Public-water combined transportation inventory path optimization method based on adaptive large-scale neighborhood search

Abstract

The invention relates to a public-water combined transport inventory path optimization method based on adaptive large-scale neighborhood search. The method comprises the following steps: S1, constructing an inventory routing decision model; s2, constructing a transportation volume model; s3, constructing an initial solution; s4, an operator selection mechanism; s5, repairing a judgment mechanism; s6, designing an operator; s7, a solution accepting strategy; s8, an algorithm stopping condition; according to the public-water combined transportation inventory path optimization method, the public-water combined transportation scene of the commercial vehicle, the route scheme of comprehensive transportation and inventory management are considered for overall optimization, and the inventory path optimization problem is divided into a transportation route optimization module and a transportation volume optimization module; a transportation route optimization module with high solving difficulty of an accurate algorithm is solved by using a heuristic method, and a transportation volume module with relatively easy solving is solved by using the accurate algorithm, so that the efficiency and quality of the algorithm are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of public water combined transport inventory path optimization, and in particular to a public water combined transport inventory path optimization method based on adaptive large-scale neighborhood search. BACKGROUND

[0002] With the global automotive industry electrification transformation, China's new energy vehicle export volume has increased by more than 50% for three consecutive years, and the export volume will break through 2 million in 2024, accounting for more than 10% of global new energy vehicle trade volume. The cost pressure of the traditional logistics mode mainly based on highway transportation is highlighted, and it has been unable to meet the transportation demand in some scenarios.

[0003] The domestic "Transportation Power Construction Program" promotes the transfer of long-distance freight transportation to railways and waterways. In whole vehicle transportation, the multimodal transport mode combining roll-on / roll-off ships and highway transportation has become a trend. As a special transport ship, the design capacity of roll-on / roll-off ships is much higher than that of branch line transport vehicles. The maximum design capacity of sea roll-on / roll-off ships reaches 8000 vehicles, and the single transport capacity of inland roll-on / roll-off ships also reaches 800 vehicles, which can effectively reduce the cost of trunk line transportation. Highway transportation can realize door-to-door distribution and make up for the lack of flexibility of waterway transportation.

[0004] However, this two-stage transportation form combining waterway and highway brings difficulties to overall planning. As a central warehouse, the center warehouse needs to consider inventory control and trunk and branch line transport capacity planning; trunk line transportation involves ship scheduling, transport capacity and route arrangement; branch line transportation considers branch line transport vehicle route and transport capacity. Therefore, it is urgent to develop a trunk and branch line transport scheme for two-level inventory path optimization to reduce short-term transportation and inventory costs.

[0005] Therefore, the present application proposes a public water combined transport inventory path optimization method based on adaptive large-scale neighborhood search to solve the above problems. SUMMARY

[0006] The technical problem to be solved by the present application is to provide a public water combined transport inventory path optimization method based on adaptive large-scale neighborhood search to reduce the short-term transportation and inventory costs of vehicle public water combined transport.

[0007] To solve the above technical problems, the technical scheme of the present application is as follows: a public water combined transport inventory path optimization method based on adaptive large-scale neighborhood search, the innovation point of which is as follows: S1, inventory routing decision model construction: based on problem description, taking trunk line transportation cost, branch line transportation cost, central warehouse inventory cost and store inventory cost as objective function, an inventory routing decision model is constructed by minimizing the objective function; S2, transportation quantity model construction: under the inventory routing decision model, fix the route related variables, construct the transportation quantity model after parameterization, relax the constraint that the demand of the store must be fully met within the cycle, and set the penalty coefficient, the transportation quantity model takes the minimization of the trunk transportation cost, inventory cost and penalty cost as the objective function; S3, initial solution construction: based on the problem description, design and construct heuristic algorithm to generate initial feasible solution; S4, operator selection mechanism: apply the destruction operator, trunk repair operator, branch repair operator and transportation quantity model to an original solution to obtain a new solution, the operator selection uses the roulette mechanism, and the weights of the destruction operator and the repair operator are associated, the trunk repair operator and the branch repair operator are selected under the consideration of the used destruction operator; S5, repair judgment mechanism: the branch transportation problem is regarded as a one-dimensional knapsack problem, the latest delivery cycle of the trunk is obtained by minimizing the number of cycles with transportation, and the latest trunk delivery cycle model is established; S6, operator design: according to the characteristics that the departure time of the trunk transportation ship affects the transportation quantity along the way and the time when the central warehouse receives goods from the trunk, the destruction operator is designed through the departure time of the trunk transportation ship, and according to the characteristics that the trunk transportation may pass through other nodes or appear return transportation on the way to the next node, the distance-based destruction operator and repair operator are designed; S7, solution acceptance strategy: in each iteration, the simulated annealing framework is used to decide whether the new solution replaces the original solution to continue iteration; S8, algorithm stopping condition: the comprehensive condition is used to judge whether the algorithm reaches the stopping condition, when any condition in the comprehensive condition is met, the algorithm will end running, and the optimal solution of the cost obtained is taken as the overall optimal solution.

