A cross-layer shuttle vehicle access system scheduling decision method

Abstract

The present application relates to a kind of cross-layer shuttle vehicle access system scheduling decision method, it includes the following steps: S1: with the minimum total operation time of warehouse-out and the maximum comprehensive stability of shelf as goal establishes cross-layer shuttle vehicle access system integrated optimization mathematical model;S2: design an improved multi-objective wild horse optimization algorithm;S3: analyze the equipment conflict problem in the process of system execution task, propose a kind of conflict detection and correction strategy and integrate into the algorithm in step S2;S4: the method in step S2 and S3 is used to solve the multi-objective integrated optimization model in step S1. Compared with prior art, this method simultaneously solves the task sequencing, equipment scheduling and location allocation problem of cross-layer shuttle vehicle access system, to obtain the Pareto solution set scheme with better comprehensive stability of shelf, shorter batch order warehouse-out operation time and no conflict of equipment operation for manager decision-making.

Description

Technical Field

[0001] This invention relates to the field of warehouse system scheduling and optimization, and in particular to a scheduling decision method for a cross-level shuttle access system. Background Technology

[0002] Cross-level shuttle access systems are a new type of automated storage system that has evolved from traditional automated storage and retrieval systems in recent years. This system consists of shuttles and elevators within multi-level racking, offering advantages such as high space utilization and storage capacity, higher throughput, and greater flexibility to adapt to external order fluctuations. Compared to layer-restricted shuttle access systems, cross-level shuttle access systems allow shuttles to access goods across levels via elevators, providing greater flexibility and robustness while reducing the number of shuttles required. However, because shuttles may need to cross levels during task execution, conflicts can arise when multiple shuttles are performing tasks. Different task sequencing, equipment scheduling, and storage location allocation schemes lead to different operating routes for shuttles and elevators, thus affecting the efficiency of the system in executing batch order tasks. The allocation scheme of aisles and storage locations directly affects the structural stability of the entire racking system. Therefore, task sequencing, coordinated scheduling of shuttles and elevators, and storage location allocation are of great importance.

[0003] Currently, there has been relevant research on traditional automated storage and retrieval systems both domestically and internationally. However, most existing technologies only consider task scheduling or location allocation in isolation, with few simultaneously optimizing task sequencing, equipment scheduling, and location allocation. Research on scheduling decisions for cross-level shuttle access systems is even scarcer, and few consider the conflict issues that arise when shuttles perform tasks. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a scheduling decision method for a cross-level shuttle access system.

[0005] The objective of this invention can be achieved through the following technical solutions: A scheduling decision-making method for a cross-level shuttle storage system solves the integrated optimization problem of task sequencing, equipment scheduling, and storage location allocation in the system, resulting in a Pareto solution with good overall rack stability, short batch task outbound operation time, and no equipment conflicts. The method includes the following steps: S1: Establish a multi-objective integrated optimization mathematical model for a cross-level shuttle access system with the goals of minimizing total outbound operation time and maximizing overall shelf stability; S2: Design an improved multi-objective Mustang optimization algorithm; S3: Analyze the device conflict problem during the system's task execution process, propose a conflict detection and correction strategy, and integrate it into the algorithm of step S2; S4: Solve the multi-objective integrated optimization model in step S1 using the methods in steps S2 and S3 to obtain the integrated optimization scheme for task sequencing, equipment scheduling, and cargo location allocation.

[0006] Preferably, the objective function of the multi-objective ensemble optimization mathematical model includes: total outbound operation time and overall shelf stability. The total outbound operation time function is: in For alleyway The completion time of the task queue. Number the lanes For alleyway The length of the task queue.

[0007] The aforementioned maximum overall stability of the shelving is expressed as: in, Indicates a lane The overall stability of the shelving unit The larger the value, the worse the overall stability of the shelving unit.

[0008] Preferably, in the outbound operation time function The calculation is as follows: in, Indicates cargo hoist Execution tunnel The last task The starting time, Indicates cargo hoist Execution tunnel The last task The time taken. Indicates a lane Equipped with a cargo lift.

[0009] Preferably, the aforementioned Calculation in progress Indicates cargo hoist Execution tunnel The task queue r Task The starting time, The calculation is as follows: in, Indicates shuttle Execute the task The moment of arrival at the outbound storage location level Indicates a lane Equipped with j Taiwan shuttle bus, Indicates task Located in the shuttle The task queue One task, express Execute the task The end moment.

[0010] For non-cross-layer tasks ,in, express Execute the task The starting moment. The calculation formula is: in, express Execute the task The end moment.

[0011] For cross-level outbound tasks , With shuttle elevator Execute the task The start time Consistent, Indicates a lane Equipped with a shuttle lift, Indicates task Located in the shuttle elevator The task queue One task, Indicates shuttle lift Execute the task The time it takes up. The calculation formula is: in, Indicates shuttle lift Execute the task The end moment.

