An embedded cost estimation method and system for electric truck route and charging coordination

Abstract

The application discloses an embedded cost estimation electric truck path and charging coordination method and system, relates to the field of combinatorial optimization and operations research, and constructs a three-component cost estimation function including distance cost, time cost and energy cost, wherein the energy cost component explicitly introduces the space-time distribution characteristics of node electricity price and the path-level arbitrage potential index, is used for quickly evaluating the influence of path structure adjustment on the overall cost without solving the charging and discharging optimization sub-problems; and on the basis, the cost estimation mechanism is embedded into the adaptive large neighborhood search framework, a set of destruction and repair operators matched with the electric freight vehicle fleet operation structure is designed, and the search process is guided to realize the cooperative optimization between path planning and charging decision. Through the deep integration of cost estimation and search decision, the application significantly reduces the calculation complexity while maintaining the solution quality, so that the city-level large-scale electric freight vehicle path-charging coordination scheduling problem which is difficult to solve by the traditional method can obtain a high-quality solution within a reasonable time, and has good engineering applicability and popularization value.

Description

Technical Field

[0001] This invention relates to the fields of combinatorial optimization and operations research, specifically to a method and system for coordinating the routing and charging of electric freight vehicles with embedded cost estimation. Background Technology

[0002] The electric freight truck routing problem is an important variant of the Vehicle Routing Problem (VRP), with wide applications in logistics, urban freight, and cold chain transportation. Compared to the traditional VRP, electric freight truck routing requires additional consideration of factors such as charging station selection, charging schedules, battery state of charge (SOC) constraints, and driving range limitations, significantly increasing the problem's complexity. The difficulty is further increased when considering charging / discharging optimization under time-varying nodal electricity prices and arbitrage opportunities for vehicles to feed electricity back into the grid (V2G).

[0003] From a problem structure perspective, the route-charging coordination and scheduling problem of electric freight fleets can be modeled as a two-level optimization problem: the upper-level problem involves decision-making regarding route selection and charging / discharging arrangements from the perspective of logistics operators, while the lower-level problem involves market clearing and the formation of nodal marginal prices from the perspective of distribution network operators. This two-level structure results in complex interdependencies between the upper-level decision variables (charging / discharging quantities) and the lower-level decision variables (nodal prices), making it difficult for traditional hierarchical solution methods to obtain the globally optimal solution. Current solution methods have the following shortcomings: First, exact algorithms have poor scalability. While exact solutions based on Mixed Integer Linear Programming (MILP) can guarantee a globally optimal solution, their computational complexity increases exponentially with the problem size. For real-world problems, exact algorithms often take hours or even days to solve, making it difficult to meet the efficiency requirements of practical operations. This is especially true when considering time-varying electricity prices and V2G functionality; the additional variables and constraints introduced by time-period discretization cause the problem size to expand dramatically, making exact solutions even more difficult.

[0004] Second, traditional metaheuristic algorithms lack charging awareness. Metaheuristic algorithms such as Adaptive Large Neighborhood Search (ALNS), Genetic Algorithm (GA), and Ant Colony Optimization (ACO) are widely used to solve various vehicle routing problems and have good scalability. However, the design of existing algorithms mainly focuses on path structure optimization, and their destruction / repair operators do not fully consider the special characteristics of charging decisions. For example, the traditional worst-case removal operator only considers distance and time factors, ignoring the impact of node charging capacity on cost; the greedy insertion operator is also usually based on the criterion of nearest distance or earliest time, and cannot effectively respond to spatiotemporal differences in electricity prices.

[0005] Third, the computational cost of cost assessment is too high. During the iterative process of heuristic algorithms, the objective function value of candidate solutions needs to be evaluated after each removal or insertion operation. For the electric freight fleet scheduling problem, accurately assessing the total cost of a route structure requires solving a subproblem involving charge-discharge optimization, which is itself a mixed integer programming problem. When the algorithm iterates a large number of times or the problem size is large, the computational burden of repeatedly solving subproblems becomes excessive, severely impacting algorithm efficiency.

[0006] Fourth, the spatiotemporal response capability to electricity prices is insufficient. Existing heuristic algorithms typically only consider distance and time costs in their cost estimation functions (or fitness functions), failing to effectively incorporate the spatiotemporal distribution characteristics of node electricity prices. This results in the search process being unable to perceive spatial (different node prices) and temporal (different time periods) differences in charging costs, making it difficult to guide the search towards the optimal electricity cost. In V2G scenarios, ignoring arbitrage opportunities can lead to significant economic losses.

[0007] Fifth, the operator design lacks specificity. The electric freight fleet scheduling problem has unique structural characteristics: the path consists of multiple travel segments, each always ending at a depot node; charging decisions are tightly coupled with dwell time; and energy constraints limit the continuous travel distance. The existing ALNS algorithm's general destroy / repair operator design fails to fully utilize these structural characteristics, limiting the algorithm's search efficiency and solution quality.

[0008] In summary, the aforementioned shortcomings of existing technologies lead to the following fundamental technical problems: in the route and charging coordination and scheduling problem of electric freight fleets, how to achieve efficient solution of large-scale instances while ensuring solution quality, and at the same time enable the search process to effectively perceive and respond to the spatiotemporal distribution characteristics of node electricity prices. Summary of the Invention

[0009] To address the shortcomings mentioned in the background art, the present invention aims to provide a method and system for coordinating the routing and charging of electric freight vehicles with embedded cost estimation.