[0008] Further, the expression of the minimization objective function of the inventory routing decision model is: Formula 1: Wherein, is the trunk transportation cost, is the branch transportation cost, is the central warehouse inventory cost, is the store inventory cost; The calculation formula of the trunk transportation cost is: Formula 2: Wherein, is the trunk transportation ship whether it departs from the node in the cycle goes to the node ; is the trunk transportation ship Whether to proceed with transportation; For mainline transport vessels In the cycle Load capacity; For the trunk line from the node Transport unit products to nodes The cost per cycle; The fixed cost of using a mainline transport vessel; For the set of base warehouse nodes; For the central warehouse node set; For assembling mainline transport vessels; It is a periodic set; The formula for calculating feeder transport costs is: Formula 3: in, For branch line transport vehicles Is it in the cycle? From node Transported to the node ; For the branch line from the node Transported to node The cost; A collection of store nodes; For the collection of feeder transport vehicles; The formula for calculating the cost of central warehouse inventory is: Formula 4: in, For nodes In the cycle Inventory levels; For nodes Single-cycle inventory cost per unit of product held; The formula for calculating store inventory costs is: Formula 5: in, For stores In the cycle The amount of stock shortage; For stores Cost per unit of product per single-cycle stockout.

[0009] Furthermore, the inventory routing decision model includes the following constraints: Constraint 1: In trunk line transportation, it is necessary to ensure line flow balance, expressed as: Formula 6: in, For the trunk line from the node Transported to node The required number of cycles; Constraint 2: Mainline transport vessels have capacity constraints; the transport volume cannot exceed the mainline transport vessel's capacity. The expression is: Formula 7: in, This is the maximum capacity of a mainline transport vessel; Constraint 3: It is necessary to ensure the balance of transport volume on the mainline transport vessels, as expressed in the following expression: Formula 8: in, As auxiliary parameters, The value is 1 when it is a base warehouse and -1 when it is a central warehouse; For period Mainline transport ships At the node Loading / unloading volume; Constraint 4: Loading and unloading of cargo can only take place when the main transport vessel reaches the base warehouse or central warehouse. The expression is as follows: Formula 9: Constraint 5: In branch line transportation, route continuity must be ensured, as expressed in the following expression: Formula 10: in, For stores Whether it is fixed depends on the central warehouse node Supply; Constraint 6: Stores must be supplied by a fixed central warehouse, expressed as: Formula 11: Constraint 7, simultaneously avoiding the occurrence of sub-loops, is expressed as: Formula 12: in, As an auxiliary variable used for sub-loop elimination, it represents the branch line transport vehicle. In the cycle Reaching the node The cumulative access order at that time; Represent a constant, taking values ; Constraint 8: Within the same period, a store can be visited at most once, expressed as: Formula 13: Constraint 9: Branch line transport vehicles visiting stores must depart from the central warehouse to which the store belongs. The expression is: Formula 14: Constraint 10: A feeder transport vehicle can only serve one central warehouse, meaning it can only depart from one central warehouse at most once. The expression is: Formula 15: Constraint 11: Feeder vehicles used for feeder transport have a capacity limit, expressed as: Formula 16: in, For stores In the cycle Quantity of goods received; Maximum capacity of feeder transport vehicles Constraint 12: Branch route decisions must be consistent with transport volume decisions, expressed as: Equation 17: Constraint 13: The central warehouse is set with a certain storage capacity, which must satisfy the maximum storage capacity constraint, expressed as: Formula 18: in, Central warehouse Maximum capacity; Constraint 14: Stores are allowed to have stockouts, but stockouts must be replenished, and all demand must be met within the planning period. The expression is: Formula 19: in, For stores In the cycle Demand; Constraint 15: Each node must satisfy the inventory balance constraint, expressed as: Formula 20: Equation 21: Equation 22: in, For the base warehouse Maximum output in a single cycle.

[0010] Furthermore, the objective function expression of the transport volume model is: Equation 23: in, This represents the cost per unit of cargo required by a mainline transport vessel for each transport cycle. Penalty cost for store ending inventory; Penalty costs for store stockouts at the end of the period; This refers to the store's ending inventory. Due to stockouts at the end of the period; The constraints of the transport volume model include constraints 2-4 and 11-15, and also apply to the variables in the constraints. , Perform parameterization processing.

[0011] Furthermore, the heuristic algorithm is divided into a matching module, a branch line transportation module, a trunk line transportation module, and a transportation volume model application module; The matching module randomly matches each store with the nearest central warehouse whose turnover capacity is not fully utilized, until all stores are matched to the central warehouse. The branch line transportation module arranges branch line transportation vehicles for each central warehouse starting from the last cycle to transport goods to the matching stores. The transportation volume does not exceed the capacity of the branch line transportation vehicles and the remaining demand of the stores, until the demand of all stores can be met through the central warehouse. Starting from the first cycle, the trunk transportation module sequentially searches for central warehouses that are experiencing shortages based on the current transportation volume, and arranges for trunk transport vessels currently in transit to transport goods to the central warehouses, or arranges for new trunk transport vessels to load goods at the nearest base warehouse before transporting them to the central warehouses. The transportation volume model application module obtains a complete initial feasible solution by parameterizing the routing decision and solving it using the transportation volume model without changing the routing decision.