[0012] The calculation formula is: in, and These represent the travel time of the shuttle elevator from its current floor to the floor where the shuttle is located, and the travel time from the floor where the shuttle is located to the floor where the target cargo location is located, respectively. This indicates the time it takes for the shuttle car lift to load the shuttle car. This indicates the time it takes for the shuttle car to be unloaded by the shuttle car elevator.

[0013] Preferably, the aforementioned In the calculation Indicates cargo hoist Execution tunnel The r Task The time taken The calculation is as follows: in, and These represent the travel time of the cargo hoist from the I / O point of the aisle to the floor where the target cargo location is located, and the travel time from the floor where the target cargo location is located back to the I / O point of the aisle, respectively. This indicates the time it takes for the cargo elevator to load cargo. This indicates the time it takes for the cargo elevator to unload cargo.

[0014] For shuttle Execute the task The time occupied for non-cross-layer outbound tasks is calculated as follows: This indicates the one-way travel time of the shuttle vehicle in the horizontal direction between the I / O point on its current floor and the target storage location. Indicates the time it takes for the shuttle to load cargo containers. This indicates the time it takes for the shuttle to unload the cargo container.

[0015] For cross-level outbound tasks The calculation is as follows: The travel time of the vertical elevator from shelf z1 to shelf z2. The calculation is as follows: , in, Indicates the height of the shelf unit. Z The total number of shelves. To increase the acceleration of the booster, This refers to the maximum travel speed of the hoist.

[0016] The travel time of the horizontal shuttle from column y1 to column y2 of the shelf. The calculation is as follows: , in, Indicates the width of the shelf unit. C This represents the total number of columns on the shelf. For the acceleration of the shuttle, This represents the maximum speed of the shuttle.

[0017] Preferably, the tunnel The overall stability function of the shelving group is: in, Indicates a lane The structural stability of the shelving unit Indicates a lane The overturning resistance of the shelving unit and The weights for the structural stability and anti-tipping properties of the shelving group are respectively, and they meet the following requirements. .

[0018] The structural stability function of the aforementioned shelving group is: in, Z The total number of shelves. C This represents the total number of columns on the shelf. x Indicates the shelf number. y Indicates the shelf row number. z Indicates the shelf level number. Indicates the occupancy status of the storage space. If the storage space is occupied, take 1; otherwise, take 0. Indicates stored in The quality of the goods at the storage location. This indicates the height of the shelving unit.

[0019] The overturning resistance function of the aforementioned shelving group is: Preferably, the constraints of the established mathematical model are: in, For the task In the alley The total number of storage locations in the alleyway The first in n One storage location was selected as the task. i When the goods are shipped out of the warehouse, Take 1; otherwise, Take 0. I This indicates the total number of tasks in the batch. J This indicates the total number of shuttle cars equipped in each alleyway. (When an order task...) alleyway The first in j Taiwan shuttle During execution, Take 1; otherwise, Set to 0. When the task... Cargo elevator During execution, Take 1; otherwise, Take 0. Indicates shuttle The length of the task queue. For alleyway The length of the task queue.

[0020] Preferably, the improvements in step S2 specifically include: S21: Design encoding and decoding methods for the task sorting segment, elevator selection segment, shuttle selection segment, and cargo location allocation segment respectively.

[0021] S22: Introduces a non-dominated ranking method, redefines the selection methods for stallions, foals, and habitats, and modifies the herd's election leadership behavior.

[0022] S23: Design a double outcry segmentation multi-point crossover mechanism to improve the mating and reproduction behavior of horse herds and increase the global search capability of the algorithm.

[0023] S24: Introduce the Levy flight strategy to improve the herd guidance behavior and increase the algorithm's ability to escape local optima.

[0024] Preferably, the encoding and decoding method for the task sorting segment, elevator selection segment, shuttle selection segment, and storage location allocation segment in step S21 includes: to solve the task scheduling and storage location allocation problem, each solution includes four segments: namely, the task sorting segment, elevator selection segment, shuttle selection segment, and storage location allocation segment. The length of each segment's encoding is equal to the total number of tasks in the batch order. The elevator and shuttle selection segments constitute the equipment selection segment. The task sorting segment uses random key encoding, while the equipment selection segment and storage location allocation segment use real number encoding. The encodings of the same sequential position in the four segments are grouped together and change along with the task sorting. The random key value range of the task sorting segment is... Tasks with smaller random keys are ranked higher. The encoding value range for the elevator selection segment is... The range of coded values ​​for the shuttle selection segment is: .in, For the number of lanes, The number of shuttle cars assigned to each lane. The code for the equipment selection segment is rounded down to obtain the decoded result. For example, if a task's code 3.42 in the hoist selection segment is rounded down to 3, then the task will be retrieved by hoist #3 and added to the queue of lane #3. If the code 2.77 in the shuttle car selection segment is rounded down to 2, then the task will be handled by shuttle car #2 in lane #3. The range of the code values ​​for the location allocation segment is... ,in, This represents the number of storage locations for this task in the selected aisle. The encoded value is rounded down and decoded. For example, 4.09 rounded down to 4 indicates that the outbound storage location assigned to this task is storage location #4 in the storage location set of the selected aisle.

