A logistics park pickup vehicle scheduling method based on improved SPEA2

Abstract

The present application relates to a kind of based on improved SPEA2 logistics park pickup vehicle scheduling method.Based on the multi-objective optimization algorithm of SPEA2, logistics park scheduling time and pickup vehicle operation time are optimized simultaneously.First, by finding the placement of target goods, according to the running speed of yard crane and the placement of goods, the pickup time of pickup vehicle in each yard is calculated.Secondly, the insertion decoding method is used to insert the pickup vehicle into the idle time of yard.In addition, the neighborhood search strategy of exchanging critical path is also used to select the pickup vehicle operation on the critical path for exchange in the iteration process of algorithm.This method solves the pickup vehicle scheduling problem of logistics park with mixed storage structure, effectively utilizes the idle time of yard by insertion decoding, and avoids falling into local optimal solution by critical path neighborhood search, thereby reducing the running time of logistics park and pickup vehicle operation time, and improving the efficiency of pickup vehicle of logistics park.

Description

Technical Field

[0001] This invention relates to the field of pickup vehicle scheduling, and in particular to a logistics park pickup vehicle scheduling method based on an improved SPEA2. Background Technology

[0002] Currently, my country's steel production ranks first in the world. However, according to statistics from relevant institutions, my country's steel logistics costs are 5% higher than those of developed countries, reaching 11%. With the significant increase in the throughput of the steel logistics industry, higher demands are placed on logistics efficiency. Among these demands, the scheduling of pickup vehicles in logistics parks is a key factor determining scheduling efficiency and logistics throughput. Logistics parks with mixed storage structures, where each yard contains different types of steel products, provide flexible yard selection for each pickup operation. In this case, the yard allocation and loading sequence of each vehicle must be considered simultaneously in pickup vehicle scheduling, which greatly increases the complexity of scheduling.

[0003] With the development of IoT technology, effectively acquiring vehicle-related information has become possible. In this context, it is crucial to utilize this data to optimize scheduling mechanisms to meet the operational needs of both vehicles and parking lots. Currently, the management of pickup vehicle queues at the park largely relies on manual, first-come, first-served scheduling. Due to a lack of scientific optimization, this leads to scheduling chaos and low customer satisfaction. Therefore, developing a pickup vehicle scheduling algorithm that simultaneously considers the efficiency of both the logistics park's overall scheduling and the efficiency of pickup vehicle scheduling is essential.

[0004] Traditional pickup vehicle scheduling algorithms typically only consider the operational efficiency of the logistics park while neglecting the operational efficiency of the pickup vehicles themselves, and they are unsuitable for scheduling pickup vehicles in logistics parks with mixed storage structures. This invention, based on the SPEA2 multi-objective optimization algorithm, aims to minimize both the logistics park completion time and the pickup vehicle time. First, an insertion-based decoding method is proposed to optimize the pickup vehicle sequence and reduce yard idle time. Second, a neighborhood search strategy for the critical path is proposed to improve the quality of the solution.

[0005] Chinese Patent CN103745021A, entitled "Comprehensive Planning Method for Vehicle Pickup Routes," discloses a comprehensive planning method for pickup vehicle routes. Based on an improved CW route planning algorithm incorporating time windows and vehicle type constraints, it uses distance from the OEM and pickup frequency as constraints to select suppliers. The improved CW algorithm is then used to solve the problem, yielding the corresponding planned routes and vehicle loading rates. However, this route planning does not consider the mixed-loading characteristics of logistics park yards. Each yard in a logistics park uses mixed steel storage, containing various types of steel, allowing pickup trucks to retrieve target goods from different yards. Therefore, this method is not suitable for solving the pickup vehicle scheduling problem in logistics parks with mixed-loading structures.

