A multi-pricing model path optimization method suitable for a car-free delivery scene

Abstract

The application provides a multi-pricing model path optimization method suitable for a non-vehicle carrying scene, characterized in that two three-section coding methods for contract carrying capacity optimization and outsourcing carrying capacity optimization are included, the three-layer coding method well integrates key information such as train number, train type, pricing mode and distribution sequence in the same chromosome, and gives an integrated solution method for an order distribution problem, a pricing mode selection problem and a path optimization problem without destroying the optimization structure of the metaheuristic.

Description

Technical Field

[0001] This invention relates to the field of non-vehicle-operated logistics platform technology, and in particular to a multi-pricing model path optimization method applicable to non-vehicle-operated scenarios. Background Technology

[0002] In recent years, the asset-light logistics (AVC) model, relying on urban logistics SaaS platforms, has gained increasing attention from industry practitioners. In this model, the carrier does not own the transportation capacity; instead, they integrate professional or idle urban capacity to achieve timely and low-cost order delivery. However, the operational model reveals a highly non-standardized transportation capacity, posing significant challenges to operation and dispatch. When an AVC platform hires non-self-operated delivery vehicles, there will be inconsistencies in vehicle types (including load capacity, mileage, and fuel type). Furthermore, the probability of urban delivery drivers accepting orders is largely influenced by the pricing model. Summary of the Invention

[0003] The purpose of this invention is to propose an integrated multi-pricing model path optimization method for order allocation, pricing mode selection, and path optimization problems in the context of non-vehicle carrier scenarios.

[0004] To achieve the above objectives, this invention proposes a multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios, including two three-segment coding methods for contract capacity optimization and outsourced capacity optimization.

[0005] The contract capacity optimization belongs to the first stage of optimization, which adopts a three-segment coding, including a vehicle selection layer, a vehicle number generation layer, and a delivery route layer. The vehicle selection layer and the vehicle number generation layer both use the interval mapping method to uniformly map vehicles and vehicle numbers to specific intervals on [0,1], thereby determining specific (vehicle, vehicle number) pairs. The third layer is the delivery route layer, where the value of each customer on the route layer is the priority of the corresponding customer's delivery. Combining the (vehicle, vehicle number) pairs, the sequence of customers served by specific vehicles and vehicle numbers is decoded. For solutions with discontinuous departure times or not starting from a small vehicle number, they can be corrected by rearranging the vehicle numbers of a certain vehicle in ascending order, starting from 1.

[0006] The outsourced transportation capacity optimization belongs to the second stage of optimization, which adopts a three-segment coding, including a vehicle selection layer, a pricing method selection layer, and a delivery route layer. The vehicle selection layer and the pricing method selection layer use an interval mapping method to determine specific (vehicle type, pricing method) pairs. The third layer is the delivery route layer, where the value of each customer in the route layer is the priority of the corresponding customer's delivery. Combining the (vehicle type, pricing method) pairs, the customer sequence served by specific vehicles and trips is decoded. The second stage three-segment coding splits the determined routes for (vehicle type, pricing method) using an improved SPLIT algorithm to ensure the legality of each route. The improved SPLIT algorithm makes the following improvements to the original SPLIT algorithm: the splitting criteria simultaneously consider vehicle capacity and maximum transit time constraints, as shown in the pseudocode below.

[0007]

[0008]

[0009] Both stages employ a population-based continuous coding metaheuristic algorithm.

[0010] Furthermore, the improved SPLIT algorithm is a combination of the traditional SPLIT algorithm and vehicle load constraints and maximum working time constraints.

[0011] Furthermore, the fitness function of the first-stage three-segment coding consists of two parts: a cost item and a penalty item. The cost item includes daily fixed wages and fuel consumption costs. The penalty item includes volume overload penalty, mass overload penalty, and overtime penalty.

[0012] Furthermore, the daily fixed wage is:

[0013] Fuel consumption cost:

[0014] Furthermore, the fitness function of the second-stage three-stage coding is the sum of the per-use cost and the per-task cost.

