Simulation optimization system and method for distribution path of electric logistics vehicle

Abstract

The invention discloses a simulation optimization system and method for a distribution path of an electric logistics vehicle, and the system comprises a vehicle module which is used for simulating the dynamic behavior and performance characteristics of the electric logistics vehicle, and is used for simulating the driving environment of the electric logistics vehicle in the distribution process. The road module is used for accurately reflecting the current road condition, the station module is used for managing charging, scheduling and task distribution of the electric logistics vehicle, and the road network module is used for ensuring that data and events among different modules can be effectively transmitted. According to the method, the discrete event simulation model considering the vehicle-pile-road interaction influence is constructed, the key dynamic random interaction influence of the electric vehicle logistics distribution system is accurately described, and the accuracy and the calculation efficiency are both considered. The evaluation, configuration, path optimization and other aspects of the electric vehicle logistics distribution system are remarkably improved, the problem that precision and efficiency are difficult to consider in an existing method is effectively solved, and the method has important practical application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the distribution technology field of electric logistics vehicles, and in particular to an electric logistics vehicle distribution path simulation optimization system and method. BACKGROUND

[0002] At present, the modeling field of electric logistics vehicle distribution system is still mainly based on analytical modeling methods, and the research using simulation methods is relatively less. In the existing research, the complex interaction relationship among electric vehicles, charging piles and road traffic is often ignored when constructing the model. This interaction is multidimensional: from the perspective of electric vehicles, the increase in vehicle fleet size and distribution frequency will exacerbate road traffic congestion and increase charging demand due to the range constraint and the need for mid-way charging; from the perspective of charging piles, the number of charging piles affects the queuing time of vehicles at charging stations, which in turn affects the vehicle distribution time and the satisfaction of customer time windows; from the perspective of road traffic, traffic conditions affect the distribution time and power consumption of electric vehicles, which in turn affect the charging time. In addition, logistics demand and traffic conditions are influenced by factors such as time, weather, and economy, and have significant dynamic and random characteristics. Therefore, the complex dynamic and random interaction relationship among vehicles, piles and roads cannot be ignored.

[0003] However, the existing technology has the following deficiencies and defects in the simulation modeling of electric logistics vehicle path planning system: Lack of precision: the existing simulation model fails to accurately describe the dynamic and random interaction between vehicles, charging facilities and roads, resulting in an inability to accurately obtain key performance indicators such as vehicle travel time, power consumption, charging demand, queuing time, charging time of electric vehicle distribution system. These performance indicators are crucial for evaluating distribution efficiency, enterprise cost and customer satisfaction.

[0004] Low computational efficiency: existing discrete-time simulation methods or object-oriented simulation methods are limited by simulation efficiency and are difficult to apply to large-scale systems. At the same time, the existing electric vehicle distribution simulation method often lacks computational efficiency when dealing with large-scale systems, limiting the application of the model in simulation optimization.

[0005] Limited optimization capability: existing electric vehicle distribution optimization methods have a computational complexity that grows exponentially with problem size when faced with actual scale logistics networks, resulting in only small-scale idealized scenarios being solvable. At the same time, these algorithms have weak modeling capabilities for real dynamic factors such as traffic congestion and charging station queues, making it difficult to meet the needs of large-scale, multi-constrained electric vehicle scheduling in actual logistics systems. SUMMARY

[0006] The embodiment of the application provides an electric logistics vehicle distribution path simulation optimization system and method to solve the technical problems of the existing electric logistics vehicle distribution system modeling method, which does not consider the interaction between the vehicle, the pile and the road, and has the problems of insufficient precision, low calculation efficiency and limited optimization capability. By developing a new simulation optimization framework, the modeling precision and calculation efficiency of the system are improved, and the optimization capability for large-scale, multi-constrained logistics networks is enhanced to meet the demand problems of actual logistics systems.

[0007] An electric logistics vehicle distribution path simulation optimization system comprises: A vehicle module is used to simulate the dynamic behavior and performance characteristics of electric logistics vehicles, to realize dynamic scheduling and path optimization by updating the vehicle state in real time and interacting with the station module and the road module; A road module is used to simulate the driving environment of electric logistics vehicles during distribution to accurately reflect the current road conditions by monitoring and updating basic parameters in real time; A station module is used to manage the charging, scheduling and task allocation of electric logistics vehicles, and the types of the station module include charging stations, demand points and distribution centers; A road network module serves as a connection hub of the simulation optimization system, and is used to initialize and configure the station module, the vehicle module and the road module to provide a unified framework for each module and ensure that data and events between different modules can be effectively transmitted.

[0008] An electric logistics vehicle distribution path simulation optimization method comprises: S1, constructing a simulation optimization model of the electric logistics vehicle distribution path and designing a simulation program; S2, establishing a target function and a constraint condition; S3, generating an initial solution based on the target function and the constraint condition by using the Floyd algorithm and the mileage saving method, and then optimizing the initial solution by using an adaptive large neighborhood search algorithm (ALNS).

