A double-layer path planning method for electric animal delivery vehicles considering V2V charging

Abstract

The application provides a double-layer path planning method for an electric animal logistics vehicle considering V2V charging, and the specific steps comprise the following steps: S1, establishing a double-layer path planning model for the electric animal logistics vehicle considering V2V charging, determining customer node information, and loading initial information of the electric animal logistics vehicle; S2, establishing a comprehensive optimization target of upper-layer path planning demand of the electric animal logistics vehicle, and planning a driving path with the lowest total distribution cost based on a beluga whale optimization algorithm of reinforcement learning; S3, establishing a lower-layer model optimization target, and selecting V2V charging demand of a small electric vehicle according to real-time power of the electric animal logistics vehicle arriving at each customer node; and S4, respectively calculating total distribution costs of the electric animal logistics vehicle before and after considering V2V charging, and finally selecting a distribution scheme with the lowest total distribution cost. The double-layer path planning model for the electric animal logistics vehicle considering V2V charging can effectively combine a distribution path problem of the electric animal logistics vehicle and a small electric vehicle power demand problem randomly generated in a driving process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of path planning, and particularly relates to a double-layer path planning method for electric logistics vehicles considering V2V charging. BACKGROUND

[0002] The electric logistics vehicle can provide wireless energy supply for small electric vehicles in a stationary state and in motion through V2V (vehicle-to-vehicle). This article mainly considers the scenario that small electric vehicles need to pass by electric logistics vehicles for wireless charging in motion to supplement the power of small electric vehicles. Due to the limited battery capacity, when small electric vehicles run out of power due to various emergencies and there is no charging pile around, they need to be transported back to the charging station for charging or call a mobile charging vehicle for charging. However, on the one hand, small electric vehicles may cause traffic congestion and affect road traffic during the waiting process; on the other hand, the cost of calling a mobile charging vehicle over a long distance is high. The electric logistics vehicle has a large size and battery capacity, so it can consider selling the excess power to small electric vehicles in need of rescue to obtain additional income while ensuring that the electric logistics vehicle has enough power to complete the driving process without changing the driving path. This not only ensures that the logistics company completes the normal distribution, but also realizes the rescue mission for small electric vehicles, improves the efficiency of road traffic, and ultimately reduces the total cost of the electric logistics vehicle.

[0003] For example, the invention patent with application number 202311309823.8 discloses a path planning method based on an improved beluga whale optimization algorithm, including the following steps: S1: multi-intelligent vehicle cooperative inspection system model construction; S1-1: obtaining intelligent vehicle position information; S1-2: calculating intelligent vehicle task urgency; S1-3: calculating vehicle performance score; S1-4: using a task allocation strategy function to intelligently allocate tasks according to the collected position information, task urgency and vehicle performance score; S2: real-time data stream integration; the real-time data stream is composed of real-time sensor data processing, and the real-time data stream in the factory is monitored in real time through the sensors installed on the intelligent vehicle; S3: improved beluga whale optimization algorithm; S3-1: exploration stage; S3-2: development stage. Compared with the prior art, the invention introduces a multi-intelligent vehicle cooperative inspection system, improves the beluga whale optimization algorithm through cooperative inspection and real-time data stream integration, and significantly improves the efficiency and coverage of factory inspection. However, the application does not perform hierarchical processing and analysis on the data, and the calculation amount is large and the efficiency is low.

[0004] Therefore, it is necessary to study a double-layer method that can comprehensively consider multiple factors to achieve optimal path planning of electric logistics vehicles and complete charging demand of small electric vehicles through V2V charging. SUMMARY

[0005] In view of the technical problem of high travel cost of existing path planning electric logistics vehicles, the application provides a double-layer path planning method for electric logistics vehicles considering V2V charging, which realizes the comprehensive optimization of path planning of electric logistics vehicles considering V2V charging, reduces the calculation amount, and effectively improves the calculation efficiency.

[0006] In order to achieve the above purpose, the technical scheme of the application is as follows: a double-layer path planning method for electric logistics vehicles considering V2V charging, comprising the following specific steps:

[0007] S1: establishing a double-layer path planning model for electric logistics vehicles considering V2V charging, and determining the initial information of the upper-layer model;

[0008] S2: establishing a comprehensive optimization target of path planning demand of the upper-layer electric logistics vehicle, and planning a driving path with the lowest total distribution cost based on the beluga whale optimization algorithm of reinforcement learning;

[0009] S3: establishing an optimization target of the lower-layer model according to the driving path with the lowest total distribution cost, and selecting V2V charging demand of small electric vehicles according to the real-time power of the electric logistics vehicle reaching each customer node;

[0010] S4: calculating the total distribution cost of the two schemes when the electric logistics vehicle considers V2V charging, and finally selecting the distribution scheme with the lowest total distribution cost.

