Stock shipyard vehicle route optimization method based on adaptive large neighborhood search

Abstract

The invention relates to the technical field of shipyard storage yard scheduling, and particularly discloses a shipyard vehicle route optimization method based on adaptive large neighborhood search and stock, comprising the following steps: establishing a mathematical model based on initial stock and demand, and generating a feasible initial delivery route in combination with the vehicle capacity and the distance between warehouses; performing iterative optimization on the initial delivery route through an adaptive large neighborhood search algorithm to obtain an optimal delivery route; based on the maximum inventory constraint, the safety inventory constraint and the vehicle capacity constraint, the optimal distribution route is applied to an interference recovery model for splitting and combination, and an optimal solution in an interference state is obtained after multiple iterations; according to the type and the influence of the interference, a corresponding recovery strategy is adopted for processing to determine a recovery scheduling scheme; the strategy comprises vehicle self-rescue and new vehicle addition, the distribution route can be quickly recovered and optimized in the face of interference events, and the stability of a supply chain and the continuity of production activities are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of shipyard yard scheduling, and more specifically, to a method for optimizing shipyard vehicle routes based on adaptive large neighborhood search and inventory. Background Art

[0002] The scope of production logistics activities of manufacturing enterprises includes the entire material flow and management activities from the entry of raw materials into the factory to the exit of finished products. Special emphasis is placed on the importance of production and distribution in supply chain management in fast-moving consumer goods manufacturers. Under the traditional operating model, the enterprise logistics system includes basic activities such as independent production, inventory and routing, as well as inbound logistics and intra-factory logistics. For enterprises producing large commodities, due to the fast production speed and large demand for materials, it is necessary to establish small warehouses near the production site and replenish them in time to avoid production stagnation. The inventory volume of these small warehouses is related to the material consumption rate, which constitutes an inventory path production problem.

[0003] The shortcomings of the prior art are:

[0004] 1. There is a lack of a universal algorithm applicable to all inventory path production problems. For small and medium-sized examples, since inventory path production is an NP-hard problem, the optimal solution or approximate optimal solution of the problem can be found by simulating natural phenomena or optimization processes. However, the algorithm solution process needs to be set according to the problem itself, which is also an area that needs to be improved.

[0005] 2. Inventory replenishment strategy is limited by the user's inventory level. Demand is dynamic and random, so it is necessary to formulate a reasonable replenishment strategy to adapt to random dynamic demand.

[0006] In view of this, the present invention provides a method for optimizing vehicle routes in an inventory shipyard based on adaptive large neighborhood search to solve the above problems. Summary of the invention

[0007] In order to overcome the problems in the prior art, the present invention proposes a method based on adaptive large neighborhood search and inventory shipyard vehicle route optimization to solve the scheduling problem of internal production logistics vehicles in large manufacturing industries after being disturbed.

[0008] The present invention provides a method for optimizing vehicle routes in a shipyard based on adaptive large neighborhood search and inventory, comprising the following steps:

[0009] A mathematical model is built based on the initial inventory and demand, combined with the vehicle capacity and the distance between warehouses to generate a feasible initial delivery route;

[0010] The initial delivery route is iteratively optimized through an adaptive large neighborhood search algorithm to obtain the optimal delivery route;

[0011] Based on the maximum inventory constraints, safety inventory constraints and vehicle capacity constraints, the optimal delivery route is applied to the interference recovery model for splitting and combining. After multiple iterations, the optimal solution under interference conditions is obtained.

[0012] According to the type and impact of the interference, appropriate recovery strategies are adopted to deal with and determine the recovery scheduling plan; strategies include vehicle self-rescue and dispatching new vehicles.

[0013] As a preferred technical solution of the present invention, a mathematical model is established within one cycle based on the delivery route cost and the vehicle cost; the mathematical model includes:

[0014] The objective function is to minimize the total cost, which includes the delivery route cost and vehicle cost;

[0015] The constraints are vehicle capacity, warehouse demand, number of vehicles and routes, where:

[0016] Vehicle capacity constraint: The delivery volume of each vehicle cannot exceed its maximum capacity.

[0017] Warehouse demand constraint: The demand of each warehouse must be met.