[0009] The beneficial effects of the application are as follows: The application proposes an optimization framework based on simulation, adopts an adaptive large neighborhood search algorithm (ALNS), obtains various performance indicators of the electric logistics vehicle distribution system through a simulation model, and then optimizes the distribution path of the electric logistics vehicle. The framework accurately simulates the interaction between the vehicle, the charging pile and the road traffic, not only improves the modeling precision, but also significantly improves the calculation efficiency. This new modeling method can effectively reveal the dynamic and random interaction mechanism of the vehicle-pile-road, and provide more accurate performance evaluation and decision optimization support for logistics enterprises. Through a more precise model, the distribution efficiency, enterprise cost and customer satisfaction can be better evaluated, thereby providing more powerful support for logistics enterprises in distribution system evaluation and decision optimization.

[0010] In addition, the application precisely depicts the key dynamic and random interaction in the electric vehicle logistics distribution system by constructing a discrete event simulation model considering the vehicle-pile-road interaction, and provides effective underlying model support for the evaluation and decision optimization of large-scale systems. The model can output detailed performance indicators, including delivery time, energy consumption, charging time and cost, and provide a visual data analysis interface to facilitate users to view simulation results and make decision support. Based on the model, combined with heuristic algorithms, the electric vehicle logistics vehicle distribution path simulation optimization method can effectively balance the interests of enterprises, customers and the public, alleviate charging queue congestion and road traffic congestion by fully considering the interests of the public, improve the overall social benefit, and ensure the sustainable development of the system. In summary, the application has brought significant improvement in the evaluation, configuration and path optimization of the electric vehicle logistics distribution system, effectively solved the problem that the existing method cannot balance precision and efficiency, and has important practical application value. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the application. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0012] Figure 1 is a structural diagram of an electric vehicle logistics vehicle distribution path simulation optimization system in an embodiment of the application; Figure 2 is a flowchart of an electric vehicle logistics vehicle distribution path simulation optimization method in an embodiment of the application. DETAILED DESCRIPTION

[0013] The technical solutions in the embodiments of the application will be described clearly and completely in the following with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0014] In an embodiment, as shown in Figure 1 An electric vehicle logistics vehicle distribution path simulation optimization system is provided, comprising: A vehicle module 100 is used to simulate the dynamic behavior and performance characteristics of electric vehicle logistics vehicles, to realize dynamic scheduling and path optimization by updating the vehicle state in real time and interacting with the station module and the road module, and to ensure that the electric vehicle logistics vehicles can maintain optimal performance and effectively reduce energy consumption in different environments.

[0015] The road module 200 is used to simulate the driving environment of electric logistics vehicles during the delivery process, so as to accurately reflect the current road conditions by monitoring and updating basic parameters in real time, thereby directly affecting the driving efficiency and energy consumption of the vehicle.

[0016] The site module 300 is used to manage the charging, scheduling and task allocation of electric logistics vehicles. The types of site module include charging stations, demand points and distribution centers. Each type of site has specific functions and attributes to meet different delivery needs, including: Charging management: The charging station is responsible for meeting the charging needs of electric logistics vehicles to ensure that the vehicles can supplement energy in time during the delivery process. The charging station receives the arrival event of the vehicle and decides whether to accept the newly arrived vehicle according to the current state of the charging pile. If a charging pile is being occupied by other vehicles, the system will select other available charging piles according to the current state to optimize resource utilization.

[0017] Task allocation: The demand point is where customers place orders and is responsible for receiving and storing goods.

[0018] The distribution center is the hub of goods and is responsible for the scheduling of vehicles and the allocation of goods. The site module dynamically adjusts the charging strategy and task allocation according to the state of the electric logistics vehicle and the scheduled delivery demand, thereby reducing the charging waiting time and optimizing resource utilization.

[0019] State monitoring and event handling: The site module monitors the usage state of the charging pile and the number of queued vehicles in real time and handles events related to the interaction of electric logistics vehicles at various sites, such as arrival events, departure events, charging completion events and charging pile state change events. The handling of these events ensures that the site module can efficiently manage resources and schedule vehicles.

[0020] The road network module 400 serves as the connection hub of the simulation optimization system and is used to initialize and configure the site module, vehicle module and road module, providing a unified framework for each module to ensure that data and events between different modules can be effectively transmitted.

[0021] It can be understood that the simulation optimization system of the electric logistics vehicle delivery path is an electric logistics vehicle delivery path simulation optimization model based on an event-driven mechanism and an O2DES framework, which comprehensively considers the dynamic and random interaction between vehicles, charging piles and road traffic. The model is composed of site modules, vehicle modules and road modules, and is initialized and triggered by the road network module to realize dynamic interaction between modules. This design aims to provide support for efficient evaluation and optimization of electric logistics vehicle delivery systems.