[0011] The initial information in the upper-layer model includes customer node position information, customer time window information and initial information of the electric logistics vehicle.

[0012] The customer node position information includes position coordinates on the customer map, and the customer time window information includes the service time window specified by the customer; the initial information of the electric logistics vehicle includes the starting position, initial power and initial load of the electric logistics vehicle.

[0013] The specific method for establishing the comprehensive optimization target of the upper-layer electric logistics vehicle path planning demand in step S2 is as follows:

[0014] S2.1: setting the comprehensive optimization target as minimizing the total distribution cost C of the electric logistics vehicle Total ;

[0015] S2.2: planning based on the beluga whale optimization algorithm of reinforcement learning, solving the problem of minimizing the total distribution cost C of the electric logistics vehicle Total , and obtaining the driving path with the lowest total distribution cost.

[0016] The total distribution cost includes the fixed cost, driving power consumption cost, time consumption cost and time window penalty cost of the electric logistics vehicle, and the total distribution cost C Total is calculated as follows:

[0017] minC Total = min(C Fix + C Travel + C Time + C Penalty );

[0018] where C Total is the total cost of distribution, C Fix is the fixed cost of electric logistics vehicle, C Travel is the driving power consumption cost, C Time is the time consumption cost, C Penalty is the time window penalty cost.

[0019] The fixed cost of electric logistics vehicle is:

[0020]

[0021] where x Oj k is a 0-1 variable, taking 1 when the kth vehicle travels from node i to node j, otherwise 0; C d is the fixed cost of a single vehicle per day, K is the set of electric logistics vehicles; N is the set of customer nodes; R is the set of charging stations;

[0022] The driving power consumption cost of electric logistics vehicle is:

[0023]

[0024] γ is the vehicle charging price; is the rated capacity of the battery of the kth electric logistics vehicle; is the maximum driving distance of the kth electric logistics vehicle; d ij is the distance between node i and node j; x ij k is a 0-1 variable, taking 1 when the kth vehicle travels from node i to node j, otherwise 0; V is the set of all nodes;

[0025] The time consumption cost of electric logistics vehicle is:

[0026]

[0027] where, is the total consumption time of all electric logistics vehicles, ξ is the penalty cost generated by time consumption;

[0028] The penalty cost of electric logistics vehicle violating the customer's time window is:

[0029]

[0030] Wherein, N is the set of all customer points, P(t ki ) is the penalty function for violating the customer time window, expressed as:

[0031]

[0032] Wherein, t ki,a is the time when the electric logistics vehicle k arrives at node i; T ki,L represents the earliest arrival time of the user time window, T ki,U represents the latest arrival time of the user time window; t ki,l represents the distance of the exponential function shifted on the coordinate horizontal axis, t ki,l is determined by the value of T ki,L ; T ki,max represents the maximum time allowed to arrive; P max represents the maximum penalty cost; q is an adjustable factor, and the value range of q is [0-1].

[0033] The specific method for planning the driving path with the lowest total distribution cost in step S2.3 is:

[0034] The population size of the beluga whale optimization algorithm is trained using the Q-learning method in reinforcement learning, and the population size is equivalent to the number of electric logistics vehicles. The expectation is calculated using Q table I, and the column of Q table I represents the population size corresponding to the training. The population size is equivalent to the number of electric logistics vehicles, and the row of Q table I represents the total distribution cost of the electric logistics vehicle in the training. The value in Q table I determines the selected population size. The optimal beluga whale population size obtained through training is 30;

[0035] The individual distance of the beluga whale population hunting in the beluga whale optimization algorithm is trained using the Q-learning method in reinforcement learning, and the individual distance is equivalent to the distance between the electric logistics vehicles. The expectation is calculated using Q table II, and the column of Q table II represents the selection of the distance between the electric logistics vehicles. The row of Q table II represents the total distribution cost of the electric logistics vehicle, and the value in Q table II determines the selected population size. The optimal beluga whale individual distance obtained through training is 0.5;

[0036] According to the training results of the population size and the individual distance, the beluga whale population size is initialized to 30, the individual distance in the exploration stage of the beluga whale population is 0.5, and the iteration number is 100 times. The total distribution cost of the electric logistics vehicle is calculated at each iteration using the beluga whale optimization algorithm, and the driving path with the lowest total distribution cost is selected when the maximum iteration number is reached.