[0010] Firstly, the objective of this invention can be achieved through the following technical solution: a method for coordinating the routing and charging of electric freight vehicles with embedded cost estimation, the method comprising the following steps: Obtain input problem data, which includes traffic network topology information, station node set, customer node set, independent charging station node set, node distance matrix, fleet parameters, customer demand parameters, service time window, and spatiotemporal node electricity price and cost coefficient of each charging node in each time period; A three-component cost estimation function is pre-constructed, wherein the three-component cost estimation function is obtained by inputting problem data based on the customer-interpolated marginal cost estimation method; the three-component cost estimation function includes a distance cost estimation component, a time cost estimation component, and an energy cost estimation component; The initial solution is constructed based on the three-component cost estimation function, generating an initial vehicle routing scheme. The CE-ALNS main loop is then iterated based on the initial vehicle routing scheme. After multiple iterations to the preset number, the complete scheduling scheme corresponding to the globally optimal path structure is output.

[0011] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the distance cost estimation component reflects the increase in detour distance caused by the insertion of customers, the time cost estimation component reflects the cost increase caused by additional travel time and service time, and the energy cost estimation component explicitly incorporates the spatiotemporal distribution characteristics of nodal electricity prices and the arbitrage potential index; The expression for the three-component cost estimation function is as follows: The distance cost estimation component is as follows: In the formula, c For customers to be inserted, k For the target path, pos The insertion position, i and j These are the predecessor and successor nodes at the insertion position, respectively. For the node i To the customer c distance, To the customer c To the node j distance, For the node i Direct to node j distance, Cost coefficient per unit distance; The time cost estimation component is: In the formula, v The average speed of the vehicle. q c For customers c The demand, Service volume per unit time This is the unit time cost coefficient; if the path k If a time window violation has already occurred, a timeout penalty will be added to that component. The energy cost estimation components are: In the formula, To insert customers c The resulting increase in energy demand for path k Path after insertion k The lowest time-of-use electricity price at charging nodes along the route. The change in the arbitrage potential index caused by the insertion. This is the maximum battery capacity. Path after insertion k Energy demand, Path before insertion k Arbitrage potential index.

[0012] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the calculation formula for the arbitrage potential index is as follows: Define the minimum time-period electricity price for rechargeable nodes along path k as: In the formula, For path k The set of rechargeable nodes traversed. T For time period sets, For nodes s During the period t Electricity price; The arbitrage potential index is defined as the average time-period electricity price difference between rechargeable nodes along the route: Construct a path based on the lowest electricity price and the arbitrage potential index. k Charging cost estimation model: In the formula, For path k Energy requirements, This represents the cost of meeting energy demand at the lowest possible electricity price. This represents the potential profit from arbitrage between nodes using the remaining battery capacity.

[0013] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the process of constructing a greedy initial solution based on the three-component cost estimation function, as follows: Initialize an empty path for each vehicle, iteratively selecting the customer and location with the lowest insertion cost from unassigned customers and inserting them until all customers are assigned, thus obtaining the initial path structure. Design a problem-customized set of destruction operators, including a random removal operator that uniformly and randomly selects customers for removal, a worst-case removal operator that selects the customer with the highest marginal removal benefit for removal, a greedy station replacement operator that randomly replaces station nodes on the path and removes all customers on that path, and a random trip removal operator that randomly selects trip segments and removes all customers within them. Design a problem-customized set of repair operators, including a random repair operator that randomly selects feasible locations for customers to be inserted, a full-cost greedy repair operator that uses a complete three-component cost estimation function to select the location with the lowest estimated cost for insertion, a transport-oriented greedy repair operator that uses only distance and time cost components for evaluation, and an energy-oriented greedy repair operator that uses only energy cost components for evaluation. Implement adaptive operator weight updates based on a roulette wheel selection and reward mechanism, where the operator selection probability is proportional to the weight, and differentiated rewards are given based on the quality of candidate solutions, and the weights are updated accordingly. Employ a simulated annealing acceptance criterion, setting an effective temperature parameter and an acceptance probability calculation formula, and setting a maximum number of iterations and a number of consecutive iterations without improvement as termination conditions.

[0014] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the method for calculating the marginal removal cost of the worst removal operator is as follows: Among them, the distance item is for removing customers. c The resulting reduction in distance costs: The time item is for removing customers. c The reduction in time cost resulting from: Charging costs are calculated separately based on whether the customer has access to charging facilities: If the customer c It has charging facilities, and the charging items are: In the formula, For vehicles in the current plan k In customers c time period t The amount of charge, To remove customers cThe resulting energy gap This refers to the highest electricity price during the designated time period for the station nodes in this route. If the customer c There are no charging facilities; the charging options are: In the formula, Energy consumption per unit distance; Worst removal operator selection removal cost Minimum specified ratio The customer removed it; The execution steps for the greedy station replacement operator are as follows: Step 1: Randomly select a path uniformly from all paths in the current solution. ; Step 2: Identify the path All station nodes, including the starting station, intermediate return station and the terminating station; Step 3: Assemble at the station D A new station, different from the current station, will be randomly selected. d' ; Step 4: Path Replace all station nodes with the new station d' ; Step 5: Set the path Remove all customer nodes from the list and add them to the set of customers to be inserted. The execution steps of the random route removal operator are as follows: Step 1: Decompose each path in the current solution into a sequence of travel segments, where a travel segment is defined as the sequence of customer nodes between two adjacent visits to a site node; Step 2: Count all travel segments for all paths and construct a travel segment set; Step 3: Randomly select a travel segment from the set of travel segments; Step 4: Remove all customer nodes in the selected travel segment and add them to the customer set to be inserted.