[0012] Furthermore, the roulette wheel betting mechanism selects an operator each time based on operator weights, with the operator weights determined by the operator in each round. The performance of the generation is determined, in The number of times the operator is applied in each iteration Applying operators yields solutions with lower costs. The number of feasible solutions obtained by applying the operator Then the operator will be used in the following... The weights in the next iteration are set to , The operator weights will be recalculated after each iteration.

[0013] Furthermore, in the latest trunk delivery cycle model, if no trunk route visits the central warehouse before the latest delivery cycle, the trunk route decision is deemed infeasible, and the central warehouse node needs to be re-inserted into the trunk route, ensuring at least one visit before the latest delivery cycle. A store can only be visited once in the same cycle, while the branch line transport vehicles have capacity limitations. Therefore, the minimum number of visits required by the store can be obtained. If the number of visits by the store in the branch route is less than the minimum number of visits, the store needs to be re-inserted into the branch route, ensuring that the number of visits is not less than the minimum number of visits. The objective function of the latest trunk line delivery cycle model is to minimize the central warehouse transportation cycle, expressed as: Formula 24: in, This indicates whether the central warehouse performs deliveries in each cycle; The latest trunk line delivery cycle model has the following constraints: Constraint 16: All store requirements matched by the central warehouse must be satisfied, expressed as: Formula 25: in, The volume of goods transported from the central warehouse to each store in each cycle; For store demand; Constraint 17: The feeder transport vehicles used for branch line delivery in the central warehouse have a vehicle capacity limit, expressed as: Equation 26: .

[0014] Furthermore, in the simulated annealing framework, if a new solution has a better cost than the original solution, the original solution will definitely be accepted; otherwise, there is a probability of accepting a worse solution and continuing the iteration. The probability of accepting a solution is... Determined by the current temperature, the expression is: Equation 27: in, The original objective function value. The new objective function value; The temperature update strategy uses an exponential method, expressed as: Equation 28: ; in, For the new temperature; The original temperature; This represents the rate of temperature decrease.

[0015] Furthermore, the comprehensive conditions include: (1) The optimal solution has not been improved after a specified number of iterations; (2) The number of operator weight adjustments reaches the specified value; (3) The algorithm reaches the specified maximum running time.

[0016] The advantages of this invention are: (1) The public-water intermodal transport inventory route optimization method of the present invention takes into account the public-water intermodal transport scenario of commercial vehicles and optimizes the route plan and inventory management of integrated transportation as a whole.

[0017] (2) This invention breaks down the inventory route optimization problem into a transportation route optimization module and a transportation volume optimization module. For the transportation route optimization module, which involves many variables and is difficult to solve with an exact algorithm, a heuristic method is used to solve it. For the transportation volume module, which is relatively easy to solve, an exact algorithm is used to solve it, thus ensuring the efficiency and quality of the algorithm. Attached Figure Description

[0018] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0019] Figure 1 This is a diagram illustrating the iterative process of the calculation example for the public-water intermodal transport inventory route optimization method of the present invention. Detailed Implementation

[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0021] Example This embodiment provides a method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search.

[0022] The problem of the road-water intermodal transport inventory routing method described in this paper is as follows: The logistics network consists of base warehouses, central warehouses, and stores. Base warehouses handle vehicle production and temporary storage, with certain production speed limitations. Central warehouses, as hubs for trunk and branch line transportation, handle vehicle storage and transfer, with certain inventory capacity limitations, and inventory costs need to be calculated. Stores sell vehicles, allowing for stockouts which must be replenished; supply and demand must be balanced within the planning period, meaning beginning inventory equals ending inventory. Store inventory and stockouts incur inventory costs and stockout costs, respectively.

[0023] Trunk line transportation is waterway transportation, using trunk line transport vessels for cross-cycle transport between base warehouses and central warehouses. This means that it takes several cycles for a trunk line transport vessel to travel from one node to another. Trunk line transport vessels have capacity limitations and are allowed to load goods at base warehouses or unload goods at central warehouses along the way. This is a cross-cycle picking and delivery transportation mode, and trunk line transportation costs are affected by the transportation cycle, volume, and distance. Feeder line transportation is road transportation, using feeder trucks to transport goods between central warehouses and stores. With vehicle capacity limitations, feeder trucks depart from the central warehouse each cycle, deliver goods to multiple stores, and return to the same central warehouse within the same cycle. Transportation costs are only affected by the transportation distance. In feeder line transportation, a store can only be visited once per cycle, and goods can only be supplied by one fixed central warehouse throughout all cycles.