[0037] The specific method for establishing the optimization target of the lower layer model in step S3 is:

[0038] S3.1: In the driving path with the lowest total distribution cost, a small electric vehicle with V2V charging demand is randomly generated between two customer nodes.

[0039] S3.2: Calculate the real-time electricity of the electric logistics vehicle on the driving path with the lowest total distribution cost at each customer node by the energy consumption calculation formula, and use it as the basis for decision-making whether to meet the charging demand of small electric vehicles. The energy consumption calculation formula is as follows:

[0040] f r =(a+gc r )(m0+m i );

[0041]

[0042]

[0043] Where c is the air resistance coefficient, A is the vehicle frontal area, a is the vehicle acceleration, p is the air density, c is the rolling resistance coefficient, g is the universal gravitational constant, m0 is the vehicle weight, v is the uniform speed, and m is the weight change of the electric logistics vehicle at node i. d r i

[0044] S3.3: According to the real-time electricity of the electric logistics vehicle arriving at each customer node, select whether to meet the charging demand of small electric vehicles.

[0045] When the electric logistics vehicle chooses to meet the V2V charging demand of small electric vehicles, calculate the real-time electricity of the electric logistics vehicle after completing V2V charging and continue driving. If the current electricity is sufficient to support the electric logistics vehicle to complete the subsequent distribution task and return to the distribution center, the total distribution cost will increase by a V2V charging benefit. If the current electricity is insufficient to support the electric logistics vehicle to complete the subsequent distribution task or cannot return to the distribution center, the distribution task will be ended, and the total distribution cost will be additionally penalized.

[0046] The specific method of the final selection of the distribution scheme with the lowest total distribution cost in step S4 is as follows:

[0047] S4.1: Calculate the driving path of the electric logistics vehicle with the lowest total distribution cost without considering V2V charging and record the value of the total distribution cost.

[0048] S4.2: Calculate the distribution cost of the electric logistics vehicle considering V2V charging, generate two schemes, and re-plan the distribution route.

[0049] S4.3: The logistics company compares the total distribution costs generated by the two schemes in the previous step when completing the distribution task, and selects the scheme with the lowest total distribution cost.

[0050] The specific method of re-planning the distribution route in step S4.2 is as follows: ​​​

[0051] The upper model and the lower model are used to calculate the driving path and the total cost of the electric logistics vehicle, the electric quantity of the electric logistics vehicle is monitored in real time when the V2V charging is considered, and the small electric vehicle with charging demand is charged, and the real-time electric quantity of the electric logistics vehicle after completing the charging is used to determine whether the electric logistics vehicle can continue to complete the distribution task;

[0052] When the electric logistics vehicle cannot continue to complete the distribution task, two schemes are generated, the first scheme is to select the charging demand of the small electric vehicle, but the number of electric logistics vehicles needs to be increased, and the distribution route needs to be re-planned, and finally the total cost of the distribution of multiple electric logistics vehicles and the V2V charging demand of the small electric vehicle is calculated;

[0053] The second scheme is to refuse to meet the charging demand of the small electric vehicle, and the total cost of the distribution task is generated when the number of electric logistics vehicles is considered.

[0054] The beneficial effects of the present application are:

[0055] 1. The double-layer path planning model of the electric logistics vehicle considering V2V charging can effectively combine the distribution path problem of the electric logistics vehicle and the temporary charging problem of the small electric vehicle in the driving process, and generate the optimal driving path of the electric logistics vehicle while solving the temporary charging problem of the small electric vehicle by V2V charging.

[0056] 2. The beluga whale optimization algorithm based on reinforcement learning can develop the best population number in the optimization stage and the best distance between beluga whale individuals in the exploration stage when optimizing the electric logistics vehicle path problem, and give the driving path of the electric logistics vehicle with the lowest total distribution cost. DETAILED DESCRIPTION

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0058] Figure 1 It is the overall framework diagram of the double-layer model of the present application.

[0059] Figure 2 It is the comparison diagram of different algorithm distribution assemblies of the present application.

