifiability of traffic data, so that various data can change with the actual situation in the actual transportation process to adapt to the ever-changing transportation costs, cross-node transportation needs, artificial road opening and other actual needs, so as to make the algorithm more universal.

[0083] S5, according to the generation strategy of the initial feasible path, determine the initial feasible path and use it as the current path;

[0084] like Figure 3 As shown in Figure 2, the steps for generating the initial feasible path specifically include:

[0085] S51, s is the starting point, t is the end point, x ij is a feasible direct point, I i is the set of feasible direct points in layer i, y i is the access point;

[0086] S52, starting from point s, use point s as the retrieval point y i , search for all feasible direct points x ij (can be reached directly, distance and time are both feasible values, i = 1...m represents the number of search layers, j = 1...n represents the node order), random and non-repeated (guaranteed to traverse all x ij ) Select a feasible direct point x ik ∈I i , as the next retrieval point y i+1 ;

[0087] Repeat this step until y i+1 = t, then a feasible path starting from point s and ending at point t is obtained; or, if all feasible direct points are searched, the output is that there is currently no feasible path starting from point s and ending at point t.

[0088] S6, arbitrarily select two nodes in the current path as modification nodes.

[0089] S7: Determine whether the modified node selection complies with the taboo rule. If not, proceed to S6 to reduce the taboo length. If yes, randomly find a new path between the two nodes and calculate the objective function value of the new path and the current path.

[0090] Among them, the taboo rule means: every time a new solution (path) is adopted, the modified node will be added to the taboo table given by S4, and subsequent modifications will be restricted. The taboo length is the maximum taboo length given by S4. If the node already exists in the taboo table when the modification is selected later, the corresponding taboo length will be reduced and no modification will be made (not in compliance with the taboo rule); when the taboo length is 0, it will be deleted from the taboo table and the modification restriction will be lifted.

[0091] Randomly select two nodes from the current path and find a path between the two nodes that is different from the original path, that is, find a new path. If there is no other path between the two nodes except the original path, change two other nodes and continue trying until a path that meets the requirements is found. The specific method for finding a new path is to bring the two nodes into the feasible path generation strategy in S5 to find a new path.

[0092] Take the path {y1,y2,y3,y4} as an example. According to the road directed graph constructed by S4, the transportation cost of each node on the path is obtained. Transshipment cost X = {x1, x2, x3,} and transportation time m∈{highway, railway, aviation}, the specific objective function is as follows:

[0093]

[0094]

[0095] Where min W is the objective function, a q is the transshipment decision variable, m q For the mode of transport, w u and w c are the time weight coefficient and the cost weight coefficient respectively

[0096] S8, judging whether to accept the new path based on the new path acceptance policy; if yes, proceed to S9; otherwise, do not change the current path and proceed to S10;

[0097] The acceptance strategy for the new path is as follows: if the objective function value of the new path is better than that of the current path, the new path is accepted; otherwise, the probability function value i is calculated based on the current temperature and the difference between the objective function values, and compared with the randomly generated random number i′∈(0,1). If i>i′, the new path is accepted, otherwise it is not accepted.

[0098] The basic form of the probability function is:

[0099]

[0100] Δ=f2-f1

[0101] f2 is the objective function value of the new path, f1 is the objective function value of the current path, and T is the initial annealing temperature given by S4. The objective function value of the new path is calculated by searching all transportation modes in a depth-first order to find the optimal transportation mode for the current node sequence.

[0102] S9, the new path is taken as the current path into the subsequent iteration and enters S10.

[0103] S10, judging whether the current temperature reaches the minimum annealing temperature given in S4, if yes, proceed to S12, otherwise proceed to S11.

[0104] S11, judging whether the number of iterations reaches the number of algorithm iterations given in S4, if so, reducing the current temperature according to the temperature reduction rate given in S4, regardless of whether it reaches the number of iterations, proceeding to S6.

[0105] S12, ending the iteration, the current path is the current optimal path, which is used as the driving path.

[0106] S13: During the driving process, if there is a transfer behavior at the next node, enter S3 and recalculate the optimal path to ensure the scientificity and rationality of the transfer behavior; if the next node is the end point, the driving process is completed.

[0107] It is worth pointing out that due to the particularity of the actual situation, the size of the objective function value is uncertain. If the same initial temperature and end temperature are used in each path planning, the algorithm may converge too quickly or too slowly. For example, when the difference in the objective function value is 1×10 2 If the initial and final temperatures are too high, the algorithm will find it difficult to accept the difference solution at the beginning, and will fall into the local optimal solution. Therefore, in the actual loop, the temperature should be appropriately adjusted to be on the same order of magnitude as the objective function value to achieve the best convergence effect.

[0108] Furthermore, due to the unique nature of multimodal transport, the objective function calculation method included in this invention requires the addition of a transshipment cost to the target value once a mode of transport changes. This cost is the time and cost incurred by loading, unloading, and handling operations during the transshipment process. This transshipment cost varies for each city and each mode of transport and is updated in real time to meet actual needs.

[0109] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal translation scripting language JavaScript, etc.