[0022] In an embodiment, the vehicle module comprises: An initialization module is configured to initialize key attributes of each electric vehicle at the beginning of the simulation, including vehicle static attributes and vehicle dynamic attributes. The vehicle static attributes include vehicle number, load capacity, and vehicle charging power threshold of the electric vehicle. The vehicle dynamic attributes include current power, current load, whether the vehicle is going to charge, driving speed, driving time, total energy consumption of the vehicle, driving distance, and driving path of the electric vehicle.

[0023] A running and energy consumption management module is configured to start the electric vehicle running according to the preset delivery path and time window, dynamically calculate the energy consumption according to the driving distance, load, and road conditions during the driving of the electric vehicle, and update the battery power of the vehicle in real time.

[0024] An event monitoring and processing module is configured to monitor the dynamic behavior of the vehicle in real time, trigger corresponding vehicle events when the vehicle arrives at a station, needs to be charged, or encounters traffic congestion, and capture the vehicle events by the road network module and process them according to the event type. The corresponding processing includes a new dispatching plan or an adjustment of the charging strategy. The vehicle events include arrival events, charging events, unloading events, state update events, and traffic congestion events.

[0025] The arrival event is defined as an event triggered when the electric vehicle arrives at a target station. The time and location of the vehicle arrival are recorded, and the vehicle state is updated to "arrival". At this time, the availability of the charging pile is checked, and it is decided whether to let the vehicle charge or unload.

[0026] The charging event is defined as an event in which the vehicle starts charging. The charging start time is recorded, and the charging state is monitored. The expected charging completion time is calculated according to the charging rate of the charging pile.

[0027] The unloading event is defined as an event in which the vehicle completes the unloading operation. The vehicle state is updated, and the unloading time is recorded.

[0028] The state update event is defined as an event in which the state of the vehicle changes during driving. It includes battery power reduction, driving time update, etc. The vehicle delivery plan is adjusted in real time according to these changes to ensure that the charging station is found in time when the power is insufficient.

[0029] The traffic congestion event is defined as an event triggered when the vehicle encounters traffic congestion during driving. The vehicle path is re-planned according to the new road conditions to avoid delaying the delivery time.

[0030] As can be understood, the vehicle module can effectively simulate the dynamic behavior of the electric vehicle under the cooperation of the above-mentioned sub-modules, realize precise energy consumption management, event-driven dispatching decision, and dynamic path optimization.

[0031] In an embodiment, the road module comprises: a parameter configuration module for initializing the basic parameters of the road at the beginning of the simulation, the basic parameters including road static attributes, road dynamic attributes, initial traffic flow and road condition information.

[0032] The road static attributes include road length, starting node of the road, terminal node of the road, road capacity; the road dynamic attributes include number of vehicles on the road.

[0033] An attribute updating module for continuously updating the road dynamic attributes according to the number of logistics vehicles entering the road. These updates ensure that the road module can accurately reflect the current road conditions.

[0034] Understandably, the road module, in cooperation with the above-mentioned sub-models, can effectively simulate the driving environment of the electric logistics vehicle in the delivery process, provide accurate road condition information for the vehicle module, and thus support the dynamic scheduling and path optimization of the vehicle. Among them, the road event describes the change of the driving environment of the electric logistics vehicle in the delivery process, mainly including vehicle arrival event and traffic condition change event.

[0035] In an embodiment, the station module comprises: An attribute initialization module for initializing the static attributes and dynamic attributes of all stations at the beginning of the simulation, the static attributes including vehicle Id index, charging station charging rate, station demand, service time and location coordinates, and the dynamic attributes including station list, demand point list, station downstream charging station list, station downstream path, warehouse vehicle list, charging pile usage state, number of queued vehicles, current charging vehicle list and charging time.

[0036] An event processing module for receiving the arrival event of the vehicle when the electric logistics vehicle arrives at the corresponding station with charging demand, updating the current state of the station, and recording the time of vehicle arrival; at this time, the event processing checks the service capacity of the station, including the number and usage state of charging piles, to decide whether to admit the newly arrived vehicle.

[0037] Among them, the station event refers to the event related to the interaction of the electric logistics vehicle with various stations (such as charging stations, demand points, etc.), including: Arrival event: the event of the electric logistics vehicle arriving at a certain station. Update the current state of the station and record the time of vehicle arrival. Check the service capacity of the station (such as the number and usage state of charging piles), and decide whether to admit the newly arrived vehicle.

[0038] Departure event: the event of the vehicle leaving the station after completing charging or unloading. Update the vehicle state and record the departure time. Plan the delivery path of the vehicle according to the electric quantity of the vehicle and the demand of the next task.

