A vehicle collaborative task offloading method in VEC

Abstract

The present invention belongs to the field of mobile communication technology, and specifically relates to a vehicle collaborative task unloading method in a VEC, comprising: establishing a system model, wherein the system model comprises a main vehicle, a computable vehicle and an incomputable vehicle; establishing a communication model according to the total time required for task unloading from the main vehicle to the computable vehicle; establishing a vehicle selection model based on multiple dynamic factors; establishing a task unloading optimization problem based on constraint conditions based on the communication model and the vehicle selection model; solving the task unloading optimization problem by using a hierarchical analysis method and a KM algorithm to obtain an optimal task unloading situation; the present invention fully considers the dynamic indicator factors of multiple edge nodes, prioritizes the computable vehicles, and adopts the KM algorithm to dynamically match the tasks, thereby effectively reducing the total calculation time of the tasks and reducing the failure rate of task unloading.

Description

Technical Field

[0001] The present invention belongs to the technical field of mobile communications and relates to a vehicle collaborative task unloading method in vehicle edge computing. Background Art

[0002] With the advent of the era of intelligent interconnection of all things, a large number of applications such as autonomous driving, car entertainment, and road management and optimization have emerged, promoting the rapid development of intelligent vehicles. Intelligent vehicles will generate massive amounts of data during operation, but their own computing power and storage resources are limited, and they cannot process this data in real time and efficiently. Especially in situations such as unmanned driving, vehicles have extremely high requirements for response time. Vehicle Edge Computing (VEC), as a distributed computing architecture, can respond to data in real time and quickly, and has the characteristics of low latency, high bandwidth and high reliability. VEC regards vehicles with computing power, storage resources and communication capabilities as edge nodes, and uses the idle resources of surrounding vehicles to provide services for other unloaded vehicles, achieving rapid response and high-speed processing of tasks.

[0003] Task offloading and resource allocation in VEC are key factors affecting vehicle task latency. Research methods for task offloading and resource allocation in VEC include: (1) Research on the joint problem of communication network and computing resource allocation: a multi-objective reinforcement learning strategy is proposed, and the K-nearest neighbor algorithm is used to select vehicle to vehicle (V2V), vehicle to road (V2R) or vehicle to infrastructure (V2I) offloading methods. (2) Research on task allocation offloading in sustainable VEC networks: the hierarchical analysis method is used to prioritize the computing tasks of the vehicle, and the optimal offloading strategy is implemented based on the deep Q learning network with the weighted sum of task processing rates as the optimization target. Most of the tasks are offloaded to the edge server of a single vehicle, roadside unit or base station. As the traffic volume and tasks increase, the distance between the vehicle and the server changes and the limited computing resources of the server will cause high network latency for task offloading, and it is impossible to calculate and transmit data results in real time. The use of collaborative task offloading between vehicles can transmit data results in real time and efficiently, avoiding changes in the network topology in VEC. In the study of vehicle collaborative task offloading: (1) For the task allocation of parked cellular V2 heterogeneous networks in VEC systems, a method for task offloading using parked vehicles as edge nodes is proposed. With the goal of maximizing the energy efficiency of parked vehicles, a reverse auction mechanism is established to encourage parked vehicles to share resources and a many-to-one matching algorithm is used for task offloading. (2) For dynamic edge node selection: a virtual edge architecture is constructed by predicting the duration of inter-vehicle links, and multiple surrounding vehicles are used to enhance the task processing capability in the vehicle network.