[0015] Furthermore, the cost per dispatch:

[0016] Cost of dispatching vehicles based on task volume:

[0017] Furthermore, the multi-pricing model path optimization method also includes constraints on vehicle load capacity, maximum working time of contract vehicles, maximum working time of vehicles priced per trip, and maximum working time of vehicles priced by workload.

[0018] Furthermore, the vehicle load constraint is as follows:

[0019]

[0020]

[0021]

[0022] The maximum working time constraint for the contracted vehicles is:

[0023]

[0024] The iterative method is as follows.

[0025]

[0026]

[0027]

[0028] Among them, i l The node following customer point i on the path; 0 - The last customer served before the vehicle returns to the parking lot;

[0029] The maximum working time constraint for vehicles priced per trip is:

[0030]

[0031] The maximum working time constraint for vehicles priced based on workload:

[0032]

[0033] Compared with existing technologies, the advantages of this invention are as follows: This invention proposes an integrated optimization algorithm for the characteristics of mixed capacity, multiple vehicle types, and multiple pricing models in non-vessel operating common carrier (NVOCC) platforms. The technical solution adopted in this invention has strong compatibility, and almost all continuous optimization metaheuristic algorithms can be introduced into the two-stage optimization framework proposed in this invention. At the same time, the three-layer encoding method effectively integrates key information such as vehicle number, vehicle type, pricing method, and delivery order into the same chromosome, without significantly disrupting the metaheuristic optimization structure. Thus, it provides an integrated solution to the order allocation problem, pricing mode selection problem, and route optimization problem in the NVOCC scenario. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the multi-pricing model and multi-vehicle routing problem in an embodiment of the present invention;

[0035] Figure 2 This is a schematic diagram of the first stage three-segment encoding-decoding in an embodiment of the present invention;

[0036] Figure 3 This is a schematic diagram of the second stage three-segment encoding-decoding in an embodiment of the present invention;

[0037] Figure 4 This is a two-stage optimization framework diagram in this embodiment of the invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be further described below.

[0039] With the widespread adoption of SaaS in the logistics industry, the asset-light transportation model is increasingly favored by practitioners due to its high efficiency and low cost operational advantages, such as... Figure 1 As shown, the problem addressed by this invention is also based on this real-world context. Due to the "tidal" phenomenon in urban delivery demand, in addition to relying on long-term, stable contracted fleets, asset-light transportation platforms also temporarily assign customers whose existing capacity cannot be served to temporary vehicles. Because outsourced drivers have different preferences, they will have different preferences for different pricing methods. For example, some drivers prefer to be charged based on delivery mileage, some prefer to be charged based on the number of deliveries, and some prefer to be charged based on the amount of work completed. The pricing models involved in this invention include: contract pricing, per-delivery pricing, and per-workload pricing, as detailed in Table 1.

[0040] Table 1 Introduction to Pricing Models

[0041]

[0042] Urban delivery scenarios involve a wide variety of vehicle types, exhibiting significant non-standardization. To better reflect actual delivery scenarios, this invention uses three common vehicle types as examples to illustrate the problems it addresses, as detailed in Table 2.

[0043] Table 2 Vehicle Model Introduction

[0044]

[0045] For ease of modeling, the problem solved by this invention (multi-pricing-model multi-type vehicle routing problem) is denoted as Multi-pricing-model multi-type VRP (MPMTVRP) and modeled on an undirected graph G = (V, A). The set of customer points is C, and the parking lot is denoted as 0; therefore, V = C ∪ {0}. All vehicles depart from the parking lot and return to the parking lot after completing their tasks. For i ∈ C, there is a demand quantity q. i Service time s i The set of available vehicle models is {T1, T2, ..., T...}. max The pricing method set is {P1, P2, ..., P}. max}, where T and P are the vehicle type and pricing method indices, respectively. The vehicle set K = {k s}∪{k ot}∪{k ov}. Where, k s For the contract vehicle index, k ot For the per-trip vehicle index, k ov This is a vehicle index priced based on task volume. The path set is denoted as... R is the set of paths. The maximum vehicle capacity is denoted as Q. k , k∈{k s}∪{k ot}∪{k ov}

[0046] Contract vehicle collection: Collection of vehicles priced per trip: Vehicles priced by task volume:

[0047] Based on the vehicle model index T, the pricing method index P, and the path set R, the pricing models corresponding to Tables 1 and 2 are given, as shown in Formula 1.