[0015] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the process of executing the CE-ALNS main loop iteration based on the initial vehicle routing scheme, as follows: Execute the CE-ALNS main loop. In each iteration, select the destroy operator and the repair operator in a roulette wheel manner. Apply the destroy operator to remove some customers and apply the repair operator to reinsert the removed customers. Solve the path-fixed subproblem for the candidate path structure to evaluate the true target value. Decide whether to accept the candidate solution according to the acceptance criterion. Update the global optimal solution and operator weights. Check the termination condition. Output the complete scheduling scheme corresponding to the global optimal path structure. The complete scheduling scheme corresponding to the global optimal path structure includes the access node sequence of each vehicle, the arrival and departure times of each node, changes in cargo loading, battery state of charge trajectory, and charging and discharging power of each time period and node.

[0016] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the set of repair operators includes: Random Repair Operator RanR: For each client to be inserted c Iterate through all locations along all paths, collect a set of locations that pass the feasibility test, and then randomly select a location from this set to place the customer. c insert; Full-cost greedy repair operator GR-ALL: using the full three-component cost estimation function Evaluate the estimated interpolation costs of all feasible locations and select the location with the lowest estimated cost to serve the customer. c insert; Transport-oriented greedy repair operator GR-TRAN: using only distance and time cost components The evaluation prioritizes reducing driving distance and operating time; the energy-oriented greedy repair operator GR-ELEC uses only the energy cost component. The evaluation prioritizes optimizing charging costs and arbitrage opportunities. Among them, the four repair operators in the repair operator set are used in a complementary manner during the search process. The full-cost greedy repair pursues the overall optimality, the transportation-oriented and energy-oriented repairs focus on different cost components, and the random repair ensures the diversity of the search.

[0017] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: the adaptive operator weight update mechanism includes: Initialize and destroy operator weight vector and repair operator weight vector Each component is initialized to 1; In each iteration, the destroy operator and the repair operator are selected respectively using a roulette wheel method. The probability of selection is: Repair Operator The selection probability is calculated similarly; differentiated rewards are given based on the quality of candidate solutions, and the reward levels are defined as follows: If the candidate solution becomes the new global optimum, the reward is... If a candidate solution improves the current solution but does not become the global optimum, the reward is... If a candidate solution overlaps with a historical solution, a reward is given. If a candidate solution is inferior but accepted, a reward is given. If a candidate solution is inferior and is rejected, there is no reward; the reward parameter satisfies... > > > >0; The weight update rules are as follows: In the formula, This is the reward value obtained in this iteration; The acceptance criteria for simulated annealing are: The formula for calculating the acceptance probability of a candidate solution is: In the formula, The objective value of the candidate solution. The objective value of the current solution. The effective temperature parameter is calibrated as follows: In the formula, The initial solution is the target value; the algorithm terminates when one of the following conditions is met: Condition 1: The maximum number of iterations is reached. Condition 2: The global optimal solution is not improved after N consecutive iterations.

[0018] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: after pre-constructing the three-component cost estimation function, performing a rapid feasibility test, including a load feasibility test to verify that the cumulative demand for the travel segment after inserting the customer does not exceed the load capacity, and an energy feasibility test to verify that the battery of the vehicle will not be depleted before reaching the next charging opportunity after inserting the customer, filtering out infeasible insertion locations; the following steps are performed: Load-bearing feasibility verification: route k Contains the position to be inserted pos The travel segment, verify the insertion of the customer. c Subsequent cumulative demand shall not exceed load capacity Q : If there is at least one travel segment that can accommodate customer c, then the load-bearing feasibility test is passed; Energy feasibility verification: Simulate the energy consumption trajectory along the path to verify that after inserting customer c, the vehicle's battery state of charge will not fall below the minimum allowable value before reaching the next rechargeable node. : In the formula, For the current position SOC Estimated value Energy consumption per unit distance and For each customer c and leaving customers c The distance; the insertion location is considered feasible only if both load feasibility and energy feasibility are approved.

[0019] Secondly, in order to achieve the above objectives, the present invention discloses an electric truck route and charging coordination system with embedded cost estimation, comprising: The data acquisition module is used to acquire input problem data, which includes traffic network topology information, station node set, customer node set, independent charging station node set, node distance matrix, fleet parameters, customer demand parameters, service time window, and spatiotemporal node electricity price and cost coefficient of each charging node in each time period. The function pre-construction module is used to pre-construct a three-component cost estimation function, wherein the three-component cost estimation function is obtained by inputting problem data based on the customer-interpolated marginal cost estimation method; the three-component cost estimation function includes a distance cost estimation component, a time cost estimation component, and an energy cost estimation component; The optimal scheduling module is used to construct an initial solution based on a three-component cost estimation function, generate an initial vehicle path scheme, and perform CE-ALNS main loop iteration based on the initial vehicle path scheme. After multiple iterations to reach a preset number, it outputs the complete scheduling scheme corresponding to the globally optimal path structure.