[0060] Figure 3 It is the experimental result diagram of different vehicle double-layer models of the present application.

[0061] Figure 4The experimental results of the two cases of the present application are compared in the following graph.

[0062] Figure 5 Q table of the present application. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort belong to the scope of protection of the present application.

[0064] Embodiment 1

[0065] As shown in the following table, a double-layer path planning method for electric animal delivery vehicles considering V2V charging includes the following specific steps: Figure 1

[0066] S1: Establish a double-layer path planning model for electric animal delivery vehicles considering V2V charging, determine the location information of all customer nodes that need to be served, the customer time window information, and load the initial information of the electric animal delivery vehicles.

[0067] Specifically, first, a double-layer path planning model for electric animal delivery vehicles considering V2V charging is established. The double-layer path planning model includes an upper model and a lower model. The upper model solves the path planning problem of electric animal delivery vehicles with time windows. The lower model adds the V2V charging demand of small electric vehicles. The electric animal delivery vehicles start from the distribution center and consider meeting the V2V charging demand of small electric vehicles while ensuring the completion of the distribution task of the upper model. The upper model includes the location information of the customer nodes that need to be served, the customer time window information, and the initial information of the electric animal delivery vehicles. The location information of the customer nodes that need to be served includes the location coordinates on the customer map. The customer time window information includes the service time window specified by the customer. The initial information of the electric animal delivery vehicles includes the starting position, the initial power, and the initial load.

[0068] First, the location of the distribution center O is determined. The service demand of the customer nodes, the location coordinates on the customer map, and the service time window specified by the customer are known. In order to complete the customer demand, K electric animal delivery vehicles are dispatched from the distribution center to serve N customer nodes. There are S charging stations in the distribution area and their locations are fixed and known.

[0069] S2: Establish a comprehensive optimization objective for the upper electric animal delivery vehicle path planning demand, and plan a driving path with the lowest total distribution cost based on the beluga whale optimization algorithm of reinforcement learning. The specific steps are as follows:

[0070] S2.1: The comprehensive optimization objective is to minimize the total distribution cost C of the electric animal delivery vehicles​Total , total delivery cost C Total including fixed cost C Fix , driving power consumption cost C Travel , time consumption cost C Time and time window penalty cost C Penalty , total delivery cost C Total is calculated as:

[0071] min C Total = min(C Fix + C Travel + C Time + C Penalty ).

[0072] S2.2: Calculate the fixed cost of dispatching all electric vehicles from the distribution center in one day.

[0073] Let C Fix denote the fixed cost of dispatching all electric vehicles from the distribution center in one day as:

[0074]

[0075] where x Oj k is a 0-1 variable, taking 1 when the kth vehicle travels from node i to node j, otherwise 0; C d is the fixed cost of a single vehicle per day, K is the set of electric vehicles; N is the set of customer nodes; R is the set of charging stations.

[0076] Calculate the driving power consumption cost of electric vehicles.

[0077] Let C Travel denote the driving power consumption cost of electric vehicles as:

[0078]

[0079] γ is the vehicle charging price; is the rated capacity of the kth electric vehicle battery; is the maximum driving distance of the kth electric vehicle; d ij is the distance between node i and node j; x ij k is a 0-1 variable, taking 1 when the kth vehicle travels from node i to node j, otherwise 0; V is the set of all nodes.

[0080] Calculate the time consumption cost of electric vehicles.

[0081] Let C Time denote the time consumption cost of electric vehicles as:

[0082]

[0083] wherein, is the total consumption time of all electric logistics vehicles, and ξ is the penalty cost generated by time consumption.

[0084] The penalty cost of electric logistics vehicles violating the customer time window is calculated.

[0085] Let C Penalty The penalty cost of electric logistics vehicles violating the customer time window is represented as:

[0086]

[0087] wherein, N is the set of all customer nodes, P(t ki ) is the penalty function for violating the customer time window, and is represented as:

[0088]

[0089] T ki,U = T ki,L + q*T ki,L

[0090] wherein, wherein t ki,a is the time when the electric logistics vehicle k arrives at the node i; T ki,L represents the earliest arrival time of the user time window, T ki,U represents the latest arrival time of the user time window; t ki,l represents the distance of the exponential function shifted on the coordinate horizontal axis, t ki,l is determined by the value of T ki,L ; T ki,max represents the maximum time allowed to arrive; P max represents the maximum penalty cost; q is an adjustable factor in the range of [0-1]; the value of q is 0.6.