[0004] With the rapid development of the intelligent connected era, the dynamic changes in the vehicle network topology on the road and the increasingly intensive task traffic, the above-mentioned existing technologies still fail to fully consider the dynamic indicator factors of various edge nodes. Summary of the invention

[0005] In order to solve the above problems of the prior art, the present invention adopts a vehicle collaborative task unloading method in VEC, comprising the following steps:

[0006] S1. Establishing a system model, wherein the system model includes a main vehicle, a computable vehicle, and a non-computable vehicle;

[0007] S2. Establishing a communication model according to the total time required for task unloading from the host vehicle to the computable vehicle;

[0008] S3. Establish a vehicle selection model based on multiple dynamic factors;

[0009] S4, establishing a task offloading optimization problem based on constraints based on the communication model and the vehicle selection model;

[0010] S5. The task offloading optimization problem is solved by using the analytic hierarchy process and the KM algorithm to obtain the optimal task offloading situation.

[0011] Step S2 of establishing a communication model includes:

[0012] S21. Calculate the transmission rate R from the host vehicle to the computable vehicle h :

[0013]

[0014] S22. Define uninstallation strategy G:

[0015] G=[g m,h ] M×H

[0016] in

[0017] S23, according to the transmission rate R h and offloading strategy G to compute subtask D m Transmission time

[0018]

[0019] S24. According to the transmission time Calculate the total transmission time t of the host vehicle trans :

[0020]

[0021] S25. Calculate subtask D according to offloading strategy G m Calculation time:

[0022]

[0023] S26. According to subtask D m Calculation time Calculation task calculation time:

[0024]

[0025] S27, calculate time t according to task comp and the total transmission time t of the host vehicle trans Calculate the total time required to offload tasks from the host vehicle to the computable vehicle:

[0026]

[0027] Where B is the channel bandwidth from the main vehicle to the computable vehicle, P h The transmission power from the host vehicle to the computable vehicle, d h is the distance from the main vehicle to the hth calculable vehicle, α is the path loss factor, σ 2 is the additional Gaussian white noise power; D m represents the mth subtask, M is the total number of subtasks, H is the total number of computable vehicles, V h represents the hth computable vehicle, f h The computing power of the computable vehicle.

[0028] Step S3 establishes a vehicle selection model including: calculating the selectivity of the computable vehicle, calculating the maximum task that the computable vehicle can perform, calculating the computing power of the computable vehicle and calculating the road overlap between the computable vehicle and the main vehicle; establishing the vehicle selection model according to the selectivity of the computable vehicle, the maximum task that the computable vehicle can perform, the computing power of the computable vehicle and the road overlap between the computable vehicle and the main vehicle.

[0029] The formula for calculating the selectivity of the vehicle in step S31 is:

[0030]

[0031] Among them, r is the communication radius of the main vehicle, d h The distance from the host vehicle to the hth calculable vehicle.

[0032] Step S32 calculates the maximum task that the vehicle can perform, including:

[0033] S321, calculating the communication connection time between the computable vehicle and the host vehicle:

[0034]

[0035] Where S is the relative displacement between the calculable vehicle and the host vehicle, ν is the velocity of the host vehicle, and ν h The speed of the vehicle can be calculated;

[0036] S322: Calculate the maximum number of tasks N that can be performed by the vehicle within the communication connection time. h :

[0037]

[0038] Among them, Q h is the remaining capacity of the hth calculable vehicle, R h The transmission rate from the host vehicle to the computable vehicle, f h The computing power of the computable vehicle.

[0039] The objective function of the task offloading optimization problem based on constraints is:

[0040]

[0041]

[0042]

[0043] C3:M<H

[0044]

[0045] Among them, constraint C1 indicates that when the subtask of the main vehicle is unloaded to the computable vehicle, its transmission time and calculation time The sum cannot exceed the communication connection time t h ; C2 represents subtask D m The tasks unloaded to the hth computable vehicle cannot exceed the limit of its executable tasks; C3 indicates that the number of subtasks of the main vehicle cannot exceed the number of computable vehicles; C4 indicates that each subtask can only be unloaded to one computable vehicle; D(t) is the set of tasks to be unloaded by the main vehicle at time t, V(t) is the set of computable vehicles of the main vehicle at time t, and D m represents the mth subtask, M is the total number of subtasks, H is the total number of computable vehicles, and f h is the computing power of the computable vehicle, N h is the maximum task that can be performed by vehicle h, g m,h For subtask D m Uninstall strategy.