[0018] Vehicle number constraint: The number of vehicles used cannot exceed the total number of available vehicles.

[0019] Route constraint: Each route must start from the main warehouse and return to the main warehouse.

[0020] As a preferred technical solution of the present invention, the logic for obtaining the initial delivery route is:

[0021] Based on the hybrid collaborative framework, warehouse demand is determined by analyzing the initial inventory suitable for warehouse distribution.

[0022] The encoding and generation algorithm of the initial solution based on inventory demand and distance is proposed to obtain feasible solutions.

[0023] The optimal vehicle route is obtained through multiple iterations of adaptive large neighborhood search.

[0024] As a preferred technical solution of the present invention, the mathematical model of the interference event includes

[0025] When a disruption event occurs, the objective function is to minimize the total cost, which includes the delivery route cost, vehicle cost, and delay cost;

[0026] The constraints are interference events, vehicle self-rescue, dispatch of new vehicles and safety inventory, among which:

[0027] Interference event constraints: Consider the impact of interference events on delivery routes.

[0028] Vehicle self-rescue constraints: The vehicle can save itself and change its route when encountering interference.

[0029] Constraint for dispatching new vehicles: When necessary, new vehicles can be dispatched to serve warehouses that have not been dispatched.

[0030] Safety stock constraint: Ensure that the inventory level in each warehouse does not fall below the safety stock value.

[0031] As a preferred technical solution of the present invention, the warehouse replenishment quantity and vehicle capacity are determined according to the optimal performance strategy to determine the warehouse distribution combination;

[0032] The distribution combination of warehouses is adjusted according to the distance between warehouses, and then the order of warehouses is determined according to the initial inventory of warehouses to integrate vehicle distribution routes;

[0033] Determine whether a disruption event has occurred on the delivery vehicle;

[0034] If there is no interference incident with the delivery vehicle, re-plan the delivery route to minimize the total cost;

[0035] If a distribution vehicle has an interference event, determine the type of interference event and determine the location of each distribution vehicle when the interference event occurs, so as to adopt the corresponding recovery strategy to handle and determine the recovery scheduling plan

[0036] As a preferred technical solution of the present invention, an optimal warehouse warning value is formulated, the optimal warehouse warning value is marked as a preset warning value, and different recovery strategies are formulated based on the preset warning value, the recovery strategies include a first strategy and a second strategy;

[0037] If the warehouse warning value is less than or equal to the preset warning value, the first strategy is selected, that is, no new vehicles are dispatched, and the delivery route is changed. For example, the vehicle can save itself by changing the route, which is suitable for interference events with smaller impacts.

[0038] If the warehouse warning value is greater than the preset warning value, the second strategy is selected, that is, to dispatch a new vehicle, which is suitable for situations where the parking waiting time is too long or there is no reasonable new route.

[0039] As a preferred technical solution of the present invention, a hybrid collaborative framework based on large neighborhood search includes: a route generation algorithm, an algorithm for determining warehouse replenishment based on a performance value adjustment strategy, and a large neighborhood removal and repair operator based on this problem.

[0040] The specific advantages of the present invention are as follows:

[0041] The present invention proposes a new random inventory path optimization problem, comprehensively considers material consumption, inventory level and vehicle path optimization, introduces safety inventory and warehouse capacity constraints, as well as detailed classification and response strategies for interference events, and designs a hybrid collaborative framework based on large neighborhood search, which effectively reduces distribution costs and improves scheduling flexibility and efficiency. In particular, it can quickly restore and optimize distribution routes in the face of interference events, ensuring the stability of the supply chain and the continuity of production activities. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 An illustrative diagram of an internal enterprise production and vehicle routing optimization method based on adaptive large neighborhood search and inventory constraints is provided for this application;

[0043] Figure 2 A disturbance management vehicle self-rescue strategy diagram for an enterprise internal production and vehicle route optimization method based on adaptive large neighborhood search and inventory constraints is provided for this application;

[0044] Figure 3 A new vehicle dispatching strategy diagram for interference management based on an adaptive large neighborhood search and inventory-constrained internal production and vehicle route optimization method provided in this application;

[0045] Figure 4 An initial roadmap for an internal enterprise production and vehicle routing optimization method based on adaptive large neighborhood search and inventory constraints provided for this application;