[0039] Charge completion event: The event when a vehicle finishes charging at a charging station. Records the time of charging completion, updates the vehicle's state of charge, and notifies the vehicle to prepare for departure.

[0040] Charging station state change event: The event when the usage state (idle or occupied) of a charging station changes. Monitors the state of each charging station and adjusts the vehicle's charging strategy based on state changes.

[0041] The definitions of these events clearly describe the key behaviors and state changes of electric delivery vehicles during simulation, providing a foundation for dynamic management and optimization of the system.

[0042] Charging management module: If a vehicle needs to charge, this module selects a usable charging station based on the current state of the charging station and starts the charging process. During charging, the start time and estimated completion time are recorded. If a charging station is being occupied by another vehicle, other available charging stations are selected based on the current state to optimize resource utilization.

[0043] Task scheduling and resource optimization module: dynamically adjusts task allocation based on the state of charge of electric delivery vehicles and the scheduled delivery requirements, including triggering a departure event after a vehicle completes charging or unloading. This module updates the vehicle's state and records the departure time. Based on the vehicle's state of charge and the requirements of the next task, the vehicle's delivery route is planned to enable it to continue performing delivery tasks after completing charging. At the same time, the module monitors charging station state change events and adjusts the vehicle's charging strategy based on state changes.

[0044] It is understandable that the site module, in cooperation with the above-mentioned sub-models, through the closed-loop mechanism of "attribute initialization → event response → resource scheduling → dynamic optimization", significantly improves the resource utilization efficiency and decision-making accuracy of the electric delivery vehicle distribution system.

[0045] In an embodiment, the road network module includes: Network initialization module: initializes the entire distribution network at the start of simulation and generates a distribution network with a preset distribution of sites, connection of roads, and traffic flow settings based on the parameters. The distribution network provides a unified framework for each module.

[0046] Time management and event scheduling module: manages simulation time through the O2DES framework, responsible for event scheduling and triggering. In the event triggering mechanism, interactions between modules are achieved through events, thereby achieving efficient data and event transmission.

[0047] When an electric delivery vehicle arrives at a certain site, the module triggers an arrival event and determines the next step event for the vehicle based on the current site type.

[0048] A state updating and coordinating module is configured to update the states of the modules in real time to ensure the accuracy and consistency of the simulation process; through the continuous state monitoring and updating of the module, the operation of each module is coordinated to ensure the efficient and stable operation of the entire simulation system.

[0049] Understandably, the road network, in cooperation with the above-mentioned sub-models, not only improves the operation efficiency of the system, but also enhances the response capability to dynamic changes, providing a solid foundation for the optimization of the electric animal logistics vehicle distribution system.

[0050] In an embodiment, the initialization and construction process of the road network is as follows: First, in the model initialization link, all station modules, vehicle modules and road modules are initialized. This step ensures that the basic structure and function of each module are established, providing a foundation for subsequent simulation operations.

[0051] Next, according to the input data, the initial values of the static attributes and dynamic attributes of each module are determined. Static attributes refer to parameters that do not change during the simulation process, such as the location of the station, the length of the road, etc.; dynamic attributes refer to parameters that change during the simulation process, such as the current power of the vehicle, the traffic flow of the road, etc.

[0052] Then, enter the road network construction phase. First, connect the station and the path to ensure that each station can be connected to other stations through the path to form a complete network structure. This step is the basis of building the road network, ensuring that vehicles can move between stations. Next, connect the path and the node, i.e. determine the downstream node of the path and the nearby station. This step further refines the connection relationship of the road network, clearly defines the moving direction of the vehicle on the path and the possible stop station, providing detailed network information for the vehicle path planning. Finally, initialize the vehicle position and place the vehicle in the distribution center. This step sets the initial state for the simulation process, ensuring that the vehicle starts from the distribution center and begins to perform the distribution task. Through this series of steps, the road network module completes the initialization and construction, providing a solid foundation for the operation of the entire electric animal logistics vehicle distribution path simulation optimization system.

[0053] The present application also provides a simulation optimization method for electric animal logistics vehicle distribution path, as shown in Figure 2 The method comprises the following steps S1-S3: S1, a discrete event simulation optimization model of the electric animal logistics vehicle distribution system is constructed, and a simulation program is designed. The discrete event simulation optimization model of the electric animal logistics vehicle distribution system is also the simulation optimization model of the electric animal logistics vehicle distribution.

[0054] In an embodiment, the step S1 comprises steps S101-S102: S101. Using the Visual Studio 2022 development environment and the C# programming language, a simulation optimization model for the delivery route of electric logistics vehicles is built based on the O2DES simulation framework. This model includes a station module, a vehicle module, and a road module. The road network module is used for initialization and event-triggered calls to achieve dynamic interaction between the modules.