[0046] Step S5 solves the task offloading optimization problem including:

[0047] S51, decomposing the task offloading optimization problem into a vehicle priority sorting problem and an offloading vehicle task allocation problem;

[0048] S52, solving the vehicle priority sorting problem;

[0049] S53. Solve the unloading vehicle task allocation problem according to the result of the vehicle priority sorting problem to obtain the unloading vehicle task allocation result.

[0050] Step S52 solves the vehicle priority sorting problem including:

[0051] S521, calculating the selectability of vehicles according to the vehicle selection model The maximum number of tasks N that can be performed by the vehicle h , the computing power of the vehicle can be calculated hAnd the road overlap Γ h These four dynamic factors establish a hierarchical model;

[0052] S522, set the judgment matrix of the hierarchical model to A=(a ij ) 4×4 , where a ij Indicates the importance comparison value between factor i and factor j;

[0053] S523, using the square root method in the approximate algorithm, respectively calculate the weight value corresponding to each dynamic factor, and normalize the calculated weight value to obtain a weight matrix W=(ω1,ω2,ω3,ω4, where W=(ω1,ω2,ω3,ω4 are the normalized weight values ​​of each dynamic factor respectively;

[0054] S524, perform consistency check on the judgment matrix A. If it passes the consistency check, use the min-max standardization algorithm to calculate the normalized matrix B of each dynamic factor = (β hi ) H×4 ; Otherwise, return to step S522 and adjust the judgment matrix A; where H is the total number of computable vehicles, β hi is the normalized result of the i-th factor of the h-th computable vehicle;

[0055] S525, calculate the priority of the computable vehicle ζ=(ξ hi ) H×4 , where ξ hi =BW T =(β hi ) H×4 (ω1,ω2,ω3,ω4) T , H is the total number of computable vehicles, ξ hi The priority of the i-th factor can be calculated for the h-th vehicle.

[0056] The road overlap degree Γ h Randomly generated from a Poisson distribution.

[0057] Solving the unloading vehicle task allocation problem includes:

[0058] S531, defining a subtask set and an attached vehicle set;

[0059] S532, according to the priority of the computable vehicles, sequentially select computable vehicles with higher priorities to add to the subsidiary vehicle set until the number of computable vehicles in the subsidiary vehicle set is the same as the number of subtasks;

[0060] S533, using a KM algorithm to match the subtasks in the subtask set with computable vehicles in the attached vehicle set, wherein the KM algorithm is a Kuhn-Mankels algorithm;

[0061] S534. If the matching is successful, the optimal unloading vehicle task allocation result is obtained; if the matching fails and there are unselected computable vehicles, return to step S533 and add the unselected computable vehicles to the attached vehicle set; if the matching fails and there are no unselected computable vehicles, the main vehicle needs to wait for the next moment to unload the task.

[0062] Beneficial effects:

[0063] The present invention comprehensively considers various dynamic characteristics such as computing power, road overlap, vehicle selectivity, and the maximum task that can be performed by computable vehicles, constructs a hierarchical model, prioritizes computable vehicles, and can give priority to the optimal computable vehicle set affected by multiple decision factors, effectively reducing the total computing time of the task; the present invention considers the dynamics of the number of collaborative vehicles and the number of tasks, and adopts a quadratic KM algorithm to dynamically match the tasks, effectively reducing the total computing time of the tasks and reducing the failure rate of task offloading. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A VEC vehicle cooperative unloading system model provided by an embodiment of the present invention;

[0065] Figure 2 A flowchart of a vehicle collaborative task unloading method in a VEC provided by an embodiment of the present invention;

[0066] Figure 3 A flowchart of an implementation method of a vehicle collaborative task unloading method in a VEC provided by an embodiment of the present invention;

[0067] Figure 4 A vehicle priority sorting flow chart provided for an embodiment of the present invention;