[0046] Figure 5 A re-planning diagram of an enterprise internal production and vehicle route optimization method based on adaptive large neighborhood search and inventory constraints is provided for this application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Figure 1An illustration of an internal production and vehicle route optimization method based on adaptive large neighborhood search and inventory constraints is shown. A reasonable inventory warning, i.e., the initial inventory, is set up. Based on the initial inventory, the demand of each sub-warehouse can be determined, and the initial feasible solution, i.e., the initial distribution route, can be generated by combining the capacity of the vehicle and the distance between the warehouses. Then, based on the adaptive large neighborhood search algorithm, the feasible solution is iterated to obtain the optimal solution, and the route corresponding to the optimal solution is the optimal distribution route. In order to obtain a recovery plan after route interruption. The scenarios in which vehicles may interfere during transportation in the shipyard are analyzed, and these interferences are combined and classified to determine the distribution combination of the warehouse. Then, different recovery strategies are designed according to different types of interference. In the strategies that require re-planning of the path, such as the first strategy and the second strategy, the path is split and combined. The split paths are divided into a set of completed distribution warehouses and a set of unfinished distribution warehouses. Enter the iterative process of adaptive large neighborhood search, and after each iteration, they are recombined with the unfinished distribution warehouses to form a new combination. Before the set of unfinished distribution warehouses enters the iterative process of adaptive large neighborhood search, the memory of the completed distribution warehouses is added, including cost and time. Through multiple iterations, the optimal interference recovery path is finally obtained. In each iteration of these two parts, the materials in the warehouse are redistributed to make full use of the vehicle capacity.

[0049] Figure 2 The vehicle self-rescue strategy for interference management is described, and the specific methods are:

[0050] When a transport vehicle encounters an interference event, it does not receive or does not need to receive assistance from a new vehicle, and responds to the interference event by changing the vehicle's own route. The first strategy is suitable for interference events that have a small impact on vehicle transportation. For example, when a vehicle breaks down, the vehicle needs to wait. At this time, the route cannot be changed in time, resulting in a greater overall delay time. The first strategy is applied to interference events in category 2. The location where the interference event occurs is assumed to be a temporary point, and the distribution route is replanned to minimize the cost and storage efficiency, excluding the location that has been delivered.

[0051] Figure 3 The strategy diagram for adding new vehicles to interfere with management is as follows:

[0052] When encountering an interference event, a new vehicle is needed to serve the unscheduled warehouse because the vehicle has been waiting for too long or there is no reasonable new route for the dispatched warehouse, such as a long route or high cost. Apply the second strategy to the interference event of category 1. Since the inventory of the general warehouse is limited, it is necessary to pick up the materials of the vehicle that has an interference event, and use the point where the failure occurs as a temporary point. The new vehicle replans the route starting from the temporary point.

[0053] Example 1: In a Chinese shipyard, the welding materials used in the construction of large cruise ships include welding rods, welding wires, carbon rods, fluxes, and gaskets. The welding materials are distributed from the main warehouse to the sub-warehouses of each workstation, and the distribution is fixed-point distribution. There are a total of 12 welding material distribution points in the factory, distributed in ten locations, namely the straight center, component 1# platform, component 2# platform, general assembly building, control center, processing 36 meters, curved surface 45 meters, platform 1, platform 8, and platform 9B. There are three transport vehicles in the welding material warehouse with a load capacity of 8t. The data in the table represent the information of each distribution point in a certain period. The first three columns are the names of all warehouses. The last five columns are the current quantities of welding rods, welding wires, carbon rods, fluxes and gaskets in each sub-warehouse when the inventory of existing materials reaches 5, and the unit of measurement is 100kg.

[0054]

[0055] The initial scheduling plan is as follows Figure 4 As shown in Figure 1, two routes and two vehicles are involved. The dispatching order of vehicle 1 is general warehouse → s1 → s2 → s4 → s5 → general warehouse. The dispatching order of vehicle 2 is general warehouse → s6 → s0 → s9 → s8 → s7 → s3 → general warehouse. In the process of assigning vehicle 1 to sub-warehouse 2, a type 2 interruption is encountered. According to the proposed algorithm, the route is replanned using strategy 1, and the results are as follows: Figure 5 As shown in the figure. After re-planning, the route of vehicle 1 is: general warehouse → s1 → s0 → s9 → s8 → s7 → s3 → general warehouse. The route of vehicle 2 remains unchanged: general warehouse → s6 → s2 → s5 → s4 → general warehouse.