[0055] S102. A simulation program based on the O2DES simulation framework is designed and developed using the C# programming language. This simulation program includes: Scheduling module: Automatically generates delivery plans based on delivery needs and the status of electric logistics vehicles, ensuring the rationality and efficiency of task allocation.

[0056] Charging module: Select the best charging station based on the electric logistics vehicle's battery status and the availability of charging piles to reduce charging waiting time and improve resource utilization.

[0057] Route planning module: Optimizes delivery routes by combining dynamic and random traffic information.

[0058] S2. Establish the objective function and constraints for optimizing the delivery route of electric logistics vehicles.

[0059] In one embodiment, step S2 includes steps S201-S204: S201. Establish an objective function for minimizing overall operating costs. The mathematical expression of the objective function is as follows: (1) (2) (3) (4) (5) (6) in, This represents the total operating cost of the logistics and distribution process. Indicates energy consumption cost, This represents the cost per unit of electricity. This represents the total energy consumption of all vehicles, and the penalty cost for violating the time window. This represents the penalty cost per unit of time. This indicates the total lateness time of all vehicles. This represents the fixed expenses for vehicles. This indicates the cost of a single electric logistics vehicle. Indicates the total number of delivery vehicles; This represents the driver's total salary. driver wage per unit time, total delivery time of the vehicle; total cost of charging piles, average daily construction cost of a single charging pile, total number of all charging piles.

[0060] S202, setting a constraint condition, the constraint condition including the following conditions: (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) wherein the constraint condition (7) indicates that a demand point can be served only once, the constraint condition (8) indicates node flow conservation, the constraint condition (9) indicates that each electric vehicle starts from the distribution center and must return to the distribution center, the constraint conditions (10) to (11) indicate the load limit of the vehicle, and the total demand transported by each vehicle does not exceed the vehicle loading capacity, the constraint conditions (12) to (13) indicate that the vehicle must wait until the time window opens when the vehicle arrives at the demand point earlier than the start time window, and the late arrival time is recorded when the vehicle arrives late, the constraint condition (14) indicates that the vehicle does not consume electricity when unloading, the constraint conditions (15) to (16) indicate the vehicle range constraint, and the constraint (17) is a decision variable description, when the vehicle has visited link(i,j), is 1, otherwise 0; when the vehicle serves the customer point , is 1, otherwise 0.

[0061] Table 1 Model parameter definition

[0062]

[0063] S3, based on the objective function and constraint conditions, an initial solution is generated by Floyd algorithm and saving mileage method, and then adaptive large neighborhood search algorithm ALNS is used for distribution path optimization. Understandably, the initial solution is generated by Floyd algorithm and saving mileage method, and then optimized by adaptive large neighborhood search algorithm (ALNS), and by selecting and adjusting the weight of each destruction and repair operator in each iteration, the optimal solution is finally obtained in the algorithm.

[0064] In an embodiment, the step S3 comprises steps S301-S302: S301, an initial solution is generated by Floyd algorithm and saving mileage method. In an embodiment, the step S301 comprises steps S3011-S3014: S3011, the shortest path of all node pairs is calculated, and Floyd algorithm is used to process the distance matrix so that the path between all node pairs is the shortest.

[0065] For each node , starting from 0, all nodes are traversed, and the algorithm checks all node pairs , if the path from to can be shortened through , that is, dist[i][k] + dist[k][j] is less than the current dist[i][j], then dist[i][j] is updated to this smaller value.

[0066] Repeat until the distance of all node pairs reaches the minimum value, and finally the matrix contains the shortest path distance between any two nodes.

[0067] S3012, a saving value list is generated and sorted: The saving value of all customer point pairs is calculated by saving mileage method, for each pair of customer points, and , where , and and are not warehouses, the larger the saving value is, the more advantageous the path of merging and= is, that is, more distance is saved.

[0068] All calculated saving values triples It is stored in a list, and this list is sorted by savings value. Sort in descending order.

[0069] S3013, Initialization Path: For each customer point, node 1 to... Create an independent initial path, with each path having the following structure: This indicates a journey starting from the warehouse and visiting the customer's location. Then return to the warehouse.

[0070] Simultaneously, the remaining capacity of each route is recorded, i.e., the amount of cargo the vehicle can still carry after serving that customer; the route data is stored in a dictionary, categorized by customer point. As keys to facilitate subsequent lookups, the value corresponding to each key contains the path sequence and the remaining capacity.

[0071] S3014, Merge Paths; by Savings Value Sort the paths from largest to smallest, merge them sequentially, and check vehicle capacity constraints to gradually build an initial path that meets the capacity limit.