[0068] Figure 5 A comparison diagram of algorithms for total task computing time provided by an embodiment of the present invention;

[0069] Figure 6 A comparison chart of task offloading failure rates provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0071] The embodiment of the present invention provides a method for unloading vehicle collaborative tasks in a VEC, such as Figure 2 As shown, including:

[0072] S1. Establishing a system model, wherein the system model includes a main vehicle, a computable vehicle, and a non-computable vehicle;

[0073] S2. Establishing a communication model according to the total time required for task unloading from the host vehicle to the computable vehicle;

[0074] S3. Establish a vehicle selection model based on multiple dynamic factors;

[0075] S4, establishing a task offloading optimization problem based on constraints based on the communication model and the vehicle selection model;

[0076] S5. The task offloading optimization problem is solved by using the analytic hierarchy process and the KM algorithm to obtain the optimal task offloading situation.

[0077] like Figure 1 As shown in Figure 1, the system model consists of a main vehicle, calculable vehicles, and uncountable vehicles. In this model, vehicles travel at different speeds. When a main vehicle is to unload a task, only calculable vehicles moving in the same direction within the main vehicle are considered. Assume that the number of calculable vehicles at time t is H, which can be expressed as a set: H(t) = {V h |h=1,2,...H}, where V h is the hth computable vehicle. The task to be unloaded by the main vehicle at time t is represented as D(t), where the task includes M independent subtask blocks, D(t) = {D m |m=1,2,...M},M<H,D m Represents the mth subtask; a computable vehicle can compute at most one subtask at a time.

[0078] The transmission rate from the main vehicle to the hth computable vehicle is:

[0079]

[0080] Where B is the channel bandwidth from the main vehicle to the computable vehicle, P h The transmission power from the host vehicle to the computable vehicle, d his the distance from the main vehicle to the hth calculable vehicle, (x,y) is the coordinate of the main vehicle, (x h ,y h ) is the coordinate of the hth computable vehicle, α is the path loss factor, σ 2 is the additive white Gaussian noise power.

[0081] The main vehicle will perform subtask D m The transfer time and calculation time of unloading to the hth computable vehicle are expressed as: and Since the time it takes to return the result is much shorter than the time it takes to transmit and calculate, the time it takes to return the result is negligible. Define the offloading decision G = [g m,h ] M×H ,in

[0082]

[0083] Therefore, subtask D m The transmission time is:

[0084]

[0085] The total transmission time of the host vehicle is then expressed as:

[0086]

[0087] Subtask D m The calculation time is:

[0088]

[0089] Among them, f h To calculate the computing power of the vehicle, the maximum value of the task computing time is:

[0090]

[0091] In summary, the total time required to offload tasks from the host vehicle to the computable vehicle is:

[0092]

[0093] The host vehicle calculates the communication time with the computable vehicles within its communication range based on the data information shared between vehicles. Assume that the position of the host vehicle is represented by a vector (x, y), and the position of the hth computable vehicle within the range of the host vehicle is (x h ,y h ), the communication radius of the main vehicle is r, the speed of the main vehicle is ν, and the vehicle speed can be calculated as ν h .

[0094] When the relative speed between the hth calculable vehicle and the host vehicle is negative, that is, ν h -ν<0, then the relative displacement of the hth calculable vehicle is: Otherwise, the relative displacement between the hth calculable vehicle and the main vehicle is:

[0095] The driving time of the hth calculable vehicle within the communication range of the host vehicle is:

[0096]

[0097] When the computable vehicle is close to the host vehicle, the communication quality between the two is better and less likely to be interrupted than when the distance is far. Therefore, the selectivity of the computable vehicle is expressed by distance. The closer the distance to the host vehicle, the higher the selectivity. The selectivity is defined as The optionality formula of the hth calculable vehicle is defined as:

[0098]

[0099] The remaining capacity of the hth computable vehicle is defined as Q h , the maximum task N that the vehicle can perform can be calculated within the communication connection time h for:

[0100]

[0101] Based on the analysis of the communication model and the vehicle selection model, the joint optimization problem is expressed as:

[0102]

[0103]

[0104]

[0105] C3:M<H

[0106]

[0107] Constraint C1: h∈V(t) indicates that when the subtask of the main vehicle is offloaded to the computable vehicle, the sum of its transmission time and computation time cannot exceed the connection time; C2: Represents subtask D m The tasks unloaded to the hth computable vehicle cannot exceed the limit of its executable tasks; C3: M < H means that the number of subtasks of the main vehicle cannot exceed the number of computable vehicles; C4: It means that each subtask can only be offloaded to one computable vehicle.

[0108] Since the number of computable vehicles, computing power, and connection time vary over time, the optimization problem is NP-hard. In order to solve this problem, the above problem is decomposed into two sub-problems, namely the vehicle priority sorting problem and the unloading vehicle task allocation problem.

[0109] 1. Vehicle priority sorting problem

[0110] like Figure 4 As shown, the present invention has an optional Computing power h , road overlap Γ h and the maximum task N that the vehicle can perform h The hierarchical model is established based on four factors; the road overlap is randomly generated by Poisson distribution, and the weights of the four factors decrease in turn. The target layer is to prioritize the computable vehicles, and the criterion layer considers f h , Γ h 、N h , the solution layer is the computable vehicle.

[0111] The judgment matrix of the target layer is set to A = (a ij ) 4×4 , where a ij As shown below:

[0112]

[0113] a ij Represents the importance comparison value between factor i and factor j.

[0114] The square root method in the approximate algorithm is used to calculate the weight value corresponding to each factor and normalize it to obtain the weight matrix W = (ω1, ω2, ω3, ω4), where

[0115]

[0116] The judgment matrix is ​​subjected to consistency check. The consistency check requires normalizing the judgment matrix, and then calculating the ratio of the consistency index value to the random index value, and judging whether the obtained ratio value is less than 0.1. If it is less than 0.1, the consistency check is passed.

[0117] If the consistency check is passed, the vehicle can be calculated using the min-max standardization algorithm to calculate the vehicle f h , Γ h 、N h The normalized matrix B = (β hi ) H×4 ,in

[0118]

[0119]

[0120]

[0121]

[0122] They are f h , Γ h 、N h The normalized value; for The maximum and minimum values ​​of f max 、f min f h The maximum and minimum values ​​of N max 、N min N h The maximum and minimum values ​​of .

[0123] If the judgment matrix does not pass the consistency test, the priority of the four factors is judged, and a is calculated based on the selected number. ij And manually adjust the judgment matrix, then return to the previous step to recalculate the weight matrix;

[0124] The priority of the calculable vehicle is calculated as ζ=(ξ hi ) H×4 , where ξ hi is the priority of the i-th factor of the h-th computable vehicle, and ξ hi =BW T =(β hi ) H×4 (ω1,ω2,ω3,ω4) T

[0125] Right now

[0126]

[0127] in, They are f h , Γ h 、N h The normalized value.

[0128] 2. Unloading vehicle task allocation

[0129] like Figure 3As shown in the figure, the task and vehicle allocation problem is converted into an optimal matching problem, and a subtask set and an attached vehicle set are defined. After the computable vehicles are prioritized, the computable vehicles with large priorities are selected in turn according to their priorities to join the attached vehicle set until the number of computable vehicles in the attached vehicle set is the same as the number of subtasks; the subtasks are matched with the computable vehicles in the attached vehicle set using the KM algorithm to minimize the total task unloading calculation time. If the matching fails and there are still selectable cooperative vehicles, the task is matched twice to improve the success rate of task unloading; if the matching fails and there are no unselected computable vehicles, the main vehicle needs to wait for the next moment to unload the task or unload the task to the base station. Finally, the optimal match that minimizes the total task calculation time is obtained.