[0091] S2.3: The beluga optimization algorithm based on reinforcement learning plans the driving path with the lowest total delivery cost.

[0092] The training function is defined to complete the delivery task while outputting the reward value, and then the electric logistics vehicle starts the delivery task, and the Q value table is updated. The electric logistics vehicle completes the customer node delivery task according to the comprehensive optimization goal of minimizing the total delivery cost of the electric logistics vehicle. In each training period, the electric logistics vehicle will select an action randomly with an ε probability according to the current state by using a simple ε-greedy method. In this application, ε takes a fixed value of 0.1. After the electric logistics vehicle selects the appropriate action according to the current state, the corresponding reward is obtained and the state is updated, and then the corresponding Q value is updated by revising the Bellman equation, which is represented as:

[0093] Q(s, a) <- (1 - a) · Q(s, a) + a[r + ymax(Q(s', a'))];

[0094] wherein: s is the current state; a is the action taken; r is the immediate reward obtained by the current state s after action a is taken to move to state s'; s' is the next state moved to; a' is the best action taken in the next state; a is the learning rate, 0 < a < 1, a indicates how much trust is given to the improved update part, generally the value is between 0.001-0.1, the value taken in the present application is 0.1; y is the discount factor, the selected y value should be close to 1, the value taken in the present application is 0.9. After defining all the parameters, the training process will continue until the performance of the intelligent agent reaches the expected, i.e. the total cost of delivery is the lowest.

[0095] Specifically, as shown in Figure 5 , the Q-learning method in reinforcement learning is used to train the population size of the beluga whale optimization algorithm, the column of the Q table represents the action (corresponding to the selection of the population size in the training, the population size is equivalent to the number of electric animal flow vehicles), the row represents the state (the total cost of delivery of electric animal flow vehicles in the training), the size of the Q value determines the selected population size (the larger the Q value, the better the effect), the number of a beluga whale population is usually kept between 15-40, and the optimal beluga whale population size obtained by training is 30, and the updating method of the Q table is as shown in Figure 5 , the Q table is a 4x4 matrix.

[0096] The Q-learning method in reinforcement learning is used to train the individual spacing (usually between 0-1) of the beluga whale population hunting in the beluga whale optimization algorithm, the column of the Q table represents the action (corresponding to the selection of the individual spacing size in the training, the individual is equivalent to the electric animal flow vehicle), the row represents the state (corresponding to the total cost of delivery of electric animal flow vehicles in the training), the size of the Q value determines the selected population size (the larger the Q value, the better the effect), and the optimal beluga whale individual spacing size obtained by training is 0.5.

[0097] According to the training results of the population size and the training results of the individual spacing, the beluga whale population size is initialized to 30, the individual spacing in the beluga whale population in the exploration stage is 0.5, the optimization target is determined to be the minimization of the total cost of delivery of electric animal flow vehicles, the iteration number is 100 times, the total cost of delivery of electric animal flow vehicles is calculated for each iteration, and the generated path with the lowest total cost of delivery when the maximum iteration number is reached is selected.

[0098] S3: Establish the optimization target of the lower layer model, and select whether to meet the charging demand of the small electric vehicle according to the real-time electric quantity of the electric animal flow vehicle arriving at each customer node.

[0099] S3.1: Randomly generate a small electric vehicle with V2V charging needs between two customer nodes on the generation path with the lowest total delivery cost. Select dynamic wireless charging as the charging method, and the charging needs are known.

[0100] S3.2: The real-time battery power of the electric logistics vehicle at each customer node on the route with the lowest total delivery cost is calculated using the energy consumption calculation formula, and this serves as the basis for deciding whether to meet the charging needs of small electric vehicles. The energy consumption calculation formula and the meaning of the parameters in the formula are as follows:

[0101] f r =(a+gc) r (m0+m) i );

[0102]

[0103]

[0104] Among them, c d Let be the air drag coefficient, with a value of 0.7; A be the frontal area of ​​the vehicle, with a value of 6; a be the vehicle acceleration, with a value of 0; ρ be the air density, with a value of 1.2; and c be the air drag coefficient. r ρ is the rolling resistance coefficient, with a value of 0.01; g is the gravitational coefficient, with a value of 9.81; m0 is the vehicle's weight, with a value of 3; v is the constant speed, with a value of 50; m i Let represent the change in load of the electric logistics vehicle at node i.