[0024] The road-water intermodal transport inventory path optimization method based on adaptive large-scale neighborhood search includes the following steps: S1. Inventory routing decision model construction: Based on the above problem description, the inventory routing decision model is constructed by minimizing the objective function, with trunk line transportation cost, branch line transportation cost, central warehouse inventory cost and store inventory cost as the objective function. The expression for minimizing the objective function of the inventory routing decision model is: Formula 1: in, For trunk line transportation costs, For branch line transportation costs, For central warehouse inventory costs, For store inventory costs; The formula for calculating trunk line transportation costs is: Formula 2: in, For mainline transport vessels Is it in the cycle? Starting from node Transported to the node ; For mainline transport vessels Whether to proceed with transportation; For mainline transport vessels In the cycle Load capacity; For the trunk line from the node Transport unit products to nodes The cost per cycle; The fixed cost of using a mainline transport vessel; For the set of base warehouse nodes; For the central warehouse node set; For assembling mainline transport vessels; It is a periodic set; In feeder transportation, the cost of using feeder vehicles is related to the transportation distance. The formula for calculating feeder transportation costs is as follows: Formula 3: in, For branch line transport vehicles Is it in the cycle? From node Transported to the node ; For the branch line from the node Transported to node The cost; A collection of store nodes; For the collection of feeder transport vehicles; Central warehouses store and transfer vehicles, incurring inventory holding costs within their storage capacity. The formula for calculating central warehouse inventory costs is as follows: Formula 4: in, For nodes In the cycle Inventory levels; For nodes Single-cycle inventory cost per unit of product held; When a store sells vehicles, it needs to pay inventory holding costs when the inventory is above zero and stockout costs when there are stockouts. The formula for calculating store inventory costs is as follows: Formula 5: in, For stores In the cycle The amount of stock shortage; For stores Cost per unit of product per single-cycle stockout.

[0025] The inventory routing decision model includes the following constraints: Constraint 1: In trunk line transportation, it is necessary to ensure line flow balance, expressed as: Formula 6: in, For the trunk line from the node Transported to node The required number of cycles; Constraint 2: Mainline transport vessels have capacity constraints; the transport volume cannot exceed the mainline transport vessel's capacity. The expression is: Formula 7: in, This is the maximum capacity of a mainline transport vessel; Constraint 3: It is necessary to ensure the balance of transport volume on the mainline transport vessels, as expressed in the following expression: Formula 8: in, As auxiliary parameters, The value is 1 when it is a base warehouse and -1 when it is a central warehouse; For period Mainline transport ships At the node Loading / unloading volume; Constraint 4: Loading and unloading of cargo can only take place when the main transport vessel reaches the base warehouse or central warehouse. The expression is as follows: Formula 9: Constraint 5: In branch line transportation, route continuity must be ensured, as expressed in the following expression: Formula 10: in, For stores Whether it is fixed by the central warehouse Supply; Constraint 6: Stores must be supplied by a fixed central warehouse, expressed as: Formula 11: Constraint 7, simultaneously avoiding the occurrence of sub-loops, is expressed as: Formula 12: in, As an auxiliary variable used for sub-loop elimination, it represents the branch line transport vehicle. In the cycle Reaching the node The cumulative access order at that time; Represent a constant, taking values ; Constraint 8: Within the same period, a store can be visited at most once, expressed as: Formula 13: Constraint 9: Branch line transport vehicles visiting stores must depart from the central warehouse to which the store belongs. The expression is: Formula 14: Constraint 10: A feeder transport vehicle can only serve one central warehouse, meaning it can only depart from one central warehouse at most once. The expression is: Formula 15: Constraint 11: Feeder vehicles used for feeder transport have a capacity limit, expressed as: Formula 16: in, For stores In the cycle Quantity of goods received; Maximum capacity of feeder transport vehicles Constraint 12: Branch route decisions must be consistent with transport volume decisions, expressed as: Equation 17: Constraint 13: The central warehouse is set with a certain storage capacity, which must satisfy the maximum storage capacity constraint, expressed as: Formula 18: in, Central warehouse Maximum capacity; Constraint 14: Stores are allowed to have stockouts, but stockouts must be replenished, and all demand must be met within the planning period. The expression is: Formula 19: in, For stores In the cycle Demand; Constraint 15: Each node must satisfy the inventory balance constraint, expressed as: Formula 20: Equation 21: Equation 22: in, For the base warehouse Maximum output in a single cycle.

[0026] S2. Transportation Volume Model Construction: The inventory routing optimization problem requires solving both transportation volume decisions and routing decisions simultaneously. To facilitate operator design, the two types of decisions are distinguished. An adaptive large-scale neighborhood search algorithm is used to solve the routing decisions, and a transportation volume model is embedded within it to call the existing solver. The transportation volume model is constructed by fixing routing-related variables under the original inventory routing decision model and then parameterizing them. To allow for infeasible solutions in the algorithm and ensure its exploration space, the transportation volume model relaxes the constraint that all store demands must be met within the cycle and sets a penalty coefficient. The objective function of the transportation volume model is to minimize trunk transportation costs, inventory costs, and penalty costs. The objective function expression of the transportation volume model is: Formula 23: in, This represents the cost per unit of cargo required by a mainline transport vessel for each transport cycle. Penalty cost for store ending inventory; Penalty costs for store stockouts at the end of the period; This refers to the store's ending inventory. Due to stockouts at the end of the period; The constraints of the transportation volume model include constraints 2-4 and 11-15, which ensure the rationality of the transportation volume and also address the variables within the constraints. , Perform parameterization processing; Trunk transport route variables The parameterization process, that is, the mainline transport ship In the cycle Whether by node Transported to the node Processed as parameters to determine mainline transport vessels In the cycle Has the node been passed? This ensures that changes in the inventory of mainline transport vessels match the docking points of transport.