[0006] Chinese Patent CN202211168409.5, entitled "A Method and System for Batch Scheduling of Vehicles in Logistics Parks Considering Multiple Pickup Locations," discloses a method for batch scheduling of vehicles in logistics parks considering multiple pickup locations. This method acquires the operational characteristics of vehicles with multiple pickup locations in the logistics park and constructs a vehicle batch scheduling optimization model with the objective function of minimizing the logistics park's scheduling time. Based on this optimization model, a heuristic algorithm is used to solve for the optimal vehicle batching scheme and the corresponding scheduling plan. However, this method does not consider the scheduling efficiency of pickup vehicles. In the process of picking up goods in a logistics park, the scheduling efficiency of customer pickup vehicles is a key factor determining customer satisfaction; therefore, this method is not applicable to this invention. Summary of the Invention

[0007] The purpose of this invention is to overcome the problem that current steel logistics park pickup vehicle scheduling requires simultaneous consideration of each vehicle's yard allocation and loading sequence, which greatly increases scheduling complexity. This invention provides a logistics park pickup vehicle scheduling method based on an improved SPEA2 algorithm. Based on SPEA2's multi-objective optimization algorithm, it optimizes both the logistics park scheduling time and the pickup vehicle operation time. First, the placement location of the target goods for the pickup vehicle is located, and the pickup time of the pickup vehicle in each yard is calculated based on the yard crane's operating speed and the goods placement location. Second, for the algorithm solution, an insertion-based decoding method is proposed, inserting pickup vehicles into idle time in the yard. Furthermore, a neighborhood search strategy for exchanging critical paths is proposed. Based on the critical paths in the pickup vehicle scheduling, during the algorithm's iteration process, pickup vehicle operations on the critical paths are selected for exchange. This invention solves the problem of scheduling pickup vehicles in logistics parks with mixed storage structures. By inserting decoding, it can effectively utilize the idle time of the yard. By using the critical path neighborhood structure, it avoids the obtained feasible solution from getting trapped in local optima, reduces the operation time of the logistics park and the operation time of pickup vehicles, and promotes the efficiency of pickup vehicles in the logistics park.

[0008] To achieve the above objectives, the technical solution of this invention is: a logistics park pickup vehicle scheduling method based on an improved SPEA2, which calculates the pickup time of pickup vehicles in each yard by finding the placement location of the target goods and based on the operating speed of the yard cranes and the placement location of the goods; it employs an insertion decoding method to insert pickup vehicles into the idle time of the yard; furthermore, it adopts a neighborhood search strategy for exchanging critical paths to select pickup vehicle operations on the critical path for exchange. The specific implementation of this method is as follows:

[0009] S1. Determine the placement location of the target goods for the pickup vehicle in each storage yard, based on... Determine the pickup time t for each pickup vehicle in each yard, where d yThe distance between the storage location of the goods and the pick-up point in the yard is represented by v, the speed of the crane is represented by a, the time it takes for the crane to pick up one unit of goods is represented by a, and N is represented by the quantity of goods.

[0010] S2. Initialize the number of iterations q, and the maximum number of iterations q max Nearest neighbor threshold k, elite archive size η, population size NP, crossover probability CP, mutation probability MP;

[0011] S3. Based on the information of the pickup vehicles in the logistics park on the same day, the pickup vehicle scheduling plan is particle-coded. A two-level coding method is adopted. The operation sequence chain OSC represents the order of vehicle operations, and its length is equal to the total number of vehicle operations. The yard allocation chain YAC is used to allocate yards for each vehicle operation. Each gene of YAC is arranged according to the vehicle index and the order of vehicle operations. Each integer represents the yard number currently allocated to the corresponding operation in the set of available yards.

[0012] S4. Randomly allocate the available storage yards to each pickup vehicle operation, and generate an initial population P according to the designed coding method. q Generate an elite archive A with a size of η. q ;

[0013] S5, iterative loop q max Second-rate;

[0014] S6. Convert the yard allocation chain YAC into the yard allocation matrix Y. m [i,j] and the time matrix T[i,j], Y m [i,j] represents the yard allocated to the j-th pickup operation of vehicle i, and T[i,j] represents the time required for the j-th pickup of vehicle i;

