Unmanned aerial vehicle distribution facility equipment site selection and path integrated optimization method and system
By combining the adaptive large neighborhood search algorithm with mixed integer linear planning model, the location selection and path of drone facilities and equipment are optimized, which solves the problem that traditional methods are difficult to adapt to drone delivery needs, and achieves efficient and low-cost delivery The solution improves the overall efficiency and reliability of the drone distribution system.
Patent Information
- Application Number
- CN202410922885.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-11-01
AI Technical Summary
Traditional logistics facilities site selection and path planning methods are difficult to adapt to the special needs of drone distribution, especially when considering the joint optimization of site selection and path, which leads to high drone distribution efficiency and cost.
Adaptive large neighborhood search algorithm (ALNS) combined with mixed integer linear programming model (MILP) is used to optimize the location selection of drone facilities and equipment by constructing initial solution, corruption and repair operators, greedy iteration algorithms and simulated annealing algorithms. and the path meet the requirements of load capacity and power capacity, and achieve the goal of minimizing total cost.
It improves the overall efficiency and reliability of the drone distribution system, reduces transportation costs and improves delivery timeliness, and can meet customer needs more quickly by optimizing paths and facility locations.
Smart Images

Figure CN118886579A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of drone delivery technology, and in particular to a method and system for integrating site selection and path optimization of drone "last mile" delivery facilities and equipment based on a large-scale domain search algorithm. Background Art
[0002] In recent years, the rapid development of e-commerce and online shopping has put forward higher requirements for logistics and distribution. In particular, the "last mile" delivery, that is, the last section of the transportation process from the distribution center to the consumer's hands, has become the most time-consuming and costly part of the logistics system because it involves complex urban traffic environments and high-frequency small-batch delivery. To solve this problem, drones, as an emerging delivery tool, have gradually attracted attention because of their fast, efficient and flexible characteristics.
[0003] Drone delivery can not only bypass ground traffic congestion, but also play a unique advantage in some special terrains (such as mountainous areas and rural areas). However, with the expansion of the scale of drone delivery, how to optimize the site selection and path planning of drone delivery facilities and equipment has become a key factor in improving the efficiency of drone delivery. Traditional logistics facility site selection and path planning methods are difficult to adapt to the special needs of drone delivery. Therefore, it is of great significance to develop a drone "last mile" delivery facility equipment site selection and path integrated optimization method based on large-scale field search algorithm.
[0004] In drone delivery, site selection and path optimization involve multiple complex factors, including drone flight distance, endurance, obstacle avoidance, and fluctuations in delivery point demand. Traditional optimization methods often consider site selection and path separately, but in practical applications, the two are closely related and influence each other. Based on this, a joint optimization method that comprehensively considers site selection and path is proposed, which can effectively improve the overall efficiency and reliability of the drone delivery system.
[0005] The Adaptive Large Neighborhood Search (ALNS) algorithm is an advanced algorithm that has achieved remarkable results in optimization problems. Its uniqueness lies in its ability to automatically adjust the search strategy during the search process to adapt to the characteristics of different problems, thereby performing well in the process of finding the global optimal solution.
[0006] The adaptive large-scale neighborhood search algorithm is based on the large neighborhood search algorithm, and allows multiple pairs of destruction and repair operators to be used in a single search to obtain the neighborhood of the current solution. According to the quality of the solution obtained by each pair of destruction and repair operations, ALNS assigns them a weight, and assigns a larger weight to operators that obtain high-quality solutions, so that they are more likely to be selected in the next search. Each weight allocation is dynamically adjusted by the quality of the solution, and the weight is used to control the frequency of use of each pair of operators in the search, so as to gradually improve the quality of the current solution during the search and iteration process. In an ALNS algorithm, since there are multiple pairs of different destruction and repair operators, many different neighborhoods will be generated, and each neighborhood can be regarded as generated by the combined use of a set of destruction and repair operators. "Adaptive" is the process of dynamically adjusting weights to select destruction and repair operators.
[0007] Prior art related to the present invention
[0008] Large Neighborhood Search (LNS) is a heuristic algorithm designed to solve large-scale combinatorial optimization problems, such as vehicle routing problems or job scheduling problems. The existing technology uses the LNS algorithm to increase the diversity of initial solutions in each iteration, thereby increasing the possibility of generating the current optimal solution. The algorithm is implemented through key steps such as initial solution generation, large neighborhood search, solution destruction and repair, and acceptance of new solutions.
[0009] The moves in LNS are similar to local search, but use tree-based search and constraint propagation to evaluate the cost and legality of the moves. The moves made are often very powerful, changing a large portion of the solution. The potential to change a large portion of the solution is where the LNS gets its name from, as the size of the neighborhood often scales exponentially with the number of primitives of the solution that the move changes.
[0010] Disadvantages of the prior art 1
[0011] Compared with the ALNS algorithm, the LNS algorithm is more fixed and static in terms of operator selection. LNS usually needs to determine the large-scale neighborhood structure and the corresponding operators before the search process, which has a strong reliance on the specificity of the problem and prior information. This makes LNS likely to require more problem domain expertise when dealing with different types of combinatorial optimization problems, and lacks the flexibility of ALNS when dealing with changes in problems.