[0020] The beneficial effects of this invention are: This invention significantly reduces computational complexity while maintaining solution quality by deeply integrating cost estimation and search decision-making. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the algorithm framework of this invention; Figure 3 This is a schematic diagram showing the charging and discharging power and electricity price at different nodes of the present invention; Figure 4 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1: like Figure 1 As shown, an electric truck route and charging coordination method with embedded cost estimation is proposed, the method comprising the following steps: S101: Obtain input problem data, which includes traffic network topology information, station node set, customer node set, independent charging station node set, node distance matrix, fleet parameters, customer demand parameters, service time window, and spatiotemporal node electricity price and cost coefficient of each charging node in each time period. Specifically as follows: Input traffic network topology information: set of station nodes Customer node set (in For a subset of customers with charging facilities, (A subset of customers without charging facilities) and a set of independent charging station nodes. Distance matrix between nodes Enter fleet parameters: number of vehicles Battery capacity minimum ,maximum Initial / Termination Energy consumption per unit distance Average driving speed v Charging efficiency Discharge efficiency Load capacity Q Maximum charging power Input customer demand parameters: the demand quantity for each customer. Service Window Input spatiotemporal node electricity price: the marginal electricity price of each rechargeable node in each time period. Time period set This corresponds to the discretization of the planning time domain (e.g., |T|=96 corresponds to a 15-minute time period). Input cost coefficient: fixed deployment cost coefficient. Distance cost coefficient Time cost coefficient Time window violation penalty coefficient .

[0024] S102: Pre-construct a three-component cost estimation function, wherein the three-component cost estimation function is obtained by inputting problem data based on the customer-interpolated marginal cost estimation method; the three-component cost estimation function includes a distance cost estimation component, a time cost estimation component, and an energy cost estimation component; The distance cost estimation component reflects the increased detour distance caused by inserting customers, the time cost estimation component reflects the increased cost caused by additional travel time and service time, and the energy cost estimation component explicitly incorporates the spatiotemporal distribution characteristics of nodal electricity prices and the arbitrage potential index. The expression for the three-component cost estimation function is: The distance cost estimation component is as follows: In the formula, c For customers to be inserted, k For the target path, pos The insertion position, i and j These are the predecessor and successor nodes at the insertion position, respectively. For the node i To the customer c distance, To the customer c To the node j distance, For the node i Direct to node j distance, Cost coefficient per unit distance; The time cost estimation component is: In the formula, v The average speed of the vehicle. q c For customers c The demand, Service volume per unit time This is the unit time cost coefficient; if the path k If a time window violation has already occurred, a timeout penalty will be added to that component. The energy cost estimation components are: In the formula, To insert customers c The resulting increase in energy demand for path k Path after insertion k The lowest time-of-use electricity price at charging nodes along the route. The change in the arbitrage potential index caused by the insertion. This is the maximum battery capacity. Path after insertion k Energy requirements, Path before insertion k Arbitrage potential index.

[0025] After pre-constructing the three-component cost estimation function, a rapid feasibility test is performed, including a load feasibility test to verify that the cumulative demand during the customer's journey does not exceed the load capacity, and an energy feasibility test to verify that the vehicle's battery will not be depleted before reaching the next charging opportunity after the customer is inserted. Infeasible insertion locations are filtered out. The process is as follows: Rapid feasibility test: Load-bearing feasibility test: Route k Contains the position to be inserted pos The itinerary segment is calculated by adding the demand already allocated to customers for that segment, and then verifying and adding the customer's demand. c The demand does not exceed the load capacity Q If at least one travel segment meets the conditions, the load test passes; Energy feasibility test: trace back from the insertion point to the nearest rechargeable node (or starting station), accumulating the travel distance and energy consumption. Estimate the distance to the customer. c time SOC Then calculate the departure of the customer c The energy consumption required to reach the next rechargeable node (or the final station). If the entire journey... SOC Stay Within the specified range, the energy test passes; overall feasibility determination: the location is considered acceptable only if both the load test and the energy test pass. pos It is considered feasible.

[0026] Specifically, the following is a load-bearing feasibility test: [Regarding the route] k Contains the position to be inserted pos The travel segment, verify the insertion of the customer. c Subsequent cumulative demand shall not exceed load capacity Q : If there is at least one travel segment that can accommodate customer c, then the load-bearing feasibility test is passed; Energy feasibility verification: Simulate the energy consumption trajectory along the path to verify that after inserting customer c, the vehicle's battery state of charge will not fall below the minimum allowable value before reaching the next rechargeable node. : In the formula, For the current position SOC Estimated value Energy consumption per unit distance and For each customer c and leaving customers c The distance; the insertion location is considered feasible only if both load feasibility and energy feasibility are approved.

[0027] S103: Based on the three-component cost estimation function, a greedy initial solution is constructed to generate an initial vehicle routing scheme. Based on the initial vehicle routing scheme, the CE-ALNS main loop is iterated. After multiple iterations to reach the preset number, the complete scheduling scheme corresponding to the globally optimal path structure is output.