[0027] feeder transport route variables The parameterization process, that is, the branch line transport vehicle In the cycle Whether by node Transported to the node Processed as a store In the cycle Parameters indicating whether or not the vehicle has been accessed, and branch line transport vehicles. In the cycle Do you want to visit the store? These parameters ensure that changes in store inventory match the routes of feeder transportation.

[0028] S3. Initial Solution Construction: Based on the problem description, a heuristic algorithm is designed and constructed to generate the initial feasible solution. The heuristic algorithm is divided into a matching module, a branch line transportation module, a trunk line transportation module, and a transportation volume model application module. The matching module randomly matches each store with the nearest central warehouse that has insufficient turnover capacity, until all stores are matched to a central warehouse. The feeder transport module schedules feeder transport vehicles for each central warehouse starting from the last cycle to transport goods to the matching stores. The transport volume does not exceed the feeder transport vehicle capacity and the remaining demand of the stores, until all store demands can be met through the central warehouse. Starting from the first cycle, the trunk transportation module sequentially searches for central warehouses that are experiencing shortages based on the current transportation volume, and arranges for trunk transport vessels currently in transit to transport goods to the central warehouses, or arranges for new trunk transport vessels to load goods at the nearest base warehouse before transporting them to the central warehouses. The transportation volume model module parameterizes the routing decision and uses the transportation volume model to solve it without changing the routing decision, thus obtaining a complete initial feasible solution.

[0029] S4. Operator Selection Mechanism: This method distinguishes operators into destruction operators and repair operators. The repair operators include trunk line repair operators and branch line repair operators. A new solution is obtained by applying destruction operators, trunk line repair operators, branch line repair operators, and transportation volume models to an original solution.

[0030] The operator selection uses a roulette wheel gambling mechanism. Considering the different performance of the repair operator under different destruction operators, the weights of the destruction and repair operators are correlated. The trunk repair operator and the branch repair operator are selected based on the destruction operators used. The roulette wheel gambling mechanism selects an operator each time based on operator weights. The operator weights are determined by the operator in each round. The performance of the generation is determined, in The number of times the operator is applied in each iteration Applying operators yields solutions with lower costs. The number of feasible solutions obtained by applying the operator Then the operator will be used in the following... The weights in the next iteration are set to , The operator weights will be recalculated after each iteration.

[0031] S5. Repair Judgment Mechanism: Treat the branch line transportation problem as a one-dimensional knapsack problem, and obtain the latest delivery cycle of the trunk line by minimizing the number of transportation cycles, and establish a model for the latest trunk line delivery cycle. In the latest trunk delivery cycle model, if no trunk route visits the central warehouse before the latest delivery cycle, the trunk route decision is deemed infeasible. The central warehouse node needs to be re-inserted into the trunk route, ensuring at least one visit before the latest delivery cycle. A store can only be visited once in the same cycle, while the branch line transport vehicles have capacity limitations. Therefore, the minimum number of visits required for a store can be obtained. If the number of visits to a store in the branch route is less than the minimum number of visits, the store needs to be re-inserted into the branch route, ensuring that the number of visits is not less than the minimum number of visits.

[0032] The objective function of the latest trunk line delivery cycle model is to minimize the central warehouse transportation cycle, expressed as: Formula 24: in, This indicates whether the central warehouse performs deliveries in each cycle; The latest trunk line delivery cycle model has the following constraints: Constraint 16: All store requirements matched by the central warehouse must be satisfied, expressed as: Formula 25: in, The volume of goods transported from the central warehouse to each store in each cycle; For store demand; Constraint 17: The feeder transport vehicles used for branch line delivery in the central warehouse have a vehicle capacity limit, expressed as: Equation 26: .

[0033] The latest trunk line delivery cycle model determines the transportation volume of each store from the central warehouse in each cycle. And whether the central warehouse conducts deliveries in each cycle. By solving for the objective, we can deduce the latest cycle that the trunk route needs to access the central warehouse to satisfy the constraint that the central warehouse is not out of stock. This allows us to determine whether the central warehouse node and access cycle of the trunk route need to be re-inserted during the repair process.

[0034] S6. Operator Design: Based on the characteristics of trunk line transportation, this invention designs some unique destruction and repair operators. In trunk line transportation, the departure time of the trunk line transport vessel affects the transport volume along the route and the time it takes for the central warehouse to receive goods from the trunk line. Based on this characteristic, a destruction operator is designed using the departure time of the trunk line transport vessel. In the scenario of trunk line transportation along a river, trunk line transport vessels typically travel along the main river channel, and base warehouses and central warehouses are distributed along the river. Therefore, trunk line transportation may pass through other nodes or involve turning back on the way to the next node. Based on this characteristic, distance-based destruction and repair operators are designed to achieve better results.