[0015] S7. Decode OSC, if O i,j This is the first operation of vehicle i and is in yard M. k The first operation is the corresponding operation O. i,j Process directly at time 0 of the corresponding stockpile; if O i,j It is M k If it is the first operation of vehicle i, but not the first operation of vehicle i, then it is the previous operation O. i,j-1 Insert O directly after the completion time. i,j Otherwise, locate storage yard M. k All free time periods [TS] x ,TE x ], where TS x It is the start time of the x-th idle time period, TE x It is the end time of the x-th idle period; in yard M k During the work process, determine Oi,j Does the insertion condition max{C} satisfy? i,j-1 ,TS x}+t i,j,k ≤TE x C i,j-1 t is the completion time of the (j-1)th pickup operation of the vehicle. i,j,k It is the pickup time of the j-th operation of pickup vehicle i in yard k; if satisfied, it is inserted into the idle time [TS]. x ,TE x In the context of [ ], the start time is min{C i,j-1 ,LM k}, where LM k It is the current stockpile M k The earliest available time;

[0016] S8. Based on the decoding result of S7, calculate the logistics park's operating time C. max Maximum running time of pickup vehicle SC max ;

[0017] S9. Based on the objective function value C max With SC max Calculate the fitness value F for each individual in the population;

[0018] S10. Based on the fitness value F, from P q With A q Before environmental selection, the top η individuals are entered into Elite Profile A. q+1 ;

[0019] S11, Regarding A in the elite files q+1 Each individual undergoes crossover and mutation according to the crossover probability CP and the mutation probability MP. OSC uses POX crossover and two-point mutation, while YAC uses two-point crossover and two-point mutation. The results of the crossover and mutation are saved into P. q+1 ;

[0020] S12. Find the longest path among all paths from the start node to the end node, perform a neighborhood search by moving the pickup vehicle on the critical path, and obtain a new chromosome.

[0021] S13. Determine if the number of iterations has reached the maximum number of iterations q. max If the condition is met, stop the iteration and output the globally optimal position P. g This is used as the optimal scheduling result.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] (1) The designed insertion decoding method, under the premise of meeting the vehicle pickup timing requirements, dynamically inserts the pickup vehicles into the idle time periods of the yard that meet the insertion conditions by adjusting their order. This method can flexibly arrange the insertion time and position of vehicles to maximize the utilization of the yard's idle time. By implementing reasonable vehicle scheduling and insertion strategies, the utilization rate of the yard's idle time is effectively improved, unnecessary waiting and waste of idle time are avoided, thereby optimizing the overall efficiency of vehicle scheduling.

[0024] (2) The designed neighborhood search operator based on the critical path is more targeted than other neighborhood search techniques. By finding the critical path in the picking process and exchanging the positions of picking operations on the critical path using the first-to-first crossover principle, the quality of the solution is improved. This local search technique, by exchanging critical operations, can focus on optimizing those parts that have a greater impact on the scheduling scheme, thereby further improving efficiency and scheduling quality.

[0025] (3) The multi-objective vehicle picking scheduling algorithm proposed in this invention addresses the problem of vehicle picking time scheduling optimization in steel logistics parks with mixed storage structures. It establishes a multi-objective mixed integer programming model with the goal of minimizing customer vehicle operation time and logistics park operation time, and determines the multi-yard picking operation plan of vehicles under limited space resources.

[0026] Advantages of this invention:

[0027] Compared to the patent for "Comprehensive Planning Method for Vehicle Pickup Routes," this invention considers the mixed-loading characteristics of logistics park yards in the scenario of pickup vehicles. Since each yard in the logistics park uses mixed steel storage, containing multiple different types of steel, pickup trucks can retrieve target goods from different yards. Based on the idle time in the yards, an insert-based decoding method is designed to dynamically allocate pickup yards for vehicles. This not only reduces algorithm complexity but also improves the operational efficiency of the logistics park yards. Compared to the patent for "A Method and System for Batch Scheduling of Logistics Park Vehicles Considering Multiple Pickup Locations," this invention proposes using a multi-objective optimization algorithm based on Pareto fronts to solve the logistics park pickup vehicle scheduling scheme. This optimization objective considers not only the logistics park's operating time but also the pickup vehicle's operating time. Furthermore, based on the critical path in the pickup vehicle scheduling process, a search strategy is proposed to exchange critical operations on the critical path to improve the quality of the solution. Attached Figure Description

[0028] Figure 1 This is a flowchart of the algorithm proposed in this invention.