[0028] The process of constructing a greedy initial solution based on the three-component cost estimation function is as follows: Initialize an empty path for each vehicle, iteratively selecting the customer and location with the lowest insertion cost from unassigned customers and inserting them until all customers are assigned, thus obtaining the initial path structure. Design a problem-customized set of destruction operators, including a random removal operator that uniformly and randomly selects customers for removal, a worst-case removal operator that selects the customer with the highest marginal removal benefit for removal, a greedy station replacement operator that randomly replaces station nodes on the path and removes all customers on that path, and a random trip removal operator that randomly selects trip segments and removes all customers within them. Design a problem-customized set of repair operators, including a random repair operator that randomly selects feasible locations for customers to be inserted, a full-cost greedy repair operator that uses a complete three-component cost estimation function to select the location with the lowest estimated cost for insertion, a transport-oriented greedy repair operator that uses only distance and time cost components for evaluation, and an energy-oriented greedy repair operator that uses only energy cost components for evaluation. Implement adaptive operator weight updates based on a roulette wheel selection and reward mechanism, where the operator selection probability is proportional to the weight, and differentiated rewards are given based on the quality of candidate solutions, and the weights are updated accordingly. Employ a simulated annealing acceptance criterion, setting an effective temperature parameter and an acceptance probability calculation formula, and setting a maximum number of iterations and a number of consecutive iterations without improvement as termination conditions.

[0029] The marginal removal cost of the worst removal operator is calculated as follows: Among them, the distance item is for removing customers. c The resulting reduction in distance costs: The time item is for removing customers. c The reduction in time cost resulting from: Charging costs are calculated separately based on whether the customer has access to charging facilities: If the customer c It has charging facilities, and the charging items are: In the formula, For vehicles in the current plan k In customers c time period t The amount of charge, To remove customers c The resulting energy gap This refers to the highest electricity price during the designated time period for the station nodes in this route. If the customer c There are no charging facilities; the charging options are: In the formula, Energy consumption per unit distance; Worst removal operator selection removal cost Minimum specified ratio The customer removed it; The execution steps for the greedy station replacement operator are as follows: Step 1: Randomly select a path uniformly from all paths in the current solution. ; Step 2: Identify the path All station nodes, including the starting station, intermediate return station and the terminating station; Step 3: Assemble at the station D A new station, different from the current station, will be randomly selected. d' ; Step 4: Path Replace all station nodes with the new station d' ; Step 5: Set the path Remove all customer nodes from the list and add them to the set of customers to be inserted. The execution steps of the random route removal operator are as follows: Step 1: Decompose each path in the current solution into a sequence of travel segments, where a travel segment is defined as the sequence of customer nodes between two adjacent visits to a site node; Step 2: Count all travel segments for all paths and construct a travel segment set; Step 3: Randomly select a travel segment from the set of travel segments; Step 4: Remove all customer nodes in the selected travel segment and add them to the customer set to be inserted.

[0030] Specifically, in this embodiment, for each vehicle Initialize empty path , ,in As the starting station, To terminate the site, set the initial battery SOC to [value missing]. Initialize the unassigned customer set U = C.

[0031] Iteratively execute the following process until U is empty: for each customer For each path R k For each feasible insertion position pos Calculate the estimated cost Add random perturbation factor Used for diversification. Selection Minimum combination , will customers Insert path Location Remove from U For the constructed initial path structure Solve the path-fixed subproblem to obtain the initial objective value. The set of custom destruction operators for the design problem is as follows: (1) Random Removal Operator (RR): Let the removal ratio be . Calculate the number of customers removed. Uniformly random selection. Remove each assigned customer from their respective paths.

[0032] (2) Worst-case removal operator (WR): for each assigned customer c Calculate the marginal removal cost. After sorting, select the one with the lowest removal cost (maximum removal benefit). Remove one customer.

[0033] (3) Greedy station replacement operator (GCD): a path is selected uniformly and randomly. R k Identify all its station nodes and randomly select a different station. Perform the replacement and remove all customers from that path.

[0034] (4) Random Trip Removal Operator (RTR): Traverse all paths to identify all trip segments, uniformly and randomly select a trip segment, and remove all customers in it.

[0035] The following is a set of custom repair operators for the design problem: (1) Random Repair Operator (RanR): For each customer to be inserted, a random position is selected from all feasible positions of the path for insertion; (2) The full-cost greedy repair operator (GR-ALL) uses a complete three-component cost estimation function to evaluate all feasible locations and select the location with the minimum estimated cost; (3) Transport-oriented greedy repair operator (GR-TRAN), which uses only distance cost and time cost components for evaluation; (4) Energy-oriented greedy repair operator (GR-ELEC) uses only the energy cost component for evaluation.

[0036] Algorithm framework such as Figure 2 As shown, the adaptive operator weight update mechanism includes: Initialize and destroy operator weight vector and repair operator weight vector Each component is initialized to 1; In each iteration, the destroy operator and the repair operator are selected respectively using a roulette wheel method. The probability of selection is: Repair Operator The selection probability is calculated similarly; differentiated rewards are given based on the quality of candidate solutions, and the reward levels are defined as follows: If the candidate solution becomes the new global optimum, the reward is... If a candidate solution improves the current solution but does not become the global optimum, the reward is... If a candidate solution overlaps with a historical solution, a reward is given. If a candidate solution is inferior but accepted, a reward is given. If a candidate solution is inferior and is rejected, there is no reward; the reward parameter satisfies... > > > >0; The weight update rules are as follows: In the formula, This is the reward value obtained in this iteration; The acceptance criteria for simulated annealing are: The formula for calculating the acceptance probability of a candidate solution is: In the formula, The objective value of the candidate solution. The objective value of the current solution. The effective temperature parameter is calibrated as follows: In the formula, The initial solution is the target value; the algorithm terminates when one of the following conditions is met: Condition 1: The maximum number of iterations is reached. Condition 2: The global optimal solution is not improved after N consecutive iterations.