[0105] S3.3: Select whether to meet the charging needs of small electric vehicles based on the real-time battery level of the electric logistics vehicle upon arrival at each customer node.

[0106] When an electric logistics vehicle chooses to meet the V2V charging needs of a small electric vehicle, the real-time battery level of the electric logistics vehicle after completing V2V charging is calculated and the vehicle continues to drive. If the current battery level is sufficient to support the electric logistics vehicle to complete subsequent delivery tasks and return to the distribution center, the total delivery cost will be increased by a V2V charging benefit. If the current battery level is insufficient to support the electric logistics vehicle to complete subsequent delivery tasks or cannot return to the distribution center, the delivery task will be terminated, and the total delivery cost will be subject to an additional penalty.

[0107] S4: Calculate the total delivery cost for electric logistics vehicles before and after considering V2V charging, and finally select the delivery plan with the lowest total delivery cost.

[0108] S4.1: Calculate the electric logistics vehicle route with the lowest total delivery cost without considering V2V charging and record the total delivery cost value.

[0109] Specifically, the white whale optimization algorithm based on reinforcement learning calculates the driving path of the electric animal distribution vehicle with the lowest total distribution cost by using the upper model of the double-layer path planning model.

[0110] S4.2: Calculate the distribution cost of the electric animal distribution vehicle considering V2V charging, and whether to increase the number of electric animal distribution vehicles and re-plan the distribution route.

[0111] The upper model and the lower model are used to calculate the driving path of the electric animal distribution vehicle and the value of the total distribution cost. When V2V charging is considered, the electric quantity of the electric animal distribution vehicle is monitored in real time, and the small electric vehicle with charging demand is charged. The real-time electric quantity of the electric animal distribution vehicle after charging is used to determine whether the electric animal distribution vehicle can continue to complete the distribution task;

[0112] When the electric animal distribution vehicle cannot continue to complete the distribution task, two solutions are generated. The first solution is to select the charging demand of the small electric vehicle, but additional electric animal distribution vehicles are needed, and the distribution route is re-planned. Finally, the total distribution cost of sending multiple electric animal distribution vehicles and meeting the V2V charging demand of the electric animal distribution vehicle is calculated.

[0113] The second solution is to refuse to meet the charging demand of the small electric vehicle, and consider the total distribution cost generated when the current number of electric animal distribution vehicles meets the distribution task.

[0114] S4.3: The logistics company compares the total distribution costs generated by the two solutions in the previous step to complete the distribution task, and selects the solution with the lowest total distribution cost.

[0115] Embodiment 2

[0116] A double-layer path planning method for electric animal distribution vehicles considering V2V charging is proposed. The V2V is analyzed for different numbers of electric animal distribution vehicles. First, two cases are considered. Case 1: a single vehicle completes the distribution task of 25 customer nodes. The experimental results are shown in Table 1. Case 2: two vehicles complete the distribution task of 25 customer nodes. The experimental results are shown in Table 2. The comparison chart of the experimental results of the two cases is shown in Figure 4 Each electric animal distribution vehicle needs to return to the distribution center from the distribution center under the two cases, which includes three scenarios. Scenario 1: the path planning of the electric animal distribution vehicle does not consider V2V charging. Scenario 2: V2V charging is considered, but the distribution task of the upper model cannot be completed. At this time, the electric animal distribution vehicle will be punished by the distribution constraint. Scenario 3: V2V charging is considered and the distribution task of the upper model is completed.

[0117] Table 1 Total distribution cost results of a single vehicle completing 25 customer nodes

[0118]

[0119]

[0120] Table 2 total cost of delivery results of two vehicles completing 25 customer nodes

[0121]

[0122] The application first establishes an electric logistics vehicle double-layer path planning model considering V2V charging, determines the delivery task of the electric logistics vehicle, and trains the optimal white whale optimization algorithm parameters by using the Q learning algorithm in reinforcement learning; then uses the white whale optimization algorithm based on reinforcement learning to optimize the total delivery cost of the electric logistics vehicle completing the delivery task in the upper model and generate the driving path with the minimum total delivery cost; then the real-time power of the electric logistics vehicle at each customer node is obtained according to the energy consumption calculation formula, and it is judged whether the V2V charging demand of the small electric vehicle in the lower model is met, if the number of electric logistics vehicles cannot meet the V2V charging demand of the small electric vehicle or meet it but cannot complete the delivery task, the logistics company can increase the number of electric logistics vehicles to complete the charging demand of the small electric vehicle; finally, the logistics company selects the delivery mode with the lowest total delivery cost under the double-layer model. The application effectively combines path planning in traditional logistics problems and V2V wireless charging, does not affect the delivery task of the electric logistics vehicle during driving, and at the same time supplements the power of the small electric vehicle through the V2V mode, providing a better path planning solution.