[0035] The trunk transport vessel departure time disruption operator modifies the departure time of trunk transport vessels, affecting the arrival time of trunk transport vessels at all nodes along the route visited by the trunk transport vessels, thereby causing large-scale disruption to the original solution.

[0036] The distance-based destruction and repair operators mean that when a mainline transport ship visits multiple nodes, it inserts "connected" nodes that are close to the transport route but have not been visited into the transport route, and removes nodes that need to "detour," that is, nodes that are too far away from the nodes before and after them, from the transport route.

[0037] S7. Solution Acceptance Strategy: In each iteration, a simulated annealing framework is used to determine whether a new solution replaces the original solution for continued iteration. Within the simulated annealing framework, if the new solution has a better cost than the original solution, the original solution is always accepted; otherwise, there is a probability of accepting a worse solution for continued iteration. The probability of solution acceptance... Determined by the current temperature, the expression is: Equation 27: in, The original objective function value. The new objective function value; The temperature update strategy uses an exponential method, expressed as: Equation 28: ; in, For the new temperature; The original temperature; This represents the rate of temperature decrease.

[0038] S8. Algorithm Stopping Conditions: Synthesis conditions are used to determine if the algorithm has reached its stopping condition. When any one of the synthesis conditions is met, the algorithm will terminate and use the obtained cost-optimal solution as the overall optimal solution. Synthesis conditions include: (1) The optimal solution has not been improved after a specified number of iterations; (2) The number of operator weight adjustments reaches the specified value; (3) The algorithm reaches the specified maximum running time.

[0039] The following example uses 2 base warehouses, 2 central warehouses, and 5 stores to plan inventory routes over 3 cycles. Trunk line transport vessels are used to transport goods produced in the base warehouses to the central warehouses, and then feeder trucks transport goods stored in the central warehouses to the responsible stores in each cycle. The specific basic data used is shown in Table 1. Table 1. Basic Data Table for the Calculation Example Node number Node category Horizontal coordinate Vertical coordinate Commodity garage capacity (vehicles) Commodity garage initial inventory (vehicles) Production speed (vehicles / day) Unit inventory holding cost (yuan / vehicle·day) Unit shortage cost (yuan / vehicle·day) D0 Base warehouse 537 20 - 0 500 - - D1 Base warehouse 596 30 - 0 500 - - S2 Central warehouse 765 181 800 0 - 20 - S3 Central warehouse 108 89 800 0 - 20 - C4 Store 762 106 - 0 - 50 150 C5 Store 661 120 - 0 - 50 150 C6 Store 958 68 - 0 - 50 150 C7 Store 159 99 - 0 - 50 150 C8 Store 451 87 - 0 - 50 150 The nodes are distributed on a map with dimensions of 1000km x 200km, and the distances between them are Euclidean distances. The mainline transport ship has a transport speed of 500km / day, and any transport time between nodes less than one day is counted as one day. The maximum warehouse capacity using the mainline transport ship is 800 vehicles, and the transport cost is 0.8 yuan / vehicle / day. The maximum capacity using the feeder transport vehicle is 10 vehicles, and the unit distance cost for feeder transport is 3 yuan / km. The demand for vehicles at each store in different periods is shown in Table 2. Table 2 Store Demand Table for Each Period Store number Period 0 demand (vehicles) Period 1 demand (vehicles) Period 2 demand (vehicles) C4 1 2 1 C5 2 1 3 C6 5 5 3 C7 3 3 5 C8 5 0 2 In the algorithm parameters, set the temperature decrease rate. The initial temperature is 0.98, the target temperature is the initial solution value, and the ending temperature is 0.99. The operator weights are updated once every 50 iterations. The algorithm's termination conditions include: the optimal solution remaining unchanged for 20,000 consecutive iterations; the maximum algorithm runtime is 600 seconds; and the maximum number of operator weight updates is 50.

[0040] The above example was calculated using the public-water intermodal transport inventory route optimization method of the present invention, and the calculation results are shown in Tables 3-5.

[0041] Table 3 shows the cost results obtained from the solution. Total cost (yuan) 21,478 Trunk transportation cost (yuan) 11,114 Branch transportation cost (yuan) 7,713 Central warehouse inventory cost (yuan) 0 Store inventory cost (yuan) 2,650 Table 4 Optimized Trunk Transportation Scheme Trunk transportation vehicle number Access period Access node Loading capacity (vehicles) 1 0 D0 +6 1 1 S3 -6 2 0 D1 +26 2 1 S2 -26 3 1 D0 +5 3 2 S3 -5 4 1 D1 +4 4 2 S2 -4 Table 5 Optimized feeder transport schemes Branch transportation vehicle number Belonging central warehouse Transportation period Route 1 S2 1 C6 2 S2 1 C8 3 S2 1 C4→C5 4 S2 2 C4→C6 5 S3 1 C7 6 S3 2 C7 The computational example of the public-water intermodal transport inventory path optimization method of this invention has a running time of only 52 seconds and a final total cost of 21,478 yuan. The target value decreases during the iteration process as follows: Figure 1 As shown, Figure 1 The horizontal axis represents the number of iterations, and the vertical axis represents the total computation cost.