[0056] Example 2: Please refer to Figure 1 The present invention provides a technical solution: a method based on adaptive large neighborhood search and inventory shipyard vehicle route optimization to solve the scheduling problem of internal production logistics vehicles in large manufacturing industries after being disturbed, that is, when the vehicle encounters interference when distributing materials to the workstation, the original scheduling route is affected, and a series of strategies are used to restore or optimize the scheduling route to meet the production needs of the workstation. Based on the inventory path problem, a mathematical model of material consumption and vehicle routes in the warehouse is first proposed, that is, minimizing vehicle cost and distribution cost, and the optimal distribution route is found through the model, and then the path is applied to the interference recovery model. In addition, maximum inventory constraints, safety inventory constraints and vehicle capacity constraints are considered. Based on adaptive large neighborhood search, a hybrid collaborative framework is proposed to solve these problems. In this collaborative framework, the warehouse demand is determined by analyzing the initial inventory suitable for warehouse distribution. The encoding and generation algorithm of the initial solution based on inventory demand and distance is proposed to obtain a feasible solution. Then, the optimal vehicle route is obtained through multiple iterations of adaptive large neighborhood search. Based on the location where the interference occurs, the optimal route is split and combined, and the optimal solution under the interference state is obtained after multiple iterations.

[0057] Due to the limited capacity of the vehicle, the demand for the sub-warehouse cannot be increased. pc , that is, the distribution quantity of materials in sub-warehouse i is all set to the optimal value, so let b pc It is a variable that can be adjusted to make it close to the optimal inventory. c When , the vehicle takes the distribution center warehouse as the starting point and delivers the same packaged materials c to the branch warehouse p. The number of deliveries is the number of packages, and the space occupied is v. Consider the transport vehicle The transportation network consists of a warehouse P and a branch warehouse P. All geographical locations (coordinates) in the network are known. The consumption rate of each branch warehouse material c per unit time is the same, denoted as d pc . When formulating a distribution plan, it is necessary to consider the consumption of materials by the branch warehouse during the distribution process, and the goal is to minimize the distribution route distance and the number of vehicles. Assume that there are H types of interference events, let H = {1, 2, ..., H}. When a distribution vehicle encounters an interference event h∈H, determine the type of h, adopt the corresponding strategy, and then decide whether it is necessary to plan the route again. If the route needs to be planned again, keep the route that has been taken unchanged and only change the route that has not been taken. If the inventory of material c in the branch warehouse p is lower than the safety inventory value, then the difference below the safety inventory is used as the penalty value for distribution. The goal of the interference response model is to minimize the path cost, vehicle cost and delay cost after re-planning. The corresponding mathematical model established is as follows:

[0058] If there is no interference event, we only need to consider the delivery route cost and vehicle cost. In one cycle, we need to establish a mathematical model and related constraints, and then attach the corresponding symbol explanation table:

[0059]

[0060] A mathematical model is constructed based on the minimization of vehicle cost and distribution cost, where the first term is the route cost and the second term is the vehicle cost.

[0061] Constraints:

[0062]

[0063] The above two formulas indicate that each sub-warehouse must be visited once.

[0064]

[0065] The formula means that for each sub-depot, the number of vehicles entering the point is equal to the number of vehicles leaving the point.

[0066]

[0067] The above two items represent the number of times each vehicle departs from and returns to the main warehouse.

[0068]

[0069] The formula means that the sum of the demands of all warehouses does not exceed the capacity of the vehicle.

[0070]

[0071] The formula means that the number of vehicles assigned to a delivery task cannot exceed the total number of vehicles to ensure that there is always a certain number of vehicles as backup.

[0072]

[0073] The above two formulas represent the safety stock constraint and the maximum stock constraint for the inventory level at any given time. The left side of the inequality represents the inventory when the delivery vehicle arrives at the branch warehouse, which is equal to the initial inventory minus the consumption plus the amount of materials delivered.