[0130] like Figure 5 As shown in the figure, the minimum calculation time of tasks under different simulation time steps according to the virtual edge-cooperative vehicle unloading and cooperative distributed unloading-SQP strategies and the algorithm proposed in this paper is changed. By comparing under different simulation time steps, the SQP is a sequential quadratic programming algorithm; the calculation time of the scheme proposed in this paper is shorter than that of the virtual edge-cooperative vehicle unloading and cooperative distributed unloading-SQP strategies. Through data analysis, it can be found that when the computable vehicles are densely distributed, the total calculation time of the three comparison algorithms is relatively small. Among them, the cooperative distributed unloading-SQP strategy is not only affected by the distribution of vehicles, but also related to the number of vehicles. When the number of vehicles is large, tasks will be unloaded for the computable vehicles to which they belong, so its space complexity and task calculation time will also increase accordingly.

[0131] like Figure 6 As shown in the figure, the task failure rate comparison under different time steps. As can be seen from the figure, the task failure rate of the algorithm proposed in this paper is the lowest. Since the algorithm proposed in this paper prioritizes the computable vehicles, it can give priority to the optimal computable vehicle set affected by multiple decision factors, and provides secondary matching for task offloading failure, thus reducing the task offloading failure rate. When selecting computable vehicles, the virtual edge-cooperative vehicle offloading algorithm considers the edge power and the connected communication time of the vehicles in the virtual edge. When offloading tasks in the order of edge power, if the vehicle with higher priority has a smaller executable task, it may skip the vehicle and select a vehicle with lower priority for offloading judgment. The communication time connected to the vehicle with lower priority is shorter, which may lead to an increase in the task offloading failure rate. The collaborative distributed offloading-SQP strategy algorithm offloads tasks for all computable vehicles. In addition to being affected by the number of vehicles, it is also affected by factors such as the computing power of the offloaded vehicle. Therefore, its offloading rate is affected by vehicle environmental factors.