[0037] The process of executing the CE-ALNS main loop iteration based on the initial vehicle routing plan is as follows: Execute the CE-ALNS main loop. In each iteration, select the destroy operator and the repair operator in a roulette wheel manner. Apply the destroy operator to remove some customers and apply the repair operator to reinsert the removed customers. Solve the path-fixed subproblem for the candidate path structure to evaluate the true target value. Decide whether to accept the candidate solution according to the acceptance criterion. Update the global optimal solution and operator weights. Check the termination condition. Output the complete scheduling scheme corresponding to the global optimal path structure. The complete scheduling scheme corresponding to the global optimal path structure includes the access node sequence of each vehicle, the arrival and departure times of each node, changes in cargo loading, battery state of charge trajectory, and charging and discharging power of each time period and node.

[0038] The set of repair operators includes: Random Repair Operator RanR: For each client to be inserted c Iterate through all locations along all paths, collect a set of locations that pass the feasibility test, and then randomly select a location from this set to place the customer. c insert; Full-cost greedy repair operator GR-ALL: using the full three-component cost estimation function Evaluate the estimated interpolation costs of all feasible locations and select the location with the lowest estimated cost to serve the customer. c insert; Transport-oriented greedy repair operator GR-TRAN: using only distance and time cost components The evaluation prioritizes reducing driving distance and operating time; the energy-oriented greedy repair operator GR-ELEC uses only the energy cost component. The evaluation prioritizes optimizing charging costs and arbitrage opportunities. Among them, the four repair operators in the repair operator set are used in a complementary manner during the search process. The full-cost greedy repair pursues the overall optimality, the transportation-oriented and energy-oriented repairs focus on different cost components, and the random repair ensures the diversity of the search.

[0039] Specifically, the present invention will be further illustrated below through embodiments: Table 1 As shown in Table 1, the effectiveness of the destruction operator combination was verified on a medium-sized instance of 15 customers and 6 vehicles. Four configurations were set up: Configuration D1: using only the random removal operator RR, with a target value of $2847.32; Configuration D2: using the RR+WR combination, with a target value of $2756.45, an improvement of 3.19% compared to D1; Configuration D3: using the RR+WR+RTR combination, with a target value of $2698.73, an improvement of 2.09% compared to D2; Configuration D4: using all four operators RR+WR+RTR+GCD, with a target value of $2634.18, an improvement of 2.39% compared to D3. The results show that the complete operator combination achieves an overall improvement of 7.57% compared to the basic configuration D1, with each operator contributing complementaryly, verifying the effectiveness of the problem-customized operator design.

[0040] The effectiveness of the repair operator combination was verified on the same instance. Four configurations were set up: Configuration R1: using only the random repair operator RanR, with a target value of $2923.56; Configuration R2: using the RanR+GR-TRAN combination, with a target value of $2712.34, an improvement of 7.22% compared to R1; Configuration R3: using the RanR+GR-ELEC combination, with a target value of $2678.45, an improvement of 8.38% compared to R1; Configuration R4: using all four operators, with a target value of $2634.18, an improvement of 9.89% compared to R1. The results show that energy-directed repair (GR-ELEC) contributes the most, verifying the importance of nodal price awareness to solution quality. Full-cost greedy repair (GR-ALL) achieves the best results. The charging and discharging power and price of different nodes are shown below. Figure 3 As shown.

[0041] Comparison of the Cost Estimation Function (CE) and the Path-Fixed Subproblem Evaluation (SP) method on instances with 5-100 customers: 5 customers: CE method 1.2 minutes, SP method 4.7 minutes, speedup 3.9 times, target gap 0.69%; 10 customers: CE method 3.8 minutes, SP method 21.7 minutes, speedup 5.7 times, target gap 0.88%; 20 customers: SP method timed out (6-hour limit), CE method completed in 62.3 minutes; 50 customers: CE method completed in 187 minutes; 100 customers: CE method completed in 413 minutes. The results validate the effectiveness of the cost estimation function on large-scale instances, achieving several times speedup while maintaining solution quality (gap <1%), and making it possible to solve scales that traditional methods cannot.

[0042] Example 2: To achieve the above objective, such as Figure 4 As shown in Embodiment 1, this invention discloses an electric truck route and charging coordination system with embedded cost estimation, comprising: Data acquisition module 11 is used to acquire input problem data, which includes traffic network topology information, station node set, customer node set, independent charging station node set, node distance matrix, fleet parameters, customer demand parameters, service time window, spatiotemporal node electricity price and cost coefficient of each charging node in each time period; The function pre-construction module 12 is used to pre-construct a three-component cost estimation function, wherein the three-component cost estimation function is obtained by inputting problem data based on the customer-inserted marginal cost estimation method; the three-component cost estimation function includes a distance cost estimation component, a time cost estimation component, and an energy cost estimation component; The optimal scheduling module 13 is used to construct a greedy initial solution based on the three-component cost estimation function, generate an initial vehicle path scheme, execute the CE-ALNS main loop iteration based on the initial vehicle path scheme, and output the complete scheduling scheme corresponding to the globally optimal path structure after multiple iterations and reaching a preset number of iterations.

[0043] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.

[0044] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0045] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0046] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.