[0012] Prior art related to the present invention
[0013] The Variable Neighborhood Search (VNS) algorithm is a heuristic algorithm designed to solve combinatorial optimization problems. The algorithm defines multiple different neighborhood structures, each corresponding to a specific change mode, and then performs local searches in these structures. By dynamically switching different neighborhood structures during the local search process, the VNS algorithm increases the diversity of the search and helps to escape the local optimal solution. The algorithm is known for its simple and flexible design. Through the steps of initial solution generation, neighborhood structure definition, local search, neighborhood structure switching and iterative search, it achieves a comprehensive exploration of the solution space. The VNS algorithm has achieved good results in practical applications and is a powerful tool for solving combinatorial optimization problems.
[0014] Disadvantages of the second prior art
[0015] Compared with the ALNS algorithm, the VNS algorithm has certain shortcomings in some aspects. First, the neighborhood structure of VNS is fixed when the algorithm is initialized, resulting in a relative lack of dynamic adjustment and adaptability to the neighborhood structure. This may limit the adaptability of VNS in dealing with the dynamics and complexity of the problem. In addition, VNS is relatively simple and lacks some complex adaptive mechanisms, making its adaptability during the search process weaker. Finally, compared with the ALNS algorithm, VNS focuses more on local search and its global search capability is relatively weak. Summary of the invention
[0016] In view of the defects of the prior art, the present invention provides a method and system for integrating site selection and path optimization of unmanned aerial vehicle delivery facility equipment.
[0017] In order to achieve the above invention object, the technical solution adopted by the present invention is as follows:
[0018] A method for integrating site selection and path optimization of drone delivery facilities and equipment, comprising the following steps:
[0019] S1. Obtain customer needs, time windows and coordinates, candidate battery swap station coordinates, and drone flight speed, maximum load capacity, and battery capacity;
[0020] S2, based on the information obtained in S1, establish a mixed integer linear programming model (MILP), whose goal is to minimize the total cost, including transportation cost, UAV fixed cost and battery swap station construction cost;
[0021] S3. Based on the combination of adaptive large neighborhood search algorithm and exact solution, MILP is solved to obtain a set of routes that enable the drone to provide services to all customers and return to the warehouse, and meet the requirements of load capacity and power capacity.
[0022] Furthermore, the S3 specifically includes the following sub-steps:
[0023] S31, constructing an initial solution of a path with a battery exchange station location;
[0024] S32, initializing the weight of each operator;
[0025] S33, removing the battery exchange station in the current solution;
[0026] S34, selecting a destruction operator and a repair operator according to a roulette algorithm;
[0027] S35, using a destruction operator to remove a specified number of client nodes in the current solution;
[0028] S36, using a repair operator to reinsert the removed client into the current solution according to a specified rule;
[0029] S37, inserting the battery swap station into the current solution with minimum cost using a greedy iterative algorithm;
[0030] S38, if the current solution is feasible, optimizing the current solution through MILP solving;
[0031] S39, executing the simulated annealing algorithm to update the current solution;
[0032] S310, updating the usage weight of each operator;
[0033] S311, determine whether the iteration stop condition is met, if so, output the optimal solution, otherwise repeat steps S3 to S310 to continue iterative solution.
[0034] Furthermore, the constraints of the MILP include:
[0035] Each customer can only be visited by one drone: Ensure that each customer node is served by only one drone to avoid duplicate deliveries and waste of resources.
[0036] Battery swap station restrictions: Ensure that drones can only swap batteries at designated battery swap stations to prevent uncontrollable factors caused by battery swapping at non-designated locations.
[0037] Path continuity: Ensure the flow conservation of the drone flying from one node to another, that is, the out-degree and in-degree of the drone at a node are equal to ensure the integrity and continuity of the path.
[0038] Warehouse departure and return: Ensure that each drone departs from the warehouse and returns to the warehouse after completing the delivery tasks for all customers.
[0039] Task allocation restrictions: A drone can only be assigned one delivery task after departing from a warehouse, so as to reasonably allocate resources and plan routes.
[0040] Load tracking: Track and record the load changes of the drone to ensure that the load of the drone at each delivery point does not exceed its maximum load capacity.
[0041] Load limit: Ensure that the drone’s load does not exceed the maximum load capacity when leaving the warehouse to ensure transportation safety and efficiency.
[0042] Non-negativity of load: Ensure that the load of the drone at any node is non-negative to prevent load errors.
[0043] Power tracking: Record the power changes of the drone to ensure the traceability of the drone's power changes during flight and ensure that it has sufficient power to complete the mission.
[0044] Depart with full charge: Ensure that the drone departs from the warehouse with full charge and reset the battery to the maximum capacity to ensure maximum endurance.
[0045] Charging at a battery swap station: When the drone is swapping batteries at a battery swap station, the battery level is reset to the maximum capacity to ensure the smooth progress of subsequent delivery tasks.
[0046] Client Node Power Preservation: The UAV’s power level is kept constant at the client node to avoid unnecessary power consumption.
[0047] Non-negative power: Ensure that the power level of the drone at any node is non-negative to ensure that there is enough power to visit the next node or return to the warehouse.
[0048] Service time limit: stipulate the customer's service time to ensure that the drone reaches the customer node and completes the delivery task within the service time window.
[0049] Battery swap time limit: Specify the battery swap time to ensure the time feasibility of the drone during the battery swap operation to ensure the smooth progress of the entire delivery process.
[0050] Battery swap waiting time: Indicates the time feasibility of the drone at the battery swap station, including the possible waiting time between arriving at the battery swap station and starting the battery swap operation.
[0051] Time window feasibility: Ensure the feasibility of the time window of all nodes (including warehouses and customers) to ensure that the drone completes all tasks within the specified time.
[0052] Avoid conflicts at battery swap stations: Resolve conflicts when drones use the same battery swap station to receive services, ensuring that only one drone performs a battery swap operation at the battery swap station during the same time period.