[0048]

[0049] Among them, w1-w3 represent the daily vehicle usage cost for Jinbei, Iveco, and Foton under the contract pricing model; w4-w6 represent the single-trip vehicle dispatch cost for Jinbei, Iveco, and Foton under the per-trip pricing model; w7-w9 represent the starting price cost for Jinbei, Iveco, and Foton under the task-based pricing model; w 10 -w 12 For JinBei, Iveco, and Foton, the unit customer service cost is defined under the pay-per-task pricing model; ||R|| represents the number of customer points along the path.

[0050] The platform needs to rationally assign tasks and plan delivery routes for both its own and outsourced fleets to minimize delivery costs while meeting driving constraints and service requirements. On one hand, the platform needs to decide which delivery method and vehicle type to use for each route; on the other hand, in addition to planning the single-trip routes for outsourced vehicles, it also needs to plan the trips of contracted vehicles. To clarify the scope of application of the model in this chapter, the following assumptions are made: customer point needs are indivisible, i.e., they can only be served by the same vehicle on the same trip; each customer point can only be accessed once; vehicles have the same speed; and all customer needs must be met. For ease of reading, the symbols used in this chapter and their descriptions are listed below.

[0051] Indexes and Collections:

[0052] i, j: Customer and parking lot (0) index

[0053] w: Index of vehicle dispatches

[0054] k s Contract Vehicle Index

[0055] k o Temporary vehicle index

[0056] k ot Vehicle Index by Price

[0057] k ov Vehicle index priced by task volume

[0058] T: Vehicle Index

[0059] P: Pricing Method Index

[0060] R: Path Index

[0061] : Set of contract vehicle counts

[0062] V: The set of all points, including the customer point set C and the parking lot set 0.

[0063] K: Set of vehicles K = {k s}∪{k ot}∪{k ov}

[0064] Path collection

[0065] Contract vehicle collection

[0066] Collection of vehicles priced per trip:

[0067] Vehicle collection priced by task volume

[0068] parameter:

[0069] Contract vehicle k s Daily fixed costs

[0070] q i Demand

[0071] s i Service Hours

[0072] d ij : Length of edge (i,j)

[0073] t ij : Travel time of edge (i,j)

[0074] Dur: Maximum working time constraint

[0075] Q k Maximum vehicle capacity, k∈{k s}∪{k ot}∪{k ov}

[0076] w1-w3: Daily vehicle costs for Jinbei, Iveco, and Foton under the contract pricing model

[0077] w4-w6: Cost per trip for JinBei, Iveco, and Foton under the pay-per-trip pricing model

[0078] w7-w9: Starting costs for JinBei, Iveco, and Foton under the pay-per-task pricing model

[0079] w 10 -w 12 Unit customer service cost for JinBei, Iveco, and Foton under the pay-per-task pricing model.

[0080] Vehicle K s Fuel consumption per unit distance (liters / km)

[0081] p: Unit oil price (yuan / liter)

[0082] Decision variables:

[0083]

[0084]

[0085]

[0086]

[0087] Contract vehicle k s The time when the wth delivery leaves the parking lot

[0088] Contract vehicle k s The time of the wth delivery returning to the depot

[0089] Contract vehicle k s Time to reach customer point i in the wth delivery

[0090] Vehicles priced per trip (k) ot Time of departure from parking lot

[0091] Vehicles priced per trip (k) ot Time to return to parking lot

[0092] Vehicles priced per trip (k) ot Time to reach customer point i

[0093] Vehicles priced according to task volume (k) ov Time of departure from parking lot

[0094] Vehicles priced according to task volume (k) ov Time to return to parking lot

[0095] Vehicles priced according to task volume (k) ov Time to reach customer point i

[0096] The cost structure of MPMTVRP consists of: daily fixed wages, fuel costs, dispatch costs per trip, and dispatch costs per task.