[0072] S302. The Adaptive Large Neighborhood Search (ALNS) algorithm is used to achieve efficient search of the solution space by dynamically adjusting the weights of the destruction and repair operators. ALNS is a highly efficient metaheuristic algorithm designed to solve large-scale combinatorial optimization problems (such as vehicle routing, scheduling, and bin packing problems). It balances exploration and exploitation by dynamically adjusting the search strategy (destruction and repair operations), thereby efficiently finding high-quality solutions in complex solution spaces. ALNS uses four destruction operators to optimize path planning: random removal, worst-case removal, relevance removal, and time-window removal. The random removal operator randomly selects a certain proportion of demand points for removal without discrimination, introducing randomness to avoid the algorithm getting trapped in local optima and increasing the diversity of solutions. The worst-case removal operator focuses on identifying the demand points that have the greatest impact on the path and removes them first, thereby accelerating the convergence process and finding better solutions faster. The relevance removal operator selects removal targets based on the strength of the connection between demand points. For example, demand points that are close together and have a large demand volume are prioritized for removal because they have a greater impact on the path structure. Finally, the time window removal operator prioritizes demand points with tight time windows or those that have large time window conflicts with other demand points to optimize time management. The synergistic effect of these four disruptive operators enables the algorithm to search for high-quality solutions more efficiently in a complex solution space, while maintaining solution diversity and optimization efficiency.

[0073] In one embodiment, step S302 includes steps S3021-S3024: S3021. Set the input initial solution to the current optimal solution; initialize the weights of each destruction operator and repair operator, and assign equal initial weight values; set the total iteration count counter and the optimal solution unimproved count counter, and define the termination condition parameters as the maximum number of iterations and the maximum number of times the target value has not changed.

[0074] S3022. Enter the main loop: The loop continues until one of two termination conditions is met, namely, reaching the preset maximum number of iterations, or the number of consecutive times the optimal solution remains unimproved reaches a preset threshold; in each loop iteration, a destructive operator and a repair operator are selected using a roulette wheel selection method based on the current weight values ​​of each operator; the destructive operator removes some client points from the current solution. Repairing customer points where operators will be removed Reinsert it into the path in a better way to generate a new candidate solution.

[0075] S3023. After generating a new candidate solution, evaluate the quality of the new candidate solution and decide whether to adopt the new candidate solution according to the acceptance criteria. The new solution is then described as the new solution.

[0076] The acceptance criterion adopts a simulated annealing strategy, specifically: if the new solution is better than the current solution, it is always accepted; if the new solution is worse, it is accepted with a certain probability, and this probability of acceptance gradually decreases as the iteration progresses. If the new solution is accepted, the current solution is replaced.

[0077] S3024. Update the optimal solution: Check whether the new solution is better than the historical optimal solution through the optimization algorithm. If the objective function value of the new solution is better than the historical optimal solution, then update the historical optimal solution and reset the unimproved counter of the optimal solution.

[0078] The operator's weight is updated based on its performance in this iteration; if the operator participates in generating a solution and is accepted, its weight is increased; if the quality of the generated solution is poor, its weight is decreased.

[0079] When the main loop terminates, the algorithm returns the optimal solution found during the search process as the final result.

[0080] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0081] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A simulation optimization system for delivery routes of electric logistics vehicles, characterized in that, include: The vehicle module is used to simulate the dynamic behavior and performance characteristics of electric logistics vehicles. By updating the vehicle status in real time and interacting with the station module and road module, dynamic scheduling and route optimization are achieved. The road module is used to simulate the driving environment of electric logistics vehicles during delivery, so as to accurately reflect the current road conditions by monitoring and updating basic parameters in real time. The station module is used to manage the charging, scheduling, and task allocation of electric logistics vehicles. The types of station modules include charging stations, demand points, and distribution centers. The road network module, as the connection hub of the simulation optimization system, is used to initialize and configure the station module, vehicle module, and road module, providing a unified framework for each module and ensuring that data and events can be effectively transmitted between different modules.

2. The simulation and optimization system for electric logistics vehicle delivery routes according to claim 1, characterized in that, The vehicle module includes: The initialization module is used to initialize the key attributes of each electric logistics vehicle at the start of the simulation. The key attributes include vehicle static attributes and vehicle dynamic attributes. The vehicle static attributes include the vehicle number, load, and vehicle charging power threshold of the electric logistics vehicle; the vehicle dynamic attributes include the current battery level, current load, whether the vehicle is heading to the charging station, driving speed, driving time, total energy consumption, driving distance, and driving route of the electric logistics vehicle. The operation and energy management module is used to start the operation of electric logistics vehicles according to the preset delivery route and time window. During the operation of electric logistics vehicles, the module dynamically calculates energy consumption based on the driving distance, load and road conditions, and updates the vehicle's battery charge in real time. The event monitoring and processing module is used to monitor vehicle dynamic behavior in real time. When a vehicle arrives at a station, needs to charge, or encounters traffic congestion, it triggers the corresponding vehicle event. The vehicle event is captured by the road network module and processed accordingly based on the event type. The corresponding processing includes new scheduling plans or adjustments to charging strategies. The vehicle events include arrival events, charging events, unloading events, status update events, and traffic congestion events.