[0053] Furthermore, the specific sub-steps of S35 are as follows:
[0054] S351, Random removal:
[0055] Randomly select a specified number of client nodes from the current solution and put them into the removal set.
[0056] Ensure that the client nodes removed do not include the warehouse points.
[0057] Remove the moved-out client nodes from the current solution.
[0058] S351, Shaw removed:
[0059] Calculate the correlation index between customer points.
[0060] A pair of customer points with a lower correlation is selected according to the correlation index and placed in the removal set.
[0061] Ensure that the client nodes removed do not include the warehouse points.
[0062] Remove the moved-out client nodes from the current solution.
[0063] S351, Worst cost removal:
[0064] Calculate the removal cost for each customer point.
[0065] Select the customer point with the largest removal cost and put it into the removal set.
[0066] Ensure that the client nodes removed do not include the warehouse points.
[0067] Remove the moved-out client nodes from the current solution.
[0068] Furthermore, the specific sub-steps of S36 are as follows:
[0069] S361, Random Insertion:
[0070] Randomly select a specified number of client nodes from the eviction set.
[0071] Randomly insert the selected client node into the path of the current solution.
[0072] Make sure that the inserted client nodes satisfy the constraints on the path.
[0073] Repeat the above steps until all removed customer nodes are inserted into the current solution.
[0074] S362, Greedy Insertion:
[0075] For each client node in the removal set, calculate the cost of inserting it into each possible position in the current solution.
[0076] The location with the lowest insertion cost is chosen and the customer node is inserted there.
[0077] Make sure that the inserted client nodes satisfy the constraints on the path.
[0078] Repeat the above steps until all removed customer nodes are inserted into the current solution.
[0079] S363, Regret Insertion:
[0080] For each client node in the removal set, the cost difference between the best path and the second best path position of inserting it into the current solution, that is, the regret value, is calculated.
[0081] The customer node with the largest regret value is selected and inserted into the optimal position of the corresponding path.
[0082] Make sure that the inserted client nodes satisfy the constraints on the path.
[0083] Repeat the above steps until all removed customer nodes are inserted into the current solution.
[0084] Furthermore, the specific sub-steps of S37 are as follows:
[0085] S371, initialization: taking the path in the current solution as the initial solution and determining a set of candidate battery swap stations.
[0086] S372, loop iteration:
[0087] For each path, traverse each node on the path, including customer nodes and warehouse nodes.
[0088] For each node, the insertion cost after the current node is inserted into the battery swap station is calculated.
[0089] The battery swap stations are inserted at locations that minimize insertion costs.
[0090] Update the battery swap station locations in the route.
[0091] Repeat the above steps until the drone power of all paths is feasible.
[0092] S373. Insert cost calculation, the cost includes allocation cost and construction cost.
[0093] The allocation cost takes into account the additional costs that may be incurred by allocating a new battery swap station to a path, such as path length changes.
[0094] Construction costs take into account the costs of establishing a battery swap station at a new location, such as equipment installation, site rental, etc.
[0095] Furthermore, the specific sub-steps of S38 are as follows:
[0096] S381. Check whether the current solution satisfies all constraints.
[0097] S382: Under the premise of determining that the current solution is feasible, determine whether it needs to be optimized. If the cost of the current solution still has room for improvement, or more constraints need to be met, then optimization is required.
[0098] S383. If it is decided to optimize, first determine the objects to be optimized, including: the location of the battery swap station, the time of visiting customers and the battery swap station.
[0099] S384. Based on the established MILP model, without changing the customer access sequence, directly call the MILP solver to solve the part that needs to be optimized.
[0100] S385. Obtain optimization results obtained by the MILP solver.
[0101] S386. Apply the optimized results to the current solution to replace the original decision and form a new and better solution.
[0102] Furthermore, the specific sub-steps of S39 are as follows:
[0103] S391. Set the initial temperature and select an appropriate cooling rate.
[0104] S392, start iteration of the simulated annealing algorithm.
[0105] S393. Randomly perturb the current solution to generate a temporary solution.
[0106] S394. Calculate the objective function value of the temporary solution.
[0107] S395. If the temporary solution is better than the current solution or the global optimal solution, the temporary solution is accepted.
[0108] S396. If the temporary solution is not better than the current solution or the global optimal solution, the temporary solution is accepted with probability.
[0109] S397. Lower the temperature.
[0110] S398. Repeat the above steps until the stop condition is met.
[0111] S399. When the temperature drops to a sufficiently low level or reaches the maximum number of iterations, the iteration is stopped and the current solution or the global optimal solution is output.
[0112] Furthermore, the specific sub-steps of S310 are as follows:
[0113] At the beginning of each segment, the weights of all operators are initialized to the same value.
[0114] In each iteration, the probability of each operator being selected is calculated based on its weight, and an operator is selected to execute.
[0115] After the operator is selected, the corresponding operations are performed, including: removing customer nodes, inserting customer nodes, and updating paths.
[0116] Update the weight of each operator according to the execution result of the operator. If the operator finds the global optimal solution in the current iteration, increase the weight; if a better solution is found, increase the weight; if an acceptable solution is found, increase the weight, but the increase is less than the increase if a better solution is found.
[0117] When the number of iterations of the current segment reaches the set value, the next search process begins and the above steps are repeated until the maximum number of iterations is reached.