[0029] Figure 2 This is the vehicle scheduling scheme obtained by the algorithm proposed in this invention. Detailed Implementation

[0030] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0031] This invention provides a logistics park pickup vehicle scheduling method based on an improved SPEA2. By finding the location of the target goods, the pickup time of the pickup vehicle in each yard is calculated based on the crane operating speed and the location of the goods. An insertion decoding method is used to insert the pickup vehicle into the idle time in the yard. In addition, a neighborhood search strategy for exchanging critical paths is adopted to select pickup vehicle operations on the critical path for exchange. Figure 1 This is a flowchart of the method of the present invention. Figure 2 This is the vehicle scheduling scheme obtained by the algorithm proposed in this invention. The following are the specific implementation steps of this invention:

[0032] 1. Calculate the placement location of the target cargo for the pickup vehicle in each storage yard, based on t = (d y +2)N determines the pickup time t of the pickup vehicle in each yard, where d y The distance between the storage location of the goods and the pick-up point in the yard is represented by N. The speed of the crane is 2m / min, and the time it takes for the crane to pick up one unit of goods is 2min. N represents the quantity of goods.

[0033] 2. Initialize the number of iterations q = 0, and the maximum number of iterations q max =200, nearest neighbor threshold k=10, elite archive size η=100; population size NP=100, crossover probability CP=0.8, mutation probability MP=0.1.

[0034] 3. Based on the information of the pickup vehicles in the logistics park for the day, the pickup vehicle scheduling plan is particle-coded. A two-level coding method is adopted. The operation sequence chain OSC represents the order of vehicle operations, and its length is equal to the total number of vehicle operations. The yard allocation chain YAC is used to allocate yards for each vehicle operation. Each gene in YAC is arranged according to the vehicle index and the order of vehicle operations, and each integer represents the yard number currently allocated to that operation in the set of available yards.

[0035] 4. Randomly assign available storage yards to each pickup vehicle operation, and generate an initial population P. q .

[0036] 5. Generate an elite file A with a size of 100. q .

[0037] 6. Iterate 200 times.

[0038] 7. Convert the yard allocation chain YAC into a yard allocation matrix Y m [i,j] and the time matrix T[i,j], Ym [i,j] represents the yard allocated to the j-th pickup operation of vehicle i, and T[i,j] represents the time required for the j-th pickup of vehicle i.

[0039] 8. Decode the OSC; if O i,j This is the first operation of vehicle i and is in yard M. k The first operation is O. i,j It can be processed directly at time 0 of the stockpile; if O i,j It is M k The first operation, but not the first operation of vehicle i, can be performed in the previous operation O. i,j-1 Insert O directly after the completion time. i,j Otherwise, locate storage yard M. k All free time periods [TS] x ,TE x ], where TS x It is the start time of the x-th idle time period, TE x This is the end time of the x-th idle period. In yard M... k During the work process, determine O i,j Does it satisfy the insertion condition max{C i,j-1 ,TS x}+t i,j,k ≤TE x C i,j-1 t is the completion time of the (j-1)th pickup operation of the vehicle. i,j,k It is the pickup time of the j-th operation of pickup vehicle i in yard k. If satisfied, it can be inserted into the idle time [TS]. x ,TE x In the context of [ ], the start time is min{C i,j-1 ,LM k}wherein LM k It is the current stockpile M k The earliest available time. The pickup time of the j-th operation at yard k. If satisfied, it can be inserted into the idle time [TS]. x ,TE x In the context of [ ], the start time is min{C i,j-1 ,LM k}wherein LM k It is the current stockpile M k The earliest available time.

[0040] 9. Based on the decoding results, calculate the logistics park's operating time C. max Maximum running time of pickup vehicle SC max .

[0041] 10. Based on the objective function value C max With SC max Calculate the fitness value F for each individual in the population.

[0042] 11. Based on the fitness value F, from P q With A q The top 100 individuals selected through environmental choices will be placed into Elite File A. q+1 .