[0074]

[0075] The formula ensures that the vehicle's cumulative driving distance does not exceed its maximum driving distance.

[0076] The mathematical model of interference events is established:

[0077]

[0078] In the objective function, Δl i,r is the delay cost of branch i that may be caused by the interference event encountered by delivery vehicle r. It is the constraint that the vehicle reaches the warehouse penalty value. or ≤V rt ≤V means that the capacity of the new vehicle does not exceed the total vehicle capacity. Indicates that the vehicle in transit departs from the current location and either goes to a warehouse that has not been delivered or returns directly to the main warehouse. If a transportation failure occurs, the vehicle is assumed to return to the main warehouse, and since the transport vehicle r t Needs repairs and can no longer be used to deliver supplies. Indicates that a new vehicle is dispatched from the main depot to the location where the interference event occurs. Indicates that all vehicles must return to the parking lot. left +m new ≤m means that the number of newly dispatched vehicles and the total number of vehicles in transit do not exceed the total number of vehicles. The objective function minimizes the total cost and optimizes the warehouse performance after the interference event occurs.

[0079] Shipyard production cannot be stagnant, so the delay in distribution cannot exceed the safety stock. In order to make the above distribution problem meaningful, the delay cost Δl of branch i i The inequality also needs to be satisfied.

[0080] 0≤Δl i,r ≤ls

[0081] At the same time, if there are enough vehicles, the initial inventory of each warehouse must satisfy the inequality to ensure normal production, that is, the distribution vehicle requires the inventory of each warehouse to be On duty, deliver supplies.

[0082]

[0083] The explanation of the parameters is shown in the following table:

[0084]

[0085]

[0086] 3. Establish optimal warehouse warning values:

[0087] During the warehouse operation, the inventory of the warehouse that reaches the warning value is always less than or equal to the inventory of other warehouses, which makes the distribution volume of the warehouse that reaches the warning value always greater than or equal to the distribution volume of other warehouses. In order to complete the distribution task meaningfully, assume that the distribution volume of the warehouse that reaches the warning value is b w , the distribution volume of each warehouse should meet b i ≤ b w In the design, make b i Expand to b w , if the delivery vehicle is to be operated as fully loaded as possible, then b1+b2+…+b i ≥V, i.e. b w +b w +…+b w ≥V, if the distance cost is not considered, at least one vehicle is needed to meet the task of delivering n warehouses, then there are nb w ≥V,b w ≥V / n, so the warehouse warning value needs to be set to I w,i ≤us-V / n. If all vehicles are to participate in the distribution and the amount of materials delivered cannot exceed the sum of the capacities of all vehicles, then b1+b2+…+b i ≤mV. Similarly, the distribution volume of each warehouse is expanded to b w , at this time there is nb w ≤mV, the warehouse warning value needs to be set to I w,i≥us-mV / n. The range of inventory warning values ​​to ensure the normal operation of the warehouse is [us-mV / n,us-V / n]. The initial inventory needs to meet Therefore, if The lower limit of the inventory warning value is set to Otherwise, set

[0088] 4. Formulate corresponding strategies for different interferences, and then determine the recovery scheduling plan: the strategies include the first strategy and the second strategy; among which:

[0089] The first strategy: vehicle self-rescue, that is, vehicle self-rescue during transportation. When a transport vehicle encounters an interference event, it does not receive or does not need to receive assistance from a new vehicle, and responds to the interference event by changing the vehicle's own driving route. The first strategy is suitable for interference events that have a small impact on vehicle transportation. For example, when a vehicle breaks down, the vehicle needs to wait. At this time, the route cannot be changed in time, resulting in a greater overall delay time. The first strategy is applied to interference events in category 2, and the location where the interference event occurs is assumed to be a temporary point P. t , except for the locations that have already been delivered, re-plan the delivery routes to minimize costs and warehousing efficiency.

[0090] The second strategy: dispatch new vehicles. When encountering interference events, because the waiting time for vehicles to park is too long or there is no reasonable new route for the dispatched warehouse, such as the route is long and the cost is high, a new vehicle is needed to serve the undispatched warehouse. Apply the second strategy to the interference events of category 1. Since the inventory of the general warehouse is limited, it is necessary to pick up the materials of the vehicle that has the interference event, and use the point where the failure occurs as a temporary point. The new vehicle replans the route starting from the temporary point.