[0118] The present invention also discloses a drone delivery facility equipment site selection and path integration optimization system, which can be used to implement the above-mentioned drone delivery facility equipment site selection and path integration optimization method, specifically including:
[0119] Data acquisition module: responsible for obtaining customer needs, time windows and coordinates, candidate battery swap station coordinates, and information on the flight speed, maximum load capacity, and battery capacity of the drone from external data sources or user input.
[0120] MILP model building module: Based on the information obtained, a mixed integer linear programming model MILP is established. The goal of this model is to minimize the total cost, including transportation cost, drone fixed cost, and battery swap station construction cost.
[0121] Solution module: The established MILP model is solved using an adaptive large neighborhood search algorithm combined with an exact solution to obtain a set of routes that enable all drones to provide services to customers and return to the warehouse while meeting the requirements of load capacity and power capacity.
[0122] Optimization result display module: displays the optimization results obtained by solving the problem to the user, including the optimal path and the location information of the battery swap station, so that the user can view and analyze it.
[0123] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above-mentioned method for integrating site selection and path optimization of unmanned aerial vehicle distribution facility equipment is implemented.
[0124] The present invention also discloses a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned method for integrating site selection and path optimization of unmanned aerial vehicle distribution facilities and equipment is implemented.
[0125] Compared with the prior art, the advantages of the present invention are:
[0126] 1. When establishing the mixed integer linear programming model, the present invention comprehensively considers multiple factors such as customer service time, drone flight speed, maximum load capacity, battery capacity, etc., so as to comprehensively optimize the drone delivery process.
[0127] 2. With the goal of minimizing the total cost, the optimization goal is clearly defined, making the optimization process more intuitive and controllable.
[0128] 3. By optimizing the flight path and service time of drones, the timeliness of delivery is effectively improved, allowing customers to receive goods more quickly.
[0129] 4. The ALNS algorithm is introduced and combined with the MILP solution to achieve efficient search for the optimal solution and improve the efficiency and speed of the optimization process.
[0130] 5. Combining the ALNS algorithm with the MILP solution can effectively cope with the complexity of the UAV facility site selection-path problem and improve the global optimization capability of the solution.
[0131] 6. By optimizing the routes and facility locations of drone delivery, delivery efficiency can be effectively improved while costs can be reduced, providing better solutions for the development of the delivery industry. BRIEF DESCRIPTION OF THE DRAWINGS
[0132] Figure 1 It is a flow chart of the method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to an embodiment of the present invention.
[0133] Figure 2 It is a schematic diagram of basic parameters of a large-scale example of an embodiment of the present invention.
[0134] Figure 3 It is a diagram of the result of a large-scale calculation example of an embodiment of the present invention. DETAILED DESCRIPTION
[0135] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples.
[0136] like Figure 1 As shown, the present invention provides a method for integrating site selection and path optimization of UAV delivery facilities and equipment, comprising the following steps:
[0137] S1. Obtain customer needs, time windows and coordinates, candidate battery swap station coordinates, and drone flight speed, maximum load capacity, and battery capacity;
[0138] The specific information obtained includes: customer demand, time window for receiving service, specific location coordinates, specific location coordinates of each candidate battery exchange station, flight speed of the drone, maximum load capacity, and maximum battery capacity.
[0139] Among them, customer demand and service time window are used to ensure that the drone can meet customer needs within the specified time. The specific location coordinates are the basis for path planning and help calculate the shortest path. The location information of the candidate battery swap station determines the layout of the battery swap facilities and affects the efficiency of drone battery swap and mission planning. In addition, the flight speed, maximum load capacity and maximum battery capacity of the drone are also necessary parameters in the model. They help plan task allocation and avoid exceeding the drone's endurance, thereby improving the overall system performance. By comprehensively considering these parameters, the established model can plan the drone's mission path more comprehensively and accurately, and improve the execution efficiency of the drone system.
[0140] S2, based on the information obtained in S1, establish a mixed integer linear programming model, whose goal is to minimize the total cost, including transportation cost, drone cost and battery swap station construction cost;
[0141] First, make assumptions about the problem:
[0142] (1) All drones are of the same model and have a constant flight speed;
[0143] (2) A drone shall not perform multiple battery swap operations between two customers twice or more; (3) A drone shall not repeatedly swap batteries at the same battery swap station.
[0144] The definitions of symbols required in the present invention are shown in Table 1.
[0145] Table 1
[0146]
[0147]
[0148] The mathematical optimization model is as follows: Objective function:
[0149]
[0150] Constraints:
[0151]
[0152]
[0153] Objective function (1) minimizes the total cost, including the drone transportation cost, drone cost and battery exchange station construction cost. Constraint (2) ensures that each customer can only be visited by one drone. Constraint (3) ensures that the drone can only exchange batteries at the located battery exchange station. Constraint (4) ensures the flow conservation of each intermediate node. Constraint (5) indicates that each drone must return to the warehouse after departing from the warehouse. Constraint (6) indicates that the drone is only assigned one delivery task after departing from the warehouse. Constraint (7) is used to record and track the load changes of the drone. Constraint (8) ensures that the load of drone k does not exceed its loading capacity when leaving the warehouse. Constraint (9) indicates that the load of the drone at any node is non-negative. Constraint (10) tracks and records the changes in the power of the drone. Constraint (11) indicates that the drone departs from the warehouse with a full charge and the power is reset to Q. Constraint (12) indicates that the drone replaces the battery at the battery exchange station and the battery power level is reset to Q. Constraint (13) indicates that the drone remains at the customer node. Constraint (14) indicates that the power level of the drone at any node is non-negative, that is, it is guaranteed to have enough power to visit the next node (remaining customers or returning to the warehouse). Constraints (10)-(14) jointly deal with the limitation of the drone's battery range. Constraint (15) stipulates the limitation of customer service time on arrival time. Constraint (16) stipulates the limitation of battery replacement time on arrival time. Constraint (17) indicates the time feasibility of the drone at the battery exchange station. There may be a waiting time between the arrival of the drone at the battery exchange station and the start of the battery replacement operation. Constraint (18) ensures the feasibility of the time window of all vertices. Constraints (19) and (20) solve the conflict situation of drones using the same battery exchange station to receive services. At that time, drone k replaces the battery before drone k′, then the start battery replacement time of drone k is less than the start battery replacement time of drone k′. The form of constraint (19) is the same as that of constraint (20), but it means that drone k′ replaces the battery before drone k.