3. The simulation and optimization system for electric logistics vehicle delivery routes according to claim 2, characterized in that, The road module includes: The parameter configuration module is used to initialize the basic parameters of the road at the start of the simulation. The basic parameters include the road's static properties, road dynamic properties, initial traffic flow, and road condition information. Static road attributes include road length, starting node, ending node, and capacity; dynamic road attributes include the number of vehicles on the road. The attribute update module is used to continuously update the dynamic attributes of the road based on the number of logistics vehicles entering the road.

4. The simulation and optimization system for electric logistics vehicle delivery routes according to claim 3, characterized in that... The site module includes: The attribute initialization module is used to initialize the static and dynamic attributes of all stations at the start of the simulation. Static attributes include vehicle ID index, charging station charging rate, station demand, service time and location coordinates. Dynamic attributes include station list, demand point list, downstream charging station list, downstream path of station, warehouse vehicle list, charging pile usage status, number of vehicles in queue, current charging vehicle list and charging time. The event handling module is used to receive the arrival event of an electric logistics vehicle when it arrives at the corresponding station due to charging demand, update the current status of the station, and record the arrival time of the vehicle. At this time, the event handling module checks the service capacity of the station and decides whether to accept the newly arrived vehicle. The service capacity includes the number and usage status of charging piles. The charging management module selects an available charging station based on the current status of the charging station when a vehicle needs to be charged, and starts the charging process. During the charging process, it records the charging start time and the estimated charging completion time. If a charging station is being used by another vehicle, it selects another available charging station based on the current status to optimize resource utilization. The task scheduling and resource optimization module is used to dynamically adjust task allocation based on the electric logistics vehicle's battery status and predetermined delivery needs. This includes triggering a departure event after the vehicle completes charging or unloading; updating the vehicle status and recording the departure time; planning the vehicle's delivery route based on the vehicle's battery level and the needs of the next task so that the vehicle can continue to perform delivery tasks after charging is completed; and monitoring charging pile status change events and adjusting the vehicle's charging strategy accordingly.

5. The simulation and optimization system for electric logistics vehicle delivery routes according to claim 1, characterized in that, The road network module includes: The network initialization module is used to initialize the entire delivery network at the start of the simulation and generate the distribution of stations, road connections, and traffic flow settings of the delivery network according to preset parameters. The delivery network provides a unified framework for all modules. The time management and event scheduling module is used to manage simulation time through the O2DES framework. It is responsible for event scheduling and triggering. In the event triggering mechanism, the interaction between modules is realized through events, thereby realizing the transmission of data and events. When an electric logistics vehicle arrives at a station, this module triggers an arrival event and determines the vehicle's next event based on the current station type. The status update and coordination module is used to update the status of each module in real time to ensure the accuracy and consistency of the simulation process; through continuous status monitoring and updates, this module coordinates the operation of each module.

6. A simulation optimization method for electric logistics vehicle delivery routes, implemented using the simulation optimization system for electric logistics vehicle delivery routes as described in any one of claims 1-5, characterized in that, The simulation optimization method for the delivery route of the electric logistics vehicle includes: S1. Construct a simulation optimization model for the delivery route of electric logistics vehicles and design a simulation program; S2. Establish the objective function and constraints; S3. Based on the objective function and constraints, an initial solution is generated using the Floyd algorithm and the mileage saving method, and then optimized using the adaptive large neighborhood search algorithm ALNS.

7. The simulation optimization method for electric logistics vehicle delivery routes according to claim 6, characterized in that, Step S1 includes: S101. Using the Visual Studio 2022 development environment and the C# programming language, a simulation optimization model for the delivery route of electric logistics vehicles is built based on the O2DES simulation framework. The model includes a station module, a vehicle module, and a road module. The road network module is used for initialization and event-triggered calls to realize dynamic interaction between the modules. S102. A simulation program based on the O2DES simulation framework is designed and developed using the C# programming language. This simulation program includes: Scheduling module: Automatically generates delivery plans based on delivery needs and the status of electric logistics vehicles, ensuring the rationality and efficiency of task allocation; Charging module: Select the best charging station based on the electric logistics vehicle's battery status and the availability of charging piles to reduce charging waiting time and improve resource utilization. Route planning module: Optimizes delivery routes by combining dynamic and random traffic information.