[0043] 12. Regarding A in the elite files q+1 Each individual underwent crossover and mutation with a crossover probability of 0.8 and a mutation probability of 0.1. OSC used POX crossover and two-point mutation, while YAC used two-point crossover and two-point mutation. The results of the crossover and mutation were saved into P... q+1 .

[0044] 13. Find the longest path among all paths from the start node to the end node, perform a neighborhood search by moving the pickup vehicle on the critical path, and obtain a new chromosome.

[0045] 14. Check if the number of iterations has reached the maximum number of iterations, 200. If so, stop iterating and output the global optimal position P. g This is used as the optimal scheduling result.

[0046] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for scheduling pickup vehicles in a logistics park based on an improved SPEA2, characterized in that, By locating the target cargo and calculating the pickup time of the pickup vehicle in each yard based on the crane's operating speed and the cargo's location, an insertion-based decoding method is used to insert the pickup vehicle into the idle time of the yard. Furthermore, a neighborhood search strategy for exchanging critical paths is employed to select pickup vehicle operations on the critical path for exchange. The specific implementation of this method is as follows: S1. Determine the placement location of the target goods for the pickup vehicle in each storage yard, based on... Determine the pickup time t for each pickup vehicle at each yard, where The distance between the storage location of the goods and the pick-up point in the yard is represented by v, the speed of the crane is represented by a, the time it takes for the crane to pick up one unit of goods is represented by a, and N is represented by the quantity of goods. S2. Initialize the number of iterations q, and the maximum number of iterations. Nearest neighbor threshold k, elite archive size Population size NP, crossover probability CP, mutation probability MP; S3. Based on the information of the pickup vehicles in the logistics park on the same day, the pickup vehicle scheduling plan is particle-coded. A two-level coding method is adopted. The operation sequence chain OSC represents the order of vehicle operations, and its length is equal to the total number of vehicle operations. The yard allocation chain YAC is used to allocate yards for each vehicle operation. Each gene of YAC is arranged according to the vehicle index and the order of vehicle operations. Each integer represents the yard number currently allocated to the corresponding operation in the set of available yards. S4. Randomly allocate available storage yards to each pickup vehicle operation, and generate an initial population according to the designed coding method. Generate a capacity of Elite Archives ; S5, Iterative Loop Second-rate; S6. Convert the YAC yard allocation chain into a yard allocation matrix. With time matrix , This represents the yard to which the j-th pickup operation of vehicle i is assigned. This represents the time required for vehicle i to pick up its j-th item; S7. Decode the OSC, if This is the first operation of vehicle i and it is in the yard. The first operation, then the corresponding operation Process directly by inserting the corresponding time step 0 of the stockpile; if yes If the first operation is not the first operation of vehicle i, then the previous operation... Insert directly after the completion time Otherwise, locate the storage yard. All free time periods ,in It is the start time of the x-th idle period. It is the end time of the x-th idle period; in the storage yard During the work process, judgment Does the insertion condition meet? ,in It is the completion time of the (j-1)th pickup operation of the vehicle. It is the pickup time of the j-th operation of pickup vehicle i in yard k; If satisfied, insert it into the idle time. In the middle, the start time is ,in This is the current storage yard. The earliest available time; S8. Calculate the logistics park's operating time based on the decoding results of S7. Maximum operating time of pickup vehicle ; S9. Based on the objective function value and Calculate the fitness value F for each individual in the population; S10. Based on the fitness value F, from and Before making environmental selection Outstanding individuals were included in the elite archives. ; S11, Regarding the elite files Each individual underwent crossover and mutation according to the crossover probability CP and the mutation probability MP. OSC used POX crossover and two-point mutation, while YAC used two-point crossover and two-point mutation. The results of the crossover and mutation were saved... ; S12. Find the longest path among all paths from the start node to the end node, perform a neighborhood search by moving the pickup vehicle on the critical path, and obtain a new chromosome. S13. Determine if the number of iterations has reached the maximum number of iterations. If the condition is met, stop iterating and output the globally optimal position. This is used as the optimal scheduling result.

Images