[0154] S3, based on the combination of adaptive large neighborhood search algorithm and precise solution, the model established in S2 is solved to obtain a set of routes that can enable all drones to return to the warehouse safely and meet the requirements of load capacity and power capacity.
[0155] 1. In a specific embodiment, constructing an initial solution for a path with a battery exchange station location includes:
[0156] Step 1: Initialize a new path and add a warehouse point to the path;
[0157] Step 2: Find the customer point in the customer set that has the shortest distance to the last point in the path and meets the time window and drone load requirements;
[0158] Step 3: Remove the customer point from the set;
[0159] Step 4: Loop through steps 2 and 3. If no customer that meets the requirements is found and the customer set is not empty, add a warehouse point to the path and return to step 1.
[0160] Step 5: loop through steps 1 to 4 until the customer set is empty, generating a path that does not consider the location of the battery swap station;
[0161] Step 6: Use an iterative greedy algorithm to insert a battery swap station into the currently generated path plan.
[0162] 2. Select a destruction operator to remove a specified number of customers from the current solution. Destruction operators include random removal, Shaw removal, and worst cost removal.
[0163] Among them, random removal: randomly remove q customer points from the current solution and put them into the removal set;
[0164] Shaw removal: Use the correlation index R(i,j) to indicate the correlation between customer point i and customer point j. The lower the R(i,j) value, the more related the two customers are. R(i,j)=ψ1r i,j +ψ2d i,j +ψ3|e i -e j |+ψ4|q i -q j |. When customer point i and customer point j are on the same delivery path, r i,j =1, otherwise 0, d i,j represents the distance between customer point i and customer point j, |e j -e j | is the difference between the earliest service times of the two customers, |q i -q j | is the difference in demand between two customers, ψ1, ψ2, ψ3 and ψ4 are the set weight values;
[0165] Worst cost removal: used to select and remove customer points that cause a significant change in the target cost. The removal cost of removing customer point i is defined as: RG i =d i-1,i +d i,i+1 -d i-1,i+1 .
[0166] 3. Select a repair operator to reinsert the removed customers into the current solution according to the specified method to generate a new customer access order. Repair operators include random insertion, greedy insertion, and regretful insertion.
[0167] Among them, random insertion: select q customers from the removal set and randomly insert them into the path until q customers are inserted and form a new solution;
[0168] Greedy insertion: insert the customer from the removed set at the position with the lowest insertion cost and remove it from the set, repeating this step until the set is empty. ir It represents the cost of inserting the customer point into path r position i, which is defined as follows:
[0169] cost ir =d (i-1,i) +d (i,i+1) -d (i-1,i+1) +c penalty +t penalty (twenty one)
[0170]
[0171] Among them, i-1 and i+1 represent the front and back points of customer point i, respectively, and c penalty is the penalty for exceeding the drone's payload capacity, t penalty It is the penalty for violating the time window requirement;
[0172] Regretful insertion: Improves the greedy insertion operator by incorporating a kind of look-ahead information when selecting the request to insert. The regret value reg i is the difference between the cost of inserting the customer point into its optimal path and the cost of the second-best path. In each iteration, the regret insertion operator selects the one with the largest regret value to insert. ir ∈{1,....,m} represents the path with the rth lowest insertion cost for customer point i, that is, when k≤k′, reg i Defined as:
[0173] 4 An iterative greedy algorithm is used for site selection decisions to select suitable stations from the set of candidate battery swap stations to minimize the sum of construction cost and insertion cost and assign them to paths where power is infeasible.
[0174] Set the current path to: r k ={v0=O,v1,...,v m ,v m+1 = O′}, the insertion cost of the battery swap station The definition is as follows:
[0175]
[0176] γ1+γ2+γ3=1γ1,γ2,γ3≥0 (25)
[0177]
[0178] Among them, γ1, γ2 and γ3 are the set weight values, For insertion gain, is the insertion loss and For additional punishment, It is used to represent the power accessibility of node v. Apparently And the path r k The worst feasible node behavior is defined as: and The smaller the value of, the worse the feasibility of the current path solution; constraint (26) indicates that after the battery exchange station s is inserted into the position v′ after point v, path r k The change in feasibility state is and They represent the accessibility of point v before and after the battery swap station is inserted, and They represent the feasibility of the path before and after insertion. Constraint (27) represents the distance increased after insertion, where v n and v n+1 are the front and back nodes of v′ respectively. Constraint (28) represents the penalty for unreachable points, where M is a large number. If the switching station itself is an unreachable point, that is, This infeasible insertion scheme will be penalized.
[0179] For the infeasible path r k , define a search area L k , L k is the path r k Part of the path r k The first point that cannot be reached by electricity is added to L k Until the battery exchange station or warehouse. By calculating each candidate battery exchange station L k The cost of each feasible location in the calculation includes the allocation cost and the construction cost. The construction cost of the battery swap station that has been built (inserted) is not repeatedly calculated. The corresponding battery swap station and insertion location with the lowest cost are selected to generate a path with higher power feasibility.