8. The simulation optimization method for electric logistics vehicle delivery routes according to claim 7, characterized in that, Step S2 includes: S201. Establish an objective function for minimizing overall operating costs. The mathematical expression of the objective function is as follows: （1） （2） （3） （4） （5） （6） in, This represents the total operating cost of the logistics and distribution process. Indicates energy consumption cost, This represents the cost per unit of electricity. This represents the total energy consumption of all vehicles, and the penalty cost for violating the time window. This represents the penalty cost per unit of time. This indicates the total lateness time of all vehicles. This represents the fixed expenses for vehicles. This indicates the cost of a single electric logistics vehicle. Indicates the total number of delivery vehicles; This represents the driver's total salary. This represents the driver's wage per unit of time. Indicates the total delivery time of the vehicle; This represents the total cost of the charging station. This indicates the average daily construction cost of a single charging station. This represents the total number of all charging stations; S202. Set constraints, which include the following conditions: （7） （8） （9） （10） （11） （12） （13） （14） （15） （16） （17） （18） Among them, constraint (7) indicates that a demand point can only be served once, constraint (8) indicates that node flow is conserved, constraint (9) indicates that each electric logistics vehicle departs from the distribution center and must return to the distribution center, constraints (10) to (11) indicate the load limit of the vehicle, and the total demand transported by each vehicle does not exceed the vehicle's loading capacity, constraints (12) to (13) indicate that if the vehicle arrives at the demand point earlier than the start time window, it must wait until the time window opens, and if it arrives later, the late arrival time is recorded, constraint (14) indicates that the vehicle does not consume electricity when unloading, constraints (15) to (16) indicate the vehicle's range constraint, and constraint (17) indicates the 0-1 variable description.

9. The simulation optimization method for electric logistics vehicle delivery routes according to claim 8, characterized in that, Step S3 includes: S301. Generate an initial solution using the Floyd algorithm and the mileage-saving method, including: S3011. Calculate the shortest path for all node pairs, and use the Floyd algorithm to process the distance matrix so that the path between all node pairs is the shortest. For each node Starting from 0, traverse all nodes, and the algorithm checks all node pairs. If from arrive The path can be accessed through If the value of dist[i][k] + dist[k][j] is less than the current value of dist[i][j], then update dist[i][j] to the smaller value. Repeat this process until the distance between all pairs of nodes reaches its minimum. The matrix contains the shortest path distance between any two nodes; S3012. Generate a list of savings values ​​and sort them; All customer point pairs are calculated using the mileage saving method. The savings, for each pair of customer points, and ,in, ,and and None of it is for the warehouse, saving value The larger the value, the more likely it is to be merged. and The more advantageous the path, the more distance is saved; All calculated savings triples It is stored in a list, and this list is sorted by savings value. Sort in descending order; S3013, Initialize the path; for each customer point, node 1 to... Create an independent initial path, with each path having the following structure: This indicates a journey starting from the warehouse and visiting the customer's location. Then return to the warehouse; Simultaneously, the remaining capacity of each route is recorded, i.e., the amount of cargo the vehicle can still carry after serving that customer; the route data is stored in a dictionary, categorized by customer point. As keys to facilitate subsequent lookups, the value corresponding to each key contains the path sequence and the remaining capacity; S3014, Merge Paths; by Savings Value Sort the paths from largest to smallest, merge them sequentially, and check vehicle capacity constraints to gradually build an initial path that meets the capacity limit. S302. By using the adaptive large neighborhood search algorithm ALNS, the weights of the destruction and repair operators are dynamically adjusted to achieve efficient search of the solution space. S3021. Set the input initial solution to the current optimal solution; initialize the weights of each destruction operator and repair operator, and assign equal initial weight values; set the total iteration count counter and the optimal solution not improved counter, and define the termination condition parameters as the maximum number of iterations and the maximum number of times the target value has not changed; S3022. Enter the main loop, which continues until one of two termination conditions is met, namely, reaching the preset maximum number of iterations, or the number of consecutive times the optimal solution remains unimproved reaches a preset threshold. In each iteration, based on the current weight values ​​of each operator, a destructive operator and a repair operator are selected using a roulette wheel selection method. The destructive operator removes a portion of client points from the current solution. Repairing customer points where operators will be removed Reinsert it into the path in a better way to generate a new candidate solution; S3023. After generating a new candidate solution, evaluate the quality of the new candidate solution and decide whether to adopt the new candidate solution according to the acceptance criteria, and describe it as a new solution. The acceptance criterion adopts a simulated annealing strategy, specifically: if the new solution is better than the current solution, it is always accepted; if the new solution is worse, it is accepted with a certain probability, and this probability of acceptance gradually decreases as the iteration progresses. If the new solution is accepted, the current solution is replaced. S3024. Update the optimal solution; check whether the new solution is better than the historical optimal solution through the optimization algorithm. If the objective function value of the new solution is better than the historical optimal solution, update the historical optimal solution and reset the unimproved counter of the optimal solution. The weights of operators are updated based on their performance in this iteration; if an operator participates in generating a solution and is accepted, its weight is increased; if the quality of the generated solution is poor, its weight is decreased. When the main loop terminates, the algorithm returns the optimal solution found during the search process as the final result.

Images