[0091] Both strategies mentioned above will increase the dispatching cost. The first strategy will increase the transportation cost and delay cost of distribution; the second strategy will increase the vehicle cost, transportation cost and delay cost of distribution. Therefore, it is necessary to choose a reasonable strategy to minimize the distribution cost and delay cost for different interference events.

[0092] 5. A hybrid collaborative framework based on large neighborhood search is designed to solve the distribution model of vehicles encountering interference events, including a solution generation algorithm, a warehouse replenishment algorithm based on a performance value adjustment strategy, and a large neighborhood removal and repair operator based on this problem:

[0093] In order to make the adaptive large neighborhood search algorithm run quickly and efficiently and obtain excellent solutions, it is necessary to design a suitable initial solution based on the distance between warehouses and the current inventory. Generally speaking, the warehouses with smaller initial inventory differences are less likely to be delivered by the same vehicle in sequence, and the warehouses with larger initial inventory differences and smaller distance differences are more likely to be delivered by the same vehicle in sequence. When a delivery vehicle has served a warehouse and the time it takes to deliver to the next warehouse is less than the time it takes for the materials to be consumed to the safety inventory, the warehouse can be considered as the next delivery target. The greedy rule is used to select the next delivery warehouse, that is, before the warehouse reaches the safety inventory, the warehouse with a closer distance is preferred as the next delivery point. The solution is encoded in the form of rows. For example, when there are 10 warehouses, [2,3,1,6,5,4,8,9,10,7] is a code without a total warehouse solution. If the total warehouse is included, the encoding is [0,2,3,1,0,6,5,4,8,0,9,10,7,0], where 0 represents the total warehouse and [0,2,3,1,0] represents a delivery route.

[0094] From the vehicle distribution process, we can see that when a warehouse issues an early warning due to a shortage of materials, a vehicle needs to be arranged for distribution. The differences in production processes of each warehouse workstation lead to the fact that warehouses do not issue distribution early warnings at the same time, but in order to increase the vehicle loading rate, all warehouses are replenished with appropriate amounts of materials. The amount of materials that warehouse i needs to replenish is b i , the delivery quantity of a vehicle in the optimal delivery route is the sum of the materials required by several warehouses, that is, b1+b1+…+b i The following situations may occur during the delivery process:

[0095] Total amount of materials But b1+b1+…+b i >V. In this case, re-arranging another vehicle will increase the distribution cost. At this time, the distribution volume needs to be adjusted, and the amount of materials in the distribution warehouse should be appropriately reduced so that the total amount of materials is b1+b1+…+b i =V.

[0096] There is a total amount of material b1+b1+…+b i = V. At this time, there is no need to adjust the warehouse's delivery volume.

[0097] Total amount of materials But b1+b1+…+b i <V. In this case, increasing the distribution volume of a warehouse will far exceed the capacity of the vehicle. Not increasing the distribution volume will increase the empty rate of the vehicle. At this time, it is necessary to adjust the distribution volume and appropriately increase the distribution volume of the warehouse so that the total volume of materials is b1+b1+…+b i =V.

[0098] The strategy based on inventory cost performance allocates the distribution volume of the warehouse so that the warehouse can operate normally, achieve the best performance of warehouse inventory, and will not increase too much distribution cost. This algorithm is based on the greedy idea that warehouse managers rely on in actual decision-making, that is, the farther the distribution center is, the more materials are distributed. Since the warehouse storage performance value is only related to the target value of the inventory and the distribution volume, only these two factors are involved in the preparation of the algorithm.

[0099] After the initial solution is generated, the optimal solution to the problem needs to be found. The optimal solution to the problem is obtained based on the adaptive large neighborhood search algorithm. This is done by designing multiple groups of destruction operators and repair operators, expanding the spatial search range of the solution, improving the current solution, and scoring or weighting the operators. The better the performance of the operator, the higher the score or weight. In each iteration, the destruction and repair operators are selected and weighted according to their previous performance, and an efficient combination method is used to improve the algorithm's optimization ability, thereby finding the optimal solution.