[0123] Other structures and principles are the same as those of Embodiment 1.

[0124] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A two-layer path planning method for electric logistics vehicles considering V2V charging, characterized in that, The specific steps include: S1: Establish a two-layer path planning model for electric logistics vehicles that considers V2V charging, and determine the initial information of the upper-layer model; S2: Establish a comprehensive optimization objective for the path planning requirements of upper-level electric logistics vehicles, and plan the driving path with the lowest total delivery cost based on the reinforcement learning-based white whale optimization algorithm; S3: Establish a lower-level model optimization objective based on the driving route with the lowest total delivery cost, and determine whether the real-time power of the electric logistics vehicle upon arrival at each customer node meets the V2V charging requirements of the small electric vehicle. S4: Calculate the total delivery cost for electric logistics vehicles before and after considering V2V charging, and finally select the delivery plan with the lowest total delivery cost.

2. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 1, characterized in that, The initial information in the upper-level model includes the location information of the customer nodes that need to be served, the customer time window information, and the initial information of the electric logistics vehicles.

3. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 2, characterized in that, The customer node location information includes the location coordinates on the customer map, and the customer time window information includes the service time window specified by the customer; the initial information of the electric logistics vehicle includes the starting position of the electric logistics vehicle, the initial battery level, and the initial cargo capacity.

4. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 2 or 3, characterized in that, The specific method for establishing the comprehensive optimization objective of the upper-level electric logistics vehicle route planning requirements in step S2 is as follows: S2.1: Set the overall optimization objective as minimizing the total delivery cost C of electric logistics vehicles. Total ; S2.2: Using the Beluga optimization algorithm based on reinforcement learning to solve for minimizing the total delivery cost C of electric logistics vehicles. Total The problem is to find the driving route with the lowest total delivery cost.

5. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 4, characterized in that, The total delivery cost includes the fixed cost of the electric logistics vehicle, the cost of electricity consumption, the cost of time consumption, and the cost of time window penalties. The total delivery cost C Total The formula is: minC Total =min(C Fix +C Travel +C Time +C Penalty ); Among them, C Total For the total delivery cost, C Fix For the fixed costs of electric logistics vehicles, C Travel For the cost of electricity consumed during driving, C Time For time consumption costs, C Penalty Penalty costs for time windows.

6. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 5, characterized in that, The fixed costs of electric logistics vehicles are: Where, x Oj k The variable is 0-1, taking the value 1 when the k-th vehicle travels from node O to node j, and 0 otherwise; C d Let K be the set of electric logistics vehicles, N be the set of customer nodes, and R be the set of charging stations, representing the fixed daily cost per vehicle. The electricity cost of operating an electric logistics vehicle is: γ represents the electricity price for vehicle charging; The rated capacity of the battery of the kth electric logistics vehicle; d represents the maximum mileage of the kth electric logistics vehicle. ij x is the distance between node i and node j; ij k V is a 0-1 variable, set to 1 when the kth vehicle travels from node i to node j, and 0 otherwise; V is the set of all nodes. The time cost of electric logistics vehicles is: in, Let ξ be the total time consumed by all electric logistics vehicles, and let ξ be the penalty cost incurred due to time consumption. The penalty cost for electric logistics vehicles violating the customer's time window is: Wherein, P(t) ki,a The penalty function for violating the client's time window is expressed as: Among them, t ki,a T represents the time it takes for electric logistics vehicle k to arrive at node i; ki,L T represents the earliest arrival time of the user's time window. ki,U Indicates the latest arrival time of the user's time window; t ki,l t represents the distance the exponential function is shifted along the horizontal axis of the coordinate system. ki,l Through T ki,L The value of P is determined; max This indicates the maximum penalty cost.

7. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 6, characterized in that, The specific method for planning the driving route with the lowest total delivery cost is as follows: The Q-learning method in reinforcement learning is used to train the population size of the beluga optimization algorithm. The population size is equivalent to the number of electric logistics vehicles. The expected value is calculated using Q-table I. The columns of Q-table I represent the corresponding population size during training. The population size is equivalent to the number of electric logistics vehicles. The rows of Q-table I represent the total delivery cost of electric logistics vehicles during training. The values ​​in Q-table I determine the selected population size. The optimal beluga population size obtained through training is 30. The Q-learning method in reinforcement learning is used to train the beluga whale optimization algorithm to measure the individual spacing during beluga whale group hunting. The individual spacing is equivalent to the spacing of electric logistics workshops. The expected value is calculated using Q-table II. The columns of Q-table II represent the selection of the electric logistics workshop spacing size, and the rows of Q-table II represent the total delivery cost of electric logistics vehicles. The median of Q-table II determines the population size to be selected. The optimal beluga whale individual spacing size obtained through training is 0.

5. Based on the training results of population size and inter-individual spacing, the initial population size of beluga whales was set to 30, the inter-individual spacing in the beluga whale population during the exploration phase was set to 0.5, and the number of iterations was set to 100. The beluga whale optimization algorithm was used to calculate the total delivery cost of the electric logistics vehicle in each iteration and select the driving path with the lowest total delivery cost when the maximum number of iterations was reached.

8. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 7, characterized in that, The specific method for establishing the optimization objective of the lower-level model in step S3 is as follows: S3.1: In the driving route with the lowest total delivery cost, a small electric vehicle with V2V charging needs is randomly generated between two customer nodes; S3.2: Calculate the real-time battery level of the electric logistics vehicle at each customer node along the route with the lowest total delivery cost using the energy consumption calculation formula. This calculation will serve as the basis for deciding whether to meet the charging needs of small electric vehicles. The energy consumption calculation formula is as follows: f r =(a+gc r )(m0+m i ); Among them, c d Let A be the air drag coefficient, A be the frontal area of ​​the vehicle, a be the vehicle acceleration, ρ be the air density, and c be the air drag coefficient. r Where g is the rolling resistance coefficient, m0 is the gravitational coefficient, v is the vehicle weight, and m is the constant speed. i Let represent the change in load of the electric logistics vehicle at node i. S3.3: Based on whether the real-time battery level of the electric logistics vehicle upon arrival at each customer node meets the charging requirements of the small electric vehicle, And calculate the charging benefits or additional penalties for small electric vehicles; When an electric logistics vehicle chooses to meet the V2V charging needs of a small electric vehicle, the real-time battery level of the electric logistics vehicle after completing V2V charging is calculated and the vehicle continues to drive. If the current battery level is sufficient to support the electric logistics vehicle to complete subsequent delivery tasks and return to the distribution center, the total delivery cost will be increased by a V2V charging revenue item. If the current battery level is insufficient to support the electric logistics vehicle to complete subsequent delivery tasks or cannot return to the distribution center, the delivery task will be terminated, and the total delivery cost will be subject to an additional penalty.

9. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to claim 8, characterized in that, The specific method for selecting the delivery plan with the lowest total delivery cost in step S4 is as follows: S4.1: Calculate the electric logistics vehicle route with the lowest total delivery cost without considering V2V charging and record the total delivery cost value; S4.2: Calculate the delivery cost of electric logistics vehicles when considering V2V charging, generate two plans, and replan the delivery route; S4.3: The logistics company compares the total delivery costs incurred by the two options in the previous step when completing the delivery task, and selects the option with the lowest total delivery cost.

10. The dual-layer path planning method for electric logistics vehicles considering V2V charging according to any one of claims 6-9, characterized in that, The specific method for replanning the delivery route in step S4.2 is as follows: The upper-level model and the lower-level model are used to calculate the driving path and total delivery cost of electric logistics vehicles. When considering V2V charging, the power of electric logistics vehicles is monitored in real time, and small electric vehicles with charging needs are charged. The real-time power of electric logistics vehicles after charging is used to determine whether the electric logistics vehicle can continue to complete the delivery task. When electric logistics vehicles are unable to continue to complete delivery tasks, two options are generated. The first option is to choose to complete the charging needs of small electric vehicles, but it is necessary to increase the number of electric logistics vehicles and replan the delivery routes. Finally, the total delivery cost of dispatching multiple electric logistics vehicles and meeting the V2V charging needs of small electric vehicles is calculated. The second option is to refuse to meet the charging needs of small electric vehicles, taking into account the total delivery cost incurred when the current number of electric logistics vehicles is used to meet delivery tasks.

Images