[0180] 5. Accurately solve the mixed integer linear programming model and optimize the current solution.
[0181] Without changing the customer access order, MILP solves the following problems: the location decision of the battery swap station, the time of visiting customers and battery swap stations, and the battery swap time. Among the feasible solutions generated iteratively (power, load, and time window are feasible), half of the paths are randomly selected, the battery swap stations are deleted, and the customer access order is fixed. The station locations of the remaining paths are taken into consideration, and the station location and battery swap time are comprehensively decided.
[0182] In the solution solved by the algorithm, a path represents the flight path of a drone k. Let R be the set of all paths, FR represents the fixed path set, CR represents the path set that needs to be solved by MILP to determine the location of the battery exchange station, R = FR ∪ CR. Let Φ = {δ(0),δ(1),...,δ(i),...,δ(n r -1),δ(n r )} represents a fixed path, where δ(0) and δ(n r ) represents a warehouse, δ(1),...,δ(i),...,δ(n r -1) represents the customer or battery exchange station visited by the path, let Г = {τ(0),τ(1),...,τ(j),...,τ(n r′ -1),τ(n r′ )} represents a path that needs to be solved by MILP for location decision, where τ(0) and τ(n r′ ) represents the warehouse, τ(1),...,τ(j),...,τ(n r′ -1) represents the customer visited by this path. Parameter β(k,i,j) means that drone k first visits customer i and then visits customer j (warehouse O′). The mathematical model is as follows:
[0183] Objective function:
[0184]
[0185] Constraints:
[0186]
[0187]
[0188] (3)-(6)
[0189] (10)-(20)
[0190] For a fixed path Φ, constraint (29) is used to fix the access path of drone k. Constraint (30) determines the location decision of the battery exchange station in Φ. For the path Γ that requires MILP solution for location decision, constraint (31) indicates that the customers in Γ must be visited and only visited once. Constraints (32) and (33) fix the customer visit order of the drone through the parameter β(k,i,j) representing the visit order. Constraint (32) indicates that when β(k,i,j) = 1, drone k first visits customer i and then visits customer j, then the arrival time at i is less than the arrival time at j. The form of constraint (33) is the same as that of constraint (32), which indicates the situation where customer j is visited first and then customer i. Constraints (3)-(6) and constraints (10)-(14) only constrain path Γ. Constraints (15)-(20) constrain all paths and solve the arrival time at all customer points and the battery exchange time at the battery exchange station.
[0191] 6. Update the solution using simulated annealing algorithm.
[0192] If the temporary solution s′ is better than the current solution s or the global optimal solution s best , that is, obj(s′)≤obj(s) or obj(s′)≤obj(s best ), then the temporary solution s′ will be accepted; otherwise, the probability Accept, where T>0 represents the temperature. The temperature is from T start Initially, it decreases after each iteration, and the temperature change expression is: T = cT, where c∈(0,1) is the cooling rate.
[0193] 7. Update the weights of each operator.
[0194] The whole search process is divided into multiple segments, each segment ω includes a fixed number of iterations. i,j represents the weight of operator i in the jth segment. In each iteration, the probability of operator i being selected is: p i,j =w i,j / ∑ h∈H w h,j . Where H is the set of all operators. At the beginning of the first iteration of the first segment, the weights of all operators are initialized to the same value. When an operator is selected for use, its weight is updated based on its performance. At the end of each segment, the operator updates its weight based on its historical performance in the segment as follows: If φ i,j ≠0, w i,j+1 =(1-θ)w i,j +θσ i,j / φ i,j Otherwise, w i,j+1 =w i,j Among them, φ i,j and σi,j denote the number and fraction of times operator i is used in the jth segment. When operator i finds the global optimal solution in the jth segment iteration, then σ i,j Increase by 50. When operator i finds a better solution in the jth iteration, then σ i,j Increase by 20. When operator i finds an acceptable solution in the jth iteration, then σ i,j Increase by 10.
[0195] To facilitate understanding of the technical solution of this embodiment, a large-scale example is solved using the method of this embodiment.
[0196] The basic parameters of large-scale examples are as follows Figure 2 As shown in the figure, nodes 0 and 51 represent warehouses, nodes 1-30 represent customers, and nodes 31-50 represent candidate battery exchange stations. The corresponding service time is the battery exchange time at the battery exchange station. Figure 3 shown.
[0197] For this example, the final result is Figure 3 As shown in the figure, different colored lines are used to represent different drone delivery paths; Figure 3 As can be seen from the figure, a total of 10 drones were used for delivery, of which the delivery path of drone 1 was 0→32→28→39→4→31→51, the delivery path of drone 2 was 0→13→35→27→12→36→51, the delivery path of drone 3 was 0→23→29→51, the delivery path of drone 4 was 0→7→33→30→43→9→35→10→36→51, the delivery path of drone 5 was 0→22→38→17→37→16→51, and the delivery path of drone 6 was 0→18→20→37→16→51. The delivery path is 0→38→19→50→26→31→18→37→51. The delivery path of drone 7 is 0→32→1→33→6→51. The delivery path of drone 8 is 0→32→3→31→20→38→24→36→15→14→37→51. The delivery path of drone 9 is 0→32→5→2→33→8→43→11→35→51. The delivery path of drone 10 is 0→21→32→25→51. The final objective function value is 280256.4 yuan.