[0100] (1) Removal operator. Based on this problem, five destruction operators are introduced:

[0101] ① Worst path removal operator: select the two paths with the highest cost, then remove one warehouse from the two paths, and repeat until all warehouses on the two paths are removed.

[0102] ② Delivery removal operator: When delivering to each warehouse by vehicle, there exists b1+b1+…+b i <V, and b1+b1+…+b i When the difference with V is small, it is necessary to remove the warehouse with the smallest delivery volume and insert a warehouse with a slightly larger delivery volume to make the vehicle utilization rate higher. At the same time, there is b1+b1+…+b i >V, and b1+b1+…+b i When the difference with V is small, it is necessary to remove the warehouse with the largest delivery volume and insert a warehouse with a slightly smaller delivery volume than this warehouse so that the total delivery volume does not exceed the capacity of the vehicle too much.

[0103] ③ Neighborhood removal operator: The closer the inventory of warehouses is, the more likely it is to cause delivery delays. Therefore, it is necessary to determine the inventory of warehouses in each path and randomly remove warehouses with smaller inventory differences. If there are multiple warehouses, a strategy of randomly removing one is adopted.

[0104] ④ Position removal operator: Taking the main warehouse as the coordinate origin, divide all warehouses into four areas, check the number of warehouses belonging to each area in the path. Set the probability of removing the warehouse according to the number of areas, and then randomly remove the warehouse.

[0105] ⑤ Shortest path removal operator: Determine whether there is a shortest path among all paths. If all the warehouses it contains do not exceed the total number of paths, then remove this path.

[0106] Through the above removal method, all removed warehouses are put into a set, and then the warehouses in the set are inserted into the appropriate path through the proposed existing stock insertion, combined with the repair method based on the greedy insertion and regret insertion proposed by Ropke&Pisinger.

[0107] ① Insert existing stock: Compare the size of the existing stock of materials in the removed warehouses, and randomly insert the previously destroyed warehouses into any route in the order of existing stock, from small to large, according to the method generated by the initial solution.

[0108] ② Greedy insertion: Greedy insertion is to place the nodes to be inserted into all positions of the line in turn, and record the position with the best cost change. At this time, there is no need to consider the total amount of delivered materials, which is similar to the delivery removal operator.

[0109] ③ Regret insertion: Regret insertion is the difference between the cost calculated after inserting a warehouse into the best insertion position and the cost calculated after inserting into the second best insertion position. The larger the regret value of a warehouse, the lower the benefit of inserting it into other positions if the node is not inserted into the best position at present. Calculate the regret values ​​of n warehouses that need to be inserted in turn, find the node with the largest regret value among the current n warehouses, insert it into the line, and update the cost. Calculate the regret values ​​of n-1 warehouses that need to be inserted in the loop again, find the warehouse with the largest regret value among the n-1 warehouses that need to be inserted and insert it. Repeat the above steps until all removed warehouses are repaired.

[0110] (2) Dynamically adjust weights and select

[0111] The algorithm selects and adjusts the removal and repair operators according to the operator weights and roulette wheel method during iteration. The roulette wheel method is used to set weights for each operator and calculate the probability of occurrence to finally select the operator. The operator selection probability is calculated by the formula For the removal operator, k = 5; for the repair operator, k = 3. Record the performance of each removal and repair operator and dynamically adjust the weight w i At the beginning, each operator is set with the same weight, and then the weight is continuously updated according to the quality of the results. The weight is updated according to formula (15), where λ is the weight update coefficient, which is set to 0.4; s i is the score of the operator; μ i is the number of occurrences of the operator.

[0112]

[0113] When the iteration is completed, the scores s of the destruction and repair methods used in the last iteration are calculated using different rules i If the result is the best one in the world, then s i =30; if the result is better than the original result but worse than the optimal result, then s i =20; if the result is worse than the original result, then s i = 10. Finally, the Metropolis criterion in the simulated annealing algorithm is used to obtain the superior and inferior solutions under the selected probability.