[0132] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A vehicle collaborative task unloading method in VEC, characterized in that: The following steps are involved: S1. Establishing a system model, wherein the system model includes a main vehicle, a computable vehicle, and a non-computable vehicle; S2. Establishing a communication model according to the total time required for task unloading from the host vehicle to the computable vehicle; S3. Establish a vehicle selection model based on multiple dynamic factors; S4, establishing a task offloading optimization problem based on constraints based on the communication model and the vehicle selection model; S5. Solve the task offloading optimization problem by using the analytic hierarchy process and KM algorithm to obtain the optimal task offloading situation; Step S2 of establishing a communication model includes: S21. Calculate the transmission rate R from the host vehicle to the computable vehicle h : S22. Define uninstallation strategy G: G=[g m,h ] M×H in S23, according to the transmission rate R h and offloading strategy G to compute subtask D m Transmission time S24. According to the transmission time Calculate the total transmission time t of the host vehicle trans : S25. Calculate subtask D according to offloading strategy G m Calculation time: S26. According to subtask D m Calculation time Calculation task calculation time: S27, calculate time t according to task comp and the total transmission time t of the host vehicle trans Calculate the total time t required to offload tasks from the host vehicle to the computable vehicle total : Where B is the channel bandwidth from the main vehicle to the computable vehicle, P h The transmission power from the host vehicle to the computable vehicle, d h is the distance from the main vehicle to the hth calculable vehicle, α is the path loss factor, σ 2 is the additional Gaussian white noise power, D(t) is the set of tasks to be unloaded by the main vehicle at time t, and D m represents the mth subtask, M is the total number of subtasks, H is the total number of computable vehicles, V h represents the hth computable vehicle, f h The computing power of the computable vehicle; Step S3 of establishing a vehicle selection model includes: calculating the selectivity of computable vehicles, calculating the maximum task that the computable vehicles can perform, calculating the computing power of the computable vehicles, and calculating the road overlap between the computable vehicles and the host vehicle; establishing the vehicle selection model according to the selectivity of the computable vehicles, the maximum task that the computable vehicles can perform, the computing power of the computable vehicles, and the road overlap between the computable vehicles and the host vehicle; Calculate the selectivity of calculable vehicles The formula is: Among them, r is the communication radius of the main vehicle; The largest tasks that a computable vehicle can perform include: S321, calculating the communication connection time t between the calculable vehicle and the host vehicle h : Where S is the relative displacement between the calculable vehicle and the host vehicle, ν is the velocity of the host vehicle, and ν h The speed of the vehicle can be calculated; S322: Calculate the maximum number of tasks N that can be performed by the vehicle within the communication connection time. h : Among them, Q h is the remaining capacity of the h-th calculable vehicle; The objective function of the task offloading optimization problem based on constraints is: Among them, constraint C1 indicates that when the subtask of the main vehicle is unloaded to the computable vehicle, its transmission time and calculation time The sum cannot exceed the communication connection time t h ; C2 represents subtask D m The tasks unloaded to the hth computable vehicle cannot exceed the limit of its executable tasks; C3 means that the number of subtasks of the main vehicle cannot exceed the number of computable vehicles; C4 means that each subtask can only be unloaded to one computable vehicle, and V(t) is the set of computable vehicles of the main vehicle at time t; Step S5 solves the task offloading optimization problem including: S51, decomposing the task offloading optimization problem into a vehicle priority sorting problem and an offloading vehicle task allocation problem; S52, solving the vehicle priority sorting problem; S53, solving the unloading vehicle task allocation problem according to the result of the vehicle priority sorting problem, and obtaining the unloading vehicle task allocation result; Step S52 solves the vehicle priority sorting problem including: S521, calculating the selectability of vehicles according to the vehicle selection model The maximum task N that the vehicle can perform can be calculated h , the computing power of the vehicle can be calculated h And the road overlap Γ h These four dynamic factors establish a hierarchical model; S522, set the judgment matrix of the hierarchical model to A=(a ij ) 4×4 , where a ij Indicates the importance comparison value of factor i and factor j; S523. Using the square root method in the approximate algorithm, respectively calculate the weight value corresponding to each dynamic factor, and normalize the calculated weight value to obtain a weight matrix W=(ω1,ω2,ω3,ω4), where ω1,ω2,ω3,ω4 are respectively the normalized weight values ​​of each dynamic factor; S524, perform consistency check on the judgment matrix A. If it passes the consistency check, use the min-max standardization algorithm to calculate the normalized matrix B of each dynamic factor = (β hi ) H×4 ; Otherwise, return to step S522 and adjust the judgment matrix A; where β hi is the normalized result of the i-th factor of the h-th computable vehicle; S525, calculate the priority of the computable vehicle ζ=(ξ hi ) H×4 , where ξ hi =BW Τ =(β hi ) H×4 (ω1,ω2,ω3,ω4) Τ ,ξ hi is the priority of the i-th factor of the h-th computable vehicle; Solving the unloading vehicle task allocation problem includes: S531, defining a subtask set and an attached vehicle set; S532, according to the priority of the computable vehicles, sequentially select computable vehicles with higher priorities to add to the subsidiary vehicle set until the number of computable vehicles in the subsidiary vehicle set is the same as the number of subtasks; S533, using a KM algorithm to match the subtasks in the subtask set with computable vehicles in the attached vehicle set, wherein the KM algorithm is a Kuhn-Mankels algorithm; S534. If the matching is successful, the optimal unloading vehicle task allocation result is obtained; if the matching fails and there are unselected computable vehicles, return to step S533 and add the unselected computable vehicles to the attached vehicle set; if the matching fails and there are no unselected computable vehicles, the main vehicle needs to wait for the next moment to unload the task.

2. A method for unloading vehicle collaborative tasks in VEC according to claim 1, characterized in that: The road overlap degree Γ h Randomly generated from a Poisson distribution.

Images