Claims

1. A method for coordinating the route and charging of electric freight vehicles with embedded cost estimation, characterized in that, The method includes the following steps: Obtain input problem data, which includes traffic network topology information, station node set, customer node set, independent charging station node set, node distance matrix, fleet parameters, customer demand parameters, service time window, and spatiotemporal node electricity price and cost coefficient of each charging node in each time period; A three-component cost estimation function is pre-constructed, wherein the three-component cost estimation function is obtained by inputting problem data based on the customer-interpolated marginal cost estimation method; the three-component cost estimation function includes a distance cost estimation component, a time cost estimation component, and an energy cost estimation component; The initial solution is constructed based on the three-component cost estimation function, generating an initial vehicle routing scheme. The CE-ALNS main loop is then iterated based on the initial vehicle routing scheme. After multiple iterations to the preset number, the complete scheduling scheme corresponding to the globally optimal path structure is output.

2. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 1, characterized in that, The distance cost estimation component reflects the increased detour distance caused by inserting customers, the time cost estimation component reflects the increased cost caused by additional travel time and service time, and the energy cost estimation component explicitly incorporates the spatiotemporal distribution characteristics of nodal electricity prices and the arbitrage potential index. The expression for the three-component cost estimation function is as follows: The distance cost estimation component is as follows: In the formula, c For customers to be inserted, k For the target path, pos The insertion position, i and j These are the predecessor and successor nodes at the insertion position, respectively. For the node i To the customer c distance, To the customer c To the node j distance, For the node i Direct to node j distance, Cost coefficient per unit distance; The time cost estimation component is: In the formula, v The average speed of the vehicle. q c For customers c The demand, Service volume per unit time This is the unit time cost coefficient; if the path k If a time window violation has already occurred, a timeout penalty will be added to that component. The energy cost estimation components are: In the formula, To insert customers c The resulting increase in energy demand for path k Path after insertion k The lowest time-of-use electricity price at charging nodes along the route. The change in the arbitrage potential index caused by the insertion. This is the maximum battery capacity. Path after insertion k Energy requirements, Path before insertion k Arbitrage potential index.

3. The method for coordinating the routing and charging of electric freight vehicles with embedded cost estimation according to claim 2, characterized in that, The formula for calculating the arbitrage potential index is as follows: Define the minimum time-period electricity price for rechargeable nodes along path k as: In the formula, For path k The set of rechargeable nodes traversed. T For time period sets, For nodes s During the period t Electricity price; The arbitrage potential index is defined as the average time-period electricity price difference between rechargeable nodes along the route: Construct a path based on the lowest electricity price and the arbitrage potential index. k Charging cost estimation model: In the formula, For path k Energy requirements, This represents the cost of meeting energy demand at the lowest possible electricity price. This represents the potential profit from arbitrage between nodes using the remaining battery capacity.

4. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 1, characterized in that, The process of constructing a greedy initial solution based on the three-component cost estimation function is as follows: Initialize an empty path for each vehicle, iteratively selecting the customer and location with the lowest insertion cost from unassigned customers and inserting them until all customers are assigned, thus obtaining the initial path structure. Design a problem-customized set of destruction operators, including a random removal operator that uniformly and randomly selects customers for removal, a worst-case removal operator that selects the customer with the highest marginal removal benefit for removal, a greedy station replacement operator that randomly replaces station nodes on the path and removes all customers on that path, and a random trip removal operator that randomly selects trip segments and removes all customers within them. Design a problem-customized set of repair operators, including a random repair operator that randomly selects feasible locations for customers to be inserted, a full-cost greedy repair operator that uses a complete three-component cost estimation function to select the location with the lowest estimated cost for insertion, a transport-oriented greedy repair operator that uses only distance and time cost components for evaluation, and an energy-oriented greedy repair operator that uses only energy cost components for evaluation. Implement adaptive operator weight updates based on a roulette wheel selection and reward mechanism, where the operator selection probability is proportional to the weight, and differentiated rewards are given based on the quality of candidate solutions, and the weights are updated accordingly. Employ a simulated annealing acceptance criterion, setting an effective temperature parameter and an acceptance probability calculation formula, and setting a maximum number of iterations and a number of consecutive iterations without improvement as termination conditions.

5. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 4, characterized in that, The marginal removal cost calculation method for the worst removal operator is as follows: Among them, the distance item is for removing customers. c The resulting reduction in distance costs: The time item is for removing customers. c The reduction in time cost resulting from: Charging costs are calculated separately based on whether the customer has access to charging facilities: If the customer c It has charging facilities, and the charging items are: In the formula, For vehicles in the current plan k In customers c time period t The amount of charge, To remove customers c The resulting energy gap This refers to the highest electricity price during the designated time period for the station nodes in this route. If the customer c There are no charging facilities; the charging options are: In the formula, Energy consumption per unit distance; Worst removal operator selection removal cost Minimum specified ratio The customer removed it; The execution steps for the greedy station replacement operator are as follows: Step 1: Randomly select a path uniformly from all paths in the current solution. ; Step 2: Identify the path All station nodes, including the starting station, intermediate return station and the terminating station; Step 3: Assemble at the station D A new station, different from the current station, will be randomly selected. d' ; Step 4: Path Replace all station nodes with the new station d' ; Step 5: Set the path Remove all customer nodes from the list and add them to the set of customers to be inserted. The execution steps of the random route removal operator are as follows: Step 1: Decompose each path in the current solution into a sequence of travel segments, where a travel segment is defined as the sequence of customer nodes between two adjacent visits to a site node; Step 2: Count all travel segments for all paths and construct a travel segment set; Step 3: Randomly select a travel segment from the set of travel segments; Step 4: Remove all customer nodes in the selected travel segment and add them to the customer set to be inserted.

6. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 1, characterized in that, The process of executing the CE-ALNS main loop iteration based on the initial vehicle routing scheme is as follows: Execute the CE-ALNS main loop. In each iteration, select the destroy operator and the repair operator in a roulette wheel manner. Apply the destroy operator to remove some customers and apply the repair operator to reinsert the removed customers. Solve the path-fixed subproblem for the candidate path structure, evaluate the true target value, decide whether to accept the candidate solution according to the acceptance criterion, update the global optimal solution and operator weights, and check the termination condition. Output the complete scheduling scheme corresponding to the globally optimal path structure; The complete scheduling scheme corresponding to the global optimal path structure includes the access node sequence of each vehicle, the arrival and departure times of each node, changes in cargo loading, battery state of charge trajectory, and charging and discharging power of each time period and node.

7. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 6, characterized in that, The set of repair operators includes: Random Repair Operator RanR: For each client to be inserted c Iterate through all locations along all paths, collect a set of locations that pass the feasibility test, and then randomly select a location from this set to place the customer. c insert; Full-cost greedy repair operator GR-ALL: using the full three-component cost estimation function Evaluate the estimated interpolation costs of all feasible locations and select the location with the lowest estimated cost to serve the customer. c insert; Transport-oriented greedy repair operator GR-TRAN: using only distance and time cost components The evaluation prioritizes reducing driving distance and operating time; the energy-oriented greedy repair operator GR-ELEC uses only the energy cost component. The evaluation prioritizes optimizing charging costs and arbitrage opportunities. Among them, the four repair operators in the repair operator set are used in a complementary manner during the search process. The full-cost greedy repair pursues the overall optimality, the transportation-oriented and energy-oriented repairs focus on different cost components, and the random repair ensures the diversity of the search.

8. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 5, characterized in that, The adaptive operator weight update mechanism includes: Initialize and destroy operator weight vector and repair operator weight vector Each component is initialized to 1; In each iteration, the destroy operator and the repair operator are selected respectively using a roulette wheel method. The probability of selection is: Repair Operator The selection probability is calculated similarly; differentiated rewards are given based on the quality of candidate solutions, and the reward levels are defined as follows: If the candidate solution becomes the new global optimum, the reward is... If a candidate solution improves the current solution but does not become the global optimum, the reward is... If a candidate solution overlaps with a historical solution, a reward is given. If a candidate solution is inferior but accepted, a reward is given. If a candidate solution is inferior and is rejected, there is no reward; the reward parameter satisfies... > > > > 0; The weight update rules are as follows: In the formula, This is the reward value obtained in this iteration; The acceptance criteria for simulated annealing are: The formula for calculating the acceptance probability of a candidate solution is: In the formula, The objective value of the candidate solution. The objective value of the current solution. The effective temperature parameter is calibrated as follows: In the formula, The initial solution is the target value; the algorithm terminates when one of the following conditions is met: Condition 1: The maximum number of iterations is reached. Condition 2: The global optimal solution is not improved after N consecutive iterations.

9. The method for coordinating route and charging of electric freight vehicles with embedded cost estimation according to claim 1, characterized in that, After the pre-constructed three-component cost estimation function is established, a rapid feasibility test is performed, including a load feasibility test to verify that the cumulative demand during the travel segment after inserting the customer does not exceed the load capacity, and an energy feasibility test to verify that the vehicle's battery will not be depleted before reaching the next charging opportunity after inserting the customer, filtering out infeasible insertion locations; the process is as follows: Load-bearing feasibility verification: route k Contains the position to be inserted pos The travel segment, verify the insertion of the customer. c Subsequent cumulative demand shall not exceed load capacity Q : If there is at least one travel segment that can accommodate customer c, then the load-bearing feasibility test is passed; Energy feasibility verification: Simulate the energy consumption trajectory along the path to verify that after inserting customer c, the vehicle's battery state of charge will not fall below the minimum allowable value before reaching the next rechargeable node. : In the formula, For the current position SOC Estimated value Energy consumption per unit distance and For each customer c and leaving customers c The distance; the insertion location is considered feasible only if both load feasibility and energy feasibility are approved.

10. An electric truck route and charging coordination system with embedded cost estimation, employing the electric truck route and charging coordination method with embedded cost estimation as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire input problem data, which includes traffic network topology information, station node set, customer node set, independent charging station node set, node distance matrix, fleet parameters, customer demand parameters, service time window, and spatiotemporal node electricity price and cost coefficient of each charging node in each time period. The function pre-construction module is used to pre-construct a three-component cost estimation function, wherein the three-component cost estimation function is obtained by inputting problem data based on the customer-interpolated marginal cost estimation method; the three-component cost estimation function includes a distance cost estimation component, a time cost estimation component, and an energy cost estimation component; The optimal scheduling module is used to construct an initial solution based on a three-component cost estimation function, generate an initial vehicle path scheme, and perform CE-ALNS main loop iteration based on the initial vehicle path scheme. After multiple iterations to reach a preset number, it outputs the complete scheduling scheme corresponding to the globally optimal path structure.

Images