[0114] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0115] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing vehicle routes in shipyards based on adaptive large neighborhood search and inventory, characterized in that: The following steps are involved: A mathematical model is built based on the initial inventory and demand, combined with the vehicle capacity and the distance between warehouses to generate a feasible initial delivery route; The initial delivery route is iteratively optimized through an adaptive large neighborhood search algorithm to obtain the optimal delivery route; Based on the maximum inventory constraints, safety inventory constraints and vehicle capacity constraints, the optimal delivery route is applied to the interference recovery model for splitting and combining. After multiple iterations, the optimal solution under interference conditions is obtained. According to the type and impact of the interference, appropriate recovery strategies are adopted to deal with and determine the recovery scheduling plan; strategies include vehicle self-rescue and dispatching new vehicles.

2. The method for optimizing vehicle routes based on adaptive large neighborhood search and inventory shipyard according to claim 1, characterized in that: Based on the delivery route cost and vehicle cost, a mathematical model is established within a cycle; the mathematical model includes: The objective function is to minimize the total cost, which includes the delivery route cost and vehicle cost; The constraints are vehicle capacity, warehouse demand, number of vehicles and routes, where: Vehicle capacity constraint: The delivery volume of each vehicle cannot exceed its maximum capacity. Warehouse demand constraint: The demand of each warehouse must be met. Vehicle number constraint: The number of vehicles used cannot exceed the total number of available vehicles. Route constraint: Each route must start from the main warehouse and return to the main warehouse.

3. The method for optimizing vehicle routes based on adaptive large neighborhood search and inventory shipyard according to claim 2, characterized in that: The logic for obtaining the initial delivery route is: Based on the hybrid collaborative framework, warehouse demand is determined by analyzing the initial inventory suitable for warehouse distribution. The encoding and generation algorithm of the initial solution based on inventory demand and distance is proposed to obtain feasible solutions. The optimal vehicle route is obtained through multiple iterations of adaptive large neighborhood search.

4. The method for optimizing vehicle routes based on adaptive large neighborhood search and inventory shipyard according to claim 3 is characterized in that: The mathematical model of the interference event includes When a disruption event occurs, the objective function is to minimize the total cost, which includes the delivery route cost, vehicle cost, and delay cost; The constraints are interference events, vehicle self-rescue, dispatch of new vehicles and safety inventory, among which: Interference event constraints: Consider the impact of interference events on delivery routes. Vehicle self-rescue constraints: The vehicle can save itself and change its route when encountering interference. Constraint for dispatching new vehicles: When necessary, new vehicles can be dispatched to serve warehouses that have not been dispatched. Safety stock constraint: Ensure that the inventory level in each warehouse does not fall below the safety stock value.

5. The method for optimizing vehicle routes based on adaptive large neighborhood search and inventory shipyard according to claim 4, characterized in that: Determine the warehouse replenishment quantity and vehicle capacity based on the optimal performance strategy to determine the warehouse distribution combination; The distribution combination of warehouses is adjusted according to the distance between warehouses, and then the order of warehouses is determined according to the initial inventory of warehouses to integrate vehicle distribution routes; Determine whether a disruption event has occurred on the delivery vehicle; If there is no interference incident with the delivery vehicle, re-plan the delivery route to minimize the total cost; If an interference event occurs to a delivery vehicle, the type of interference event is determined, and the location of each delivery vehicle is determined when the interference event occurs, so that the corresponding recovery strategy can be adopted to determine the recovery scheduling plan.

6. The method for optimizing vehicle routes based on adaptive large neighborhood search and inventory shipyard according to claim 5, characterized in that: Formulate an optimal warehouse warning value, mark the optimal warehouse warning value as a preset warning value, and formulate different recovery strategies based on the preset warning value, wherein the recovery strategies include a first strategy and a second strategy; If the warehouse warning value is less than the preset warning value, the first strategy is selected, the vehicle self-rescue by changing the delivery route; If the warehouse warning value is greater than the preset warning value, the second strategy is selected to dispatch a new vehicle, which is suitable for situations where the parking waiting time is too long or there is no reasonable new route.

7. The method for optimizing vehicle routes based on adaptive large neighborhood search and inventory shipyard according to claim 6, characterized in that: The hybrid collaborative framework based on large neighborhood search includes: a route generation algorithm, an algorithm for determining warehouse replenishment based on a performance value adjustment strategy, and a large neighborhood removal and repair operator based on this problem.

Images