[0097] Daily fixed wage:

[0098] Fuel cost:

[0099] Cost per dispatch:

[0100] Cost of dispatching vehicles based on task volume:

[0101] The objective function is: F = F1 + F2 + F3 + F4 (6)

[0102] Vehicle load constraints:

[0103]

[0104]

[0105]

[0106] Maximum working time constraint for contracted vehicles:

[0107]

[0108] The iteration method is as follows:

[0109]

[0110]

[0111]

[0112] Among them, i l The node after customer point i on the path; 0- represents the last customer served before the vehicle returns to the parking lot.

[0113] Maximum working time constraint for vehicles priced per trip:

[0114]

[0115] Maximum working time constraint for vehicles priced based on workload:

[0116]

[0117] The arrival time for vehicles priced per trip and vehicles priced by workload is similar to that of contract vehicles, and will not be repeated here.

[0118] Algorithm Design

[0119] Traditional optimization methods cannot be directly used to solve the problem of this invention for the following reasons: 1) The problem involved in this invention has complex characteristics, integrating features such as multiple journeys, heterogeneous vehicle types, and multiple pricing models, requiring delivery solutions for various types of vehicles, which traditional modeling and optimization methods find difficult to solve; 2) Vehicle types in the non-vehicle carrier platform are extremely non-standard, and the number of orders is often very large.

[0120] Encoding and Decoding

[0121] To design a multi-pricing model path optimization method suitable for asset-light transportation scenarios, this invention proposes two three-segment coding methods for contract capacity optimization and outsourced capacity optimization, such as... Figure 2 and Figure 3 As shown.

[0122] The contract capacity optimization belongs to the first stage of optimization, which adopts a three-segment coding, including a vehicle selection layer, a vehicle number generation layer, and a delivery route layer. The vehicle selection layer and the vehicle number generation layer both use the interval mapping method to uniformly map vehicles and vehicle numbers to specific intervals on [0,1], thereby determining specific (vehicle, vehicle number) pairs. The third layer is the delivery route layer, where the value of each customer on the route layer is the priority of the corresponding customer's delivery. Combining the (vehicle, vehicle number) pairs, the sequence of customers served by specific vehicles and vehicle numbers is decoded. For solutions with discontinuous departure times or not starting from a small vehicle number, they can be corrected by rearranging the vehicle numbers of a certain vehicle in ascending order, starting from 1.

[0123] The outsourced transportation capacity optimization belongs to the second stage of optimization, which adopts a three-segment coding, including a vehicle selection layer, a pricing method selection layer, and a delivery route layer. The vehicle selection layer and the pricing method selection layer use an interval mapping method to determine specific (vehicle type, pricing method) pairs. The third layer is the delivery route layer, where the value of each customer in the route layer is the priority of the corresponding customer's delivery. Combining the (vehicle type, pricing method) pairs, the customer sequence served by specific vehicles and trips is decoded. The second stage three-segment coding splits the determined routes for (vehicle type, pricing method) using an improved SPLIT algorithm to ensure the legality of each route. The improved SPLIT algorithm makes the following improvements to the original SPLIT algorithm: the splitting criteria simultaneously consider vehicle capacity and maximum transit time constraints, as shown in the pseudocode below.

[0124]

[0125]

[0126] Objective function design

[0127] The first phase involves optimizing the company's own transportation capacity. The fitness function for this phase consists of two parts: a cost component and a penalty component. The cost component includes fixed costs and fuel costs; the penalty component includes penalties for volume overload, mass overload, and excessive working time.

[0128] The second phase involves optimizing outsourced transportation capacity. The fitness function for this phase is the sum of the cost per trip and the cost per task.

[0129] Neighborhood search technology

[0130] Variable neighborhood search methods can be introduced according to the optimization objectives of the first and second stages respectively to improve the quality of the solution in a targeted manner.

[0131] Optimize framework

[0132] This invention proposes a two-stage optimization method for routes using both owned capacity (contracted capacity) and outsourced capacity, as illustrated in the diagram below. Figure 4 As shown, since contracted capacity cannot guarantee the completion of delivery tasks, the algorithm fully optimizes the contracted capacity in the first stage and generates feasible delivery routes for contracted vehicles. Customers not assigned vehicles are served by outsourced vehicles, and the algorithm finally generates feasible outsourced vehicle delivery routes in the second stage. Both stages of the algorithm employ population-based continuous coding metaheuristic algorithms, such as particle swarm optimization and differential evolution. Neighborhood search can be placed in the later stages of population search to avoid affecting the evolutionary process of the population.