[0110] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including the instruction device, which implements the function specified in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0112] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0113] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. An intelligent multimodal transport path planning method based on an improved simulated annealing algorithm, characterized in that: The following steps are involved: S1, obtain real travel time, individual information and congestion information; S2, based on the actual travel time, individual information and congestion information, and using the Wide & Deep model to build a travel time prediction model; S3, inputs the individual information data provided by the logistics company and the congestion information data provided by the transportation department into the travel time prediction model, predicts the travel time of each road section, and obtains the transportation cost of each road section and the transshipment cost of each city based on the quotation of the logistics company; S4, using the travel time of each road section, the transportation cost of each road section and the transfer cost of each city obtained in S3 to construct a road directed graph, and initialize the algorithm parameters, including: initial annealing temperature, minimum annealing temperature, number of algorithm iterations, temperature drop rate, maximum taboo length, and taboo table; S5, according to the generation strategy of the initial feasible path, determine the initial feasible path and use it as the current path; S6, arbitrarily select two nodes in the current path as modification nodes; S7: Determine whether the modified node selection complies with the taboo rule. If not, proceed to S6 to reduce the taboo length. If yes, randomly find a new path between the two nodes and calculate the objective function value of the new path and the current path. S8, judging whether to accept the new path based on the new path acceptance policy; if yes, proceed to S9; otherwise, do not change the current path and proceed to S10; S9, the new path is taken as the current path into the subsequent iteration; S10, determine whether the current temperature reaches the minimum annealing temperature given in S4, if yes, go to S12, otherwise go to S11; S11, judging whether the number of iterations reaches the number of algorithm iterations given in S4, if so, reducing the current temperature according to the temperature drop rate given in S4 and proceeding to S6; S12, ending the iteration, the current path is the current optimal path, and is used as the driving path; S13: During the driving process, if there is a transfer behavior at the next node, then go to S3 and recalculate the optimal path. If the next node is the end point, the driving process is completed; The steps to building a drive time prediction model include: S21, the feature vector of the congestion information obtained in S1 Input to the Wide model, the mathematical formula is as follows: is the Wide model coefficient, φ(b) is the cross-feature of congestion information; c ki is a Boolean variable, k=1,…d1, which is 1 if feature i is part of the kth transformation and 0 otherwise; S22, input the feature vector u of the individual information into the Deep model. The Deep model is a recurrent neural network model. The hidden layer calculation formula is as follows: a l+1 =σ(W l a l +b l ) o=tanh(W 2 a 2 +β 2 ) l∈[0,1,2] is the number of hidden layers, is the activation function, o is the output layer, a l ,β l ,W l are the activation function, bias value and weight of the hidden layer l respectively; S23, merge the Wide model and the Deep model to establish the real travel time e i -s i The minimum loss function of the prediction time f(·) forms a logistic regression problem: Where, t i For road section p i The ideal driving time, is the Deep model coefficient, and β is the prediction error.

2. The intelligent multimodal transport path planning method based on the improved simulated annealing algorithm according to claim 1 is characterized in that: The individual information in S1 includes: load capacity, driving age, mileage, and driving record; the congestion information in S1 includes: road grade, driving date, weather conditions, and historical traffic volume.

3. The intelligent multimodal transport path planning method based on the improved simulated annealing algorithm according to claim 1 is characterized in that: In S5, the steps for generating the initial feasible path specifically include: S51, s is the starting point, t is the end point, x ij is a feasible direct point, I i is the set of feasible direct points in layer i, y i is the access point; S52, starting from point s, use point s as the retrieval point y i , search for all feasible direct points x ij , that is, the distance and time are both feasible values, i = 1…m represents the number of search layers, j = 1…n represents the node order, which is random and non-repeated, ensuring that all x are traversed ij , select a feasible direct point x ik ∈I i , as the next retrieval point y i+1 ; Repeat this step until y i+1 = t, then a feasible path starting from point s and ending at point t is obtained; or, if all feasible direct points are searched, the output is that there is currently no feasible path starting from point s and ending at point t.

4. The intelligent multimodal transport path planning method based on the improved simulated annealing algorithm according to claim 3 is characterized in that: In S7, the method for finding a new path is to bring the two nodes into the generation strategy of the feasible path in S5.

5. The intelligent multimodal transport path planning method based on the improved simulated annealing algorithm according to claim 4 is characterized in that: The steps for solving the objective function values ​​of the new path and the current path are: 1) Obtain the transportation cost of each node on the path based on the road directed graph constructed by S4 Transshipment cost X = {x1, x2, x3} and transportation time m∈{road, railway, aviation}; 2) The specific objective function is as follows: Where min W is the objective function, a q is the transshipment decision variable, m q For the mode of transport, w u and w c are the time weight coefficient and the cost weight coefficient respectively.

6. The intelligent multimodal transport path planning method based on the improved simulated annealing algorithm according to claim 1 is characterized in that: In S8, the acceptance strategy of the new path is: if the new path objective function value is better than the current path, then the new path is accepted; otherwise, the probability function value z is calculated based on the current temperature and the difference between the objective function value and the randomly generated random number i ' ∈(0,1) for comparison, if z>i ' , then the new path is accepted, otherwise it is not accepted.

7. The intelligent multimodal transport path planning method based on the improved simulated annealing algorithm according to claim 6 is characterized in that: The solution formula for the probability function value z is: Δ=f2-f1 Where f2 is the objective function value of the new path, f1 is the objective function value of the current path, and T is the initial annealing temperature given by S4.

8. A computer storage medium storing a readable program, characterized in that: When the program is run, the method according to any one of claims 1 to 7 is executed.

9. A device, characterized in that include: one or more processors, a memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors are caused to perform the method according to any one of claims 1 to 7.

Images