[0198] In one embodiment of the present invention, a drone delivery facility equipment site selection and path integration optimization system is provided. The system can be used to implement the above-mentioned drone delivery facility equipment site selection and path integration optimization method, specifically including:
[0199] Data acquisition module: responsible for obtaining customer needs, time windows and coordinates, candidate battery swap station coordinates, and drone flight speed, maximum load capacity, battery capacity and other information from external data sources or user input.
[0200] MILP model building module: Based on the information obtained, a mixed integer linear programming (MILP) model is established. The goal of this model is to minimize the total cost, including transportation costs, drone fixed costs, and battery swap station construction costs.
[0201] Solution module: The established MILP model is solved using an adaptive large neighborhood search algorithm combined with an exact solution to obtain a set of routes that enable the drone to serve all customers and return to the warehouse while meeting the requirements of load capacity and power capacity.
[0202] Optimization result display module: displays the optimization results obtained by solving the problem to the user, including the optimal path and the location information of the battery swap station, so that the user can view and analyze it.
[0203] In one embodiment of the present invention, a terminal device is provided, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the above-mentioned unmanned aerial vehicle distribution facility equipment site selection and path integration optimization method.
[0204] In one embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as at least one disk storage.
[0205] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the above-mentioned method for integrating site selection and path optimization of the above-mentioned drone delivery facility equipment in the above-mentioned embodiment; one or more instructions in the computer-readable storage medium are loaded and executed by the processor.
[0206] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0207] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0208] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0209] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0210] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the implementation methods of the present invention, and should be understood that the protection scope of the present invention is not limited to such special statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.
Claims
1. A method for integrating site selection and path optimization of drone delivery facilities and equipment, characterized in that: The following steps are involved: S1. Obtain customer needs, time windows and coordinates, candidate battery swap station coordinates, and drone flight speed, maximum load capacity, and battery capacity; S2, based on the information obtained in S1, establish a mixed integer linear programming model MILP, whose goal is to minimize the total cost, including transportation cost, drone fixed cost and battery swap station construction cost; S3. Based on the combination of adaptive large neighborhood search algorithm and exact solution, MILP is solved to obtain a set of routes that enable the drone to provide services to all customers and return to the warehouse, and meet the requirements of load capacity and power capacity.
2. The method for integrating site selection and route optimization of drone delivery facilities and equipment according to claim 1, characterized in that: The S3 specifically includes the following sub-steps: S31, constructing an initial solution of a path with a battery exchange station location; S32, initializing the weight of each operator; S33, removing the battery exchange station in the current solution; S34, selecting a destruction operator and a repair operator according to a roulette algorithm; S35, using a destruction operator to remove a specified number of client nodes in the current solution; S36, using a repair operator to reinsert the removed client into the current solution according to a specified rule; S37, inserting the battery swap station into the current solution with minimum cost using a greedy iterative algorithm; S38, if the current solution is feasible, optimizing the current solution through MILP solving; S39, executing the simulated annealing algorithm to update the current solution; S310, updating the usage weight of each operator; S311, determine whether the iteration stop condition is met, if so, output the optimal solution, otherwise repeat steps S3 to S310 to continue iterative solution.
3. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 1, characterized in that: The constraints of the MILP include: Each customer can only be accessed by one drone: Ensure that each customer node is served by only one drone; Battery exchange station restrictions: ensure that the drone can only be replaced at the designated battery exchange station; Path continuity: ensure that the flow of a drone flying from one node to another is conserved, that is, the out-degree and in-degree of a drone at a node are equal; Warehouse departure and return: Ensure that each drone must return to the warehouse after completing the delivery tasks for all customers after departing from the warehouse; Task assignment restrictions: A drone can only be assigned one delivery task after it departs from a warehouse; Load tracking: Track and record the load changes of the drone to ensure that the load of the drone at each delivery point does not exceed its maximum load capacity; Load limit: Ensure that the drone does not exceed the maximum load capacity when leaving the warehouse; Non-negative load: ensure that the load of the drone at any node is non-negative; Power tracking: Record the power changes of the drone to ensure the traceability of the power changes of the drone during flight and ensure that it has sufficient power to complete the mission; Depart with full power: Ensure that the drone departs from the warehouse with full power and the power is reset to the maximum capacity; Battery swap station charging: When the drone is swapping batteries at a battery swap station, the battery level is reset to the maximum capacity. Client node power retention: The drone’s power level remains constant at the client node; Non-negative power: Ensure that the power level of the drone at any node is non-negative; Service time limit: stipulate the customer's service time to ensure that the drone reaches the customer's node and completes the delivery task within the service time window; Battery replacement time limit: stipulate the battery replacement time to ensure the feasibility of the UAV in the battery replacement operation; Battery swap waiting time: indicates the feasibility of the UAV at the battery swap station, including the waiting time between arriving at the battery swap station and starting the battery swap operation; Time window feasibility: Ensure the feasibility of the time window of all nodes and ensure that the drone completes all tasks within the specified time; Avoid conflicts at battery swap stations: Resolve conflicts when drones use the same battery swap station to receive services, ensuring that only one drone performs a battery swap operation at the battery swap station during the same time period.
4. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 2, characterized in that: The specific sub-steps of S35 are as follows: S351, Random removal: Randomly select a specified number of client nodes from the current solution and put them into the removal set; Ensure that the client nodes removed do not include warehouse nodes; Remove the moved out client nodes from the current solution; S351, Shaw removed: Calculate the correlation index between customer points; Select a pair of customer points with a lower correlation according to the correlation index and put them into the removal set; Ensure that the client nodes removed do not include warehouse nodes; Remove the moved out client nodes from the current solution; S351, Worst cost removal: Calculate the removal cost for each customer point; Select the customer point with the largest removal cost and put it into the removal set; Ensure that the client nodes removed do not include warehouse nodes; Remove the moved-out client nodes from the current solution.
5. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 2, characterized in that: The specific sub-steps of S36 are as follows: S361, Random Insertion: Randomly select a specified number of client nodes from the removal set; Randomly insert the selected client node into the path of the current solution; Ensure that the inserted client nodes satisfy the constraints on the path; Repeat the above steps until all removed customer nodes are inserted into the current solution; S362, Greedy Insertion: For each client node in the removal set, calculate the cost of inserting it into each possible position of the current solution; Choose the location with the lowest insertion cost and insert the client node into it; Ensure that the inserted client nodes satisfy the constraints on the path; Repeat the above steps until all removed customer nodes are inserted into the current solution; S363, Regret Insertion: For each client node in the removal set, calculate the cost difference between the best path and the second best path position of inserting it into the current solution, that is, the regret value; Select the customer node with the largest regret value and insert it into the best position of the corresponding path; Ensure that the inserted client nodes satisfy the constraints on the path; Repeat the above steps until all removed customer nodes are inserted into the current solution.
6. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 2, characterized in that: The specific sub-steps of S37 are as follows: S371, initialization: taking the path in the current solution as the initial solution and determining a set of candidate battery exchange stations; S372, loop iteration: For each path, traverse each node on the path, including customer nodes and warehouse nodes; For each node, calculate the insertion cost after the current node is inserted into the battery swap station; Inserting battery swap stations into locations that minimize insertion costs; Update the location of battery swap stations in the route; Repeat the above steps until the power of drones on all paths is feasible; S373. Insert cost calculation, the cost includes allocation cost and construction cost; The allocation cost takes into account the additional costs that may be incurred by allocating a new battery swap station to a path; Construction costs take into account the cost of setting up a battery swap station at a new location.
7. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 2, characterized in that: The specific sub-steps of S38 are as follows: S381, check whether the current solution satisfies all constraints; S382. On the premise that the current solution is feasible, determine whether it needs to be optimized; if the cost of the current solution still has room for improvement, or more constraints need to be met, then optimization is required; S383. If it is decided to optimize, first determine the objects to be optimized, including: the location of the battery exchange station, the time of visiting customers and the battery exchange station; S384, based on the established MILP model, without changing the customer access sequence, directly call the MILP solver to solve the part that needs to be optimized; S385, obtaining the optimization result obtained by the MILP solver; S386. Apply the optimized results to the current solution to replace the original decision and form a new and better solution.
8. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 2, characterized in that: The specific sub-steps of S39 are as follows: S391, set the initial temperature and select a suitable cooling rate; S392, start iteration of simulated annealing algorithm; S393, randomly perturb the current solution to generate a temporary solution; S394, calculating the objective function value of the temporary solution; S395. If the temporary solution is better than the current solution or the global optimal solution, accept the temporary solution; S396. If the temporary solution is not better than the current solution or the global optimal solution, the temporary solution is accepted with probability; S397, lower the temperature; S398, repeat the above steps until the stop condition is met; S399. When the temperature drops to a sufficiently low level or reaches the maximum number of iterations, the iteration is stopped and the current solution or the global optimal solution is output.
9. The method for integrated optimization of site selection and path of UAV delivery facilities and equipment according to claim 2, characterized in that: The specific sub-steps of S310 are as follows: At the beginning of each segment, the weights of all operators are initialized to the same value; In each iteration, the probability of each operator being selected is calculated based on its weight, and an operator is selected to execute; After selecting the operator, the corresponding operations are performed, including: removing customer nodes, inserting customer nodes, and updating paths; Update the weight of each operator according to the execution results of the operator; if the operator finds the global optimal solution in the current iteration, increase the weight; if a better solution is found, increase the weight; if an acceptable solution is found, increase the weight, but the increase is smaller than the increase if a better solution is found; When the number of iterations of the current segment reaches the set value, the next search process begins and the above steps are repeated until the maximum number of iterations is reached.
10. An integrated optimization system for site selection and routing of drone delivery facilities, characterized by: The system can be used to implement a method for integrating site selection and path optimization of drone delivery facilities and equipment as described in any one of claims 1 to 9, specifically including: Data acquisition module: responsible for obtaining customer needs, time windows and coordinates, candidate battery swap station coordinates, and information about the flight speed, maximum load capacity, and battery capacity of the drone from external data sources or user input; MILP model building module: Based on the information obtained, a mixed integer linear programming model MILP is established. The goal of this model is to minimize the total cost, including transportation cost, drone fixed cost, and battery swap station construction cost; Solution module: Use the adaptive large neighborhood search algorithm combined with the exact solution to solve the established MILP model to obtain a set of routes that enable all drones to provide services to customers and return to the warehouse, while meeting the requirements of load capacity and power capacity; Optimization result display module: displays the optimization results obtained by solving the problem to the user, including the optimal path and the location information of the battery swap station, so that the user can view and analyze it.
Citation Information
Cited By
Unmanned aerial vehicle scheduling method and device, computer equipment and storage medium
CN119312990A
UAV dispatching method, device, computer equipment and storage medium
CN119312990B
Method for minimizing cost site selection and minimizing site number site selection
CN119579213A
Unmanned delivery method based on heterogeneous facility joint delivery
CN120146744A
Intelligent production scheduling method and device for production workshop
CN121433122A