[0133] The above are merely preferred embodiments of the present invention and do not constitute any limitation on the present invention. Any equivalent substitutions or modifications made by those skilled in the art to the technical solutions and content disclosed in the present invention without departing from the scope of the present invention shall be deemed to have remained within the protection scope of the present invention.

Claims

1. A path optimization method for multi-pricing models applicable to non-vessel operating common carrier (NVOCC) scenarios, characterized in that, This includes two three-segment coding methods for optimizing contracted capacity and optimizing outsourced capacity; The contract capacity optimization belongs to the first stage of optimization, which adopts a three-segment coding, including a vehicle selection layer, a vehicle number generation layer, and a delivery route layer. The vehicle selection layer and the vehicle number generation layer both use the interval mapping method to uniformly map vehicles and vehicle numbers to specific intervals on [0,1], thereby determining specific (vehicle, vehicle number) pairs. The third layer is the delivery route layer, where the value of each customer on the route layer is the priority of the corresponding customer's delivery. Combining the (vehicle, vehicle number) pairs, the sequence of customers served by specific vehicles and vehicle numbers is decoded. For solutions with discontinuous departure times or not starting from a small vehicle number, they can be corrected by rearranging the vehicle numbers of a certain vehicle in ascending order, starting from 1. The outsourced transportation capacity optimization belongs to the second stage of optimization, which adopts a three-segment coding, including a vehicle selection layer, a pricing method selection layer, and a delivery route layer. The vehicle selection layer and the pricing method selection layer use an interval mapping method to determine specific (vehicle type, pricing method) pairs. The third layer is the delivery route layer, where the value for each customer in the route layer is the priority of the corresponding customer's delivery. Combining the (vehicle type, pricing method) pairs, the sequence of customers served by specific vehicles and trips is decoded. The second stage three-segment coding splits the determined routes for (vehicle type, pricing method) pairs using an improved SPLIT algorithm to ensure the validity of each route. The improved SPLIT algorithm makes the following improvements to the original SPLIT algorithm: the splitting criteria simultaneously consider vehicle capacity and maximum transit time constraints, as shown in the pseudocode below:

2. The multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios according to claim 1, characterized in that, In cases where contracted capacity cannot guarantee the completion of delivery tasks, the first phase fully optimizes the contracted capacity and generates feasible delivery routes for contracted vehicles; customers who are not assigned vehicles are served by outsourced vehicles, and in the second phase, feasible outsourced vehicle delivery routes are finally generated. Both stages employ a population-based continuous coding metaheuristic algorithm.

3. The multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios according to claim 1, characterized in that, The fitness function of the first-stage three-segment coding consists of two parts: a cost item and a penalty item. The cost item includes daily fixed wages and fuel consumption costs. The penalty item includes volume overload penalty, mass overload penalty, and overtime penalty.

4. The multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios according to claim 3, characterized in that, The daily fixed wage is: Fuel consumption cost:

5. The multi-pricing model path optimization method applicable to non-vehicle carrier scenarios according to claim 1, characterized in that, The fitness function of the second-stage three-stage coding is the sum of the cost per use and the cost per task.

6. The multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios according to claim 5, characterized in that, The cost per dispatch: Cost of dispatching vehicles based on task volume:

7. The multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios according to claim 1, characterized in that, The multi-pricing model path optimization method also includes constraints on vehicle load capacity, maximum working time of contract vehicles, maximum working time of vehicles priced per trip, and maximum working time of vehicles priced by workload.

8. The multi-pricing model path optimization method applicable to non-vessel operating common carrier (NVOCC) scenarios according to claim 7, characterized in that, The vehicle load limit is: The maximum working time constraint for the contracted vehicles is: The iterative method is as follows: Among them, i l The node following customer point i on the path; 0 - The last customer served before the vehicle returns to the parking lot; The maximum working time constraint for vehicles priced per trip is: The maximum working time constraint for vehicles priced based on workload:

Images