[0042] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search, characterized by: Includes the following steps: S1. Inventory routing decision model construction: Based on the problem description, the inventory routing decision model is constructed by minimizing the objective function, with trunk line transportation cost, branch line transportation cost, central warehouse inventory cost and store inventory cost as the objective function. S2. Transportation volume model construction: Under the inventory routing decision model, fix the routing-related variables, and construct the transportation volume model after parameterization. The transportation volume model relaxes the constraint that store demand must be fully satisfied within the cycle and sets a penalty coefficient. The transportation volume model takes minimizing trunk transportation cost, inventory cost and penalty cost as the objective function. S3. Initial Solution Construction: Based on the problem description, design and construct a heuristic algorithm to generate an initial feasible solution; S4. Operator selection mechanism: Apply destruction operators, trunk line repair operators, branch line repair operators, and transport volume models to an existing solution to obtain a new solution. Operator selection uses a roulette wheel betting mechanism, which associates the weights of destruction operators and repair operators, and selects trunk line repair operators and branch line repair operators while considering the destruction operators used. S5. Repair Judgment Mechanism: Treat the branch line transportation problem as a one-dimensional knapsack problem, and obtain the latest delivery cycle of the trunk line by minimizing the number of transportation cycles, and establish a model for the latest trunk line delivery cycle. S6. Operator Design: Based on the characteristic that the departure time of the trunk transport vessel affects the transport volume along the route and the time when the central warehouse receives the goods from the trunk, a destruction operator is designed based on the departure time of the trunk transport vessel. Based on the characteristic that trunk transport may pass through other nodes or make a turnaround on the way to the next node, a distance-based destruction operator and repair operator are designed. S7. Solution acceptance strategy: In each iteration, the simulated annealing framework is used to determine whether a new solution replaces the original solution to continue the iteration; S8. Algorithm Stopping Condition: Use the synthesis condition to determine whether the algorithm has reached the stopping condition. When any of the synthesis conditions is met, the algorithm will end its operation and take the obtained cost-optimal solution as the overall optimal solution.

2. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 1, characterized in that: The expression for the minimization objective function of the inventory routing decision model is: Formula 1: in, For trunk line transportation costs, For branch line transportation costs, For central warehouse inventory costs, For store inventory costs; The formula for calculating trunk line transportation costs is: Formula 2: in, For mainline transport vessels Is it in the cycle? Starting from node Transported to the node ; For mainline transport vessels Whether to proceed with transportation; For mainline transport vessels In the cycle Load capacity; For the trunk line from the node Transport unit products to nodes The cost per cycle; The fixed cost of using a mainline transport vessel; For the set of base warehouse nodes; For the central warehouse node set; For assembling mainline transport vessels; It is a periodic set; The formula for calculating feeder transport costs is: Formula 3: in, For branch line transport vehicles Is it in the cycle? From node Transported to the node ; For the branch line from the node Transported to node The cost; A collection of store nodes; For the collection of feeder transport vehicles; The formula for calculating the cost of central warehouse inventory is: Formula 4: in, For nodes In the cycle Inventory levels; For nodes Single-cycle inventory cost per unit of product held; The formula for calculating store inventory costs is: Formula 5: in, For stores In the cycle The amount of stock shortage; For stores Cost per unit of product per single-cycle stockout.

3. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 2, characterized in that: The inventory routing decision model includes the following constraints: Constraint 1: In trunk line transportation, it is necessary to ensure line flow balance, expressed as: Formula 6: in, For the trunk line from the node Transported to node The required number of cycles; Constraint 2: Mainline transport vessels have capacity constraints; the transport volume cannot exceed the mainline transport vessel's capacity. The expression is: Formula 7: in, This is the maximum capacity of a mainline transport vessel; Constraint 3: It is necessary to ensure the balance of transport volume on the mainline transport vessels, as expressed in the following expression: Formula 8: in, As auxiliary parameters, The value is 1 when it is a base warehouse and -1 when it is a central warehouse; For period Mainline transport ships At the node Loading / unloading volume; Constraint 4: Loading and unloading of cargo can only take place when the main transport vessel reaches the base warehouse or central warehouse. The expression is as follows: Formula 9: Constraint 5: In branch line transportation, route continuity must be ensured, as expressed in the following expression: Formula 10: in, For stores Whether it is fixed depends on the central warehouse node Supply; Constraint 6: Stores must be supplied by a fixed central warehouse, expressed as: Formula 11: Constraint 7, simultaneously avoiding the occurrence of sub-loops, is expressed as: Formula 12: in, As an auxiliary variable used for sub-loop elimination, it represents the branch line transport vehicle. In the cycle Reaching the node The cumulative access order at that time; Represent a constant, taking values ; Constraint 8: Within the same period, a store can be visited at most once, expressed as: Formula 13: Constraint 9: Branch line transport vehicles visiting stores must depart from the central warehouse to which the store belongs. The expression is: Formula 14: Constraint 10: A feeder transport vehicle can only serve one central warehouse, meaning it can only depart from one central warehouse at most once. The expression is: Formula 15: Constraint 11: Feeder vehicles used for feeder transport have a capacity limit, expressed as: Formula 16: in, For stores In the cycle Quantity of goods received; Maximum capacity of feeder transport vehicles Constraint 12: Branch route decisions must be consistent with transport volume decisions, expressed as: Equation 17: Constraint 13: The central warehouse is set with a certain storage capacity, which must satisfy the maximum storage capacity constraint, expressed as: Formula 18: in, Central warehouse Maximum capacity; Constraint 14: Stores are allowed to have stockouts, but stockouts must be replenished, and all demand must be met within the planning period. The expression is: Formula 19: in, For stores In the cycle Demand; Constraint 15: Each node must satisfy the inventory balance constraint, expressed as: Formula 20: Equation 21: Equation 22: in, For the base warehouse Maximum output in a single cycle.

4. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 3, characterized in that: The objective function expression for the transportation volume model is: Equation 23: in, This represents the cost per unit of cargo required by a mainline transport vessel for each transport cycle. Penalty cost for store ending inventory; Penalty costs for store stockouts at the end of the period; This refers to the store's ending inventory. Due to stockouts at the end of the period; The constraints of the transport volume model include constraints 2-4 and 11-15, and also apply to the variables in the constraints. , Perform parameterization processing.

5. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 1, characterized in that: The heuristic algorithm is divided into a matching module, a branch line transportation module, a trunk line transportation module, and a transportation volume model application module. The matching module randomly matches each store with the nearest central warehouse whose turnover capacity is not fully utilized, until all stores are matched to the central warehouse. The branch line transportation module arranges branch line transportation vehicles for each central warehouse starting from the last cycle to transport goods to the matching stores. The transportation volume does not exceed the capacity of the branch line transportation vehicles and the remaining demand of the stores, until the demand of all stores can be met through the central warehouse. Starting from the first cycle, the trunk transportation module sequentially searches for central warehouses that are experiencing shortages based on the current transportation volume, and arranges for trunk transport vessels currently in transit to transport goods to the central warehouses, or arranges for new trunk transport vessels to load goods at the nearest base warehouse before transporting them to the central warehouses. The transportation volume model application module obtains a complete initial feasible solution by parameterizing the routing decision and solving it using the transportation volume model without changing the routing decision.

6. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 1, characterized in that: The roulette wheel betting mechanism selects an operator each time based on operator weights. The operator weights are determined by the operator in each round. The performance of the generation is determined, in The number of times the operator is applied in each iteration Applying operators yields solutions with lower costs. The number of feasible solutions obtained by applying the operator Then the operator will be used in the following... The weights in the next iteration are set to , The operator weights will be recalculated after each iteration.

7. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 1, characterized in that: In the latest trunk delivery cycle model, if no trunk route visits the central warehouse before the latest delivery cycle, the trunk route decision is deemed infeasible. The central warehouse node needs to be re-inserted into the trunk route, ensuring at least one visit before the latest delivery cycle. A store can only be visited once in the same cycle, while the branch line transport vehicles have capacity limitations. Therefore, the minimum number of visits required for a store can be obtained. If the number of visits to a store in the branch route is less than the minimum number of visits, the store needs to be re-inserted into the branch route, ensuring that the number of visits is not less than the minimum number of visits. The objective function of the latest trunk line delivery cycle model is to minimize the central warehouse transportation cycle, expressed as: Formula 24: in, This indicates whether the central warehouse performs deliveries in each cycle; The latest trunk line delivery cycle model has the following constraints: Constraint 16: All store requirements matched by the central warehouse must be satisfied, expressed as: Formula 25: in, The volume of goods transported from the central warehouse to each store in each cycle; For store demand; Constraint 17: The feeder transport vehicles used for branch line delivery in the central warehouse have a vehicle capacity limit, expressed as: Equation 26: .

8. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 1, characterized in that: In the simulated annealing framework, if a new solution has a better cost than the original solution, the original solution is always accepted; otherwise, there is a probability of accepting a worse solution and continuing the iteration. The probability of accepting a solution is... Determined by the current temperature, the expression is: Equation 27: in, The original objective function value. The new objective function value; The temperature update strategy uses an exponential method, expressed as: Equation 28: ; in, For the new temperature; The original temperature; This represents the rate of temperature decrease.

9. The method for optimizing the inventory path of public-water intermodal transport based on adaptive large-scale neighborhood search according to claim 1, characterized in that: The comprehensive conditions include: (1) The optimal solution has not been improved after a specified number of iterations; (2) The number of operator weight adjustments reaches the specified value; (3) The algorithm reaches the specified maximum running time.

Images