[0025] Preferably, in step S22, the method for selecting stallions, foals, and habitats is as follows: for herd initialization, all individuals in the herd are divided into different sets ( Stallions are preferred from The remaining stallions are randomly selected from the individuals in the remaining set. The remaining members are randomly assigned to stallions as foals, forming different herds. Afterwards, stallions within the herds are selected based on fitness. A non-dominated sort is performed on all stallions to obtain (…). ), randomly from Choose one stallion as the habitat. The formula for the herd's election leadership behavior is: in, This indicates the current location of the foal in the herd. For the horse herd The position of the stallion.

[0026] Preferably, the double-outbreak segmented multi-point crossover mechanism in step S23 includes: mature foals are separated from their original herds and mated with mature foals from different herds to produce two offspring, which are then randomly added to two different herds. The specific steps of the segmented multi-point crossover are as follows: S231: Randomly select member 1 from the task sorting segment. The code is swapped with the code at the same position as member 2. .in, This represents the number of orders.

[0027] S232: The elevator and shuttle car selection segment randomly selects member 1. The adjacent double code is swapped with the adjacent double code of member 2. .

[0028] S233: Randomly select member 1 for the cargo location allocation section. Each neighboring three-bit code, and member 2 is randomized. Three neighboring bits are swapped. .

[0029] Preferably, the Levi flight strategy in step S24 includes: performing a non-dominated sort on all stallions to obtain a Pareto set ( ),from One stallion is randomly selected as... Instead of moving towards its habitat, it opted for Levi flight. The details of Levi flight are as follows: in Indicates the step size scaling factor. d Dimensions representing an individual's location and It follows a standard normal distribution and is usually set as follows: , . Indicates stallion The new location after the move.

[0030] Preferably, in step S3, the conflict detection and correction strategy includes: recording the starting layer of the shuttle hoist and the shuttle car and the dwell layer after each task execution. Conflict detection is performed on the decoded individuals. For all shuttle cars equipped in the same roadway, for tasks of the same execution order, conflicts may occur in the following two situations: 1) The target cargo location layer of shuttle car A is occupied by shuttle car B (… ); 2) At least two shuttles have the same target cargo location layer. Conflict detection is performed on a per-lane basis. The layer of the target cargo location is obtained from the assigned cargo location number. A conflict is determined based on the target cargo location layer and the shuttle's current layer. For detected conflicts, the individual code is corrected by changing the target cargo location number, thus avoiding the conflict. For individuals that cannot be corrected using this method, a fitness penalty is applied.

[0031] Preferably, the specific steps of the conflict detection and correction strategy are as follows: Three shuttle cars, C11, C12, and C13, are installed in tunnel #1. S31: Calculate the number of detection cycles max_cycle = Number of tasks in the longest shuttle task queue; initialize conflict flags. with = 0, the floor where shuttles C11, C12, and C13 stop. ST 1. ST 2. ST 3; Order cycle 1 S32: Get the current C11, C12, C13 cycle Target storage level OT 1, OT 2, OT 3 S33: If OT 1 == ST 2 or OT 1 == ST 3. Then, retrieve the remaining storage locations in the same aisle where C11 is currently performing a task, randomly select one of these storage locations to replace the original one, and update the settings. OT 1; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let with = 1, skip to step S37 S34: If OT 2 == ST 1 or OT 2 == ST 3 or OT 2 == OT 1. Then, retrieve the remaining storage locations in the same aisle where C12 is currently executing a task, randomly select one of these storage locations to replace the original storage location, and update the settings. OT 2; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let with = 1, skip to step S37 S35: OT 3 == ST 1 or OT 3 == ST 2 or OT 3 == OT 1 or OT 3 == OT 2. Then, retrieve the remaining storage locations in the same aisle where C13 is currently performing a task, randomly select one of these storage locations to replace the original storage location, and update the settings. OT 3; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let with = 1, skip to step S37 S36: ST 1 OT 1, ST 2 OT 2, ST 3 OT 3; Order cycle cycle + 1. If cycle max_ cycle Return to step S32 S37: End, Output with and the revised cargo allocation section Preferably, in step S4, the complete steps of the solution method are as follows: S41: Set population size N stallion ratio P.S. Crossover probability PC Number of iterations S42: Randomly initialize the population and calculate the fitness function for each individual. , And perform conflict detection and correction to obtain with ,if with = 1, then ,in yes Penalty coefficient S43: Perform a non-dominated ranking of all individuals based on their fitness, and obtain the Pareto set of all individuals. Select stallions and assign foals to form a herd. S44: Perform a non-dominated sort on all stallions to obtain the Pareto set of stallions. Randomly from Choose a location for a stallion as its habitat. WH ,from One stallion is randomly selected as... S45: Calculation ,in The coefficients decrease linearly from 1 to 0. The maximum number of iterations for the algorithm S46: For each herd Calculate the adaptive mechanism vector in, , , It is a vector consisting of 0s and 1s, with the same dimension as the variable. and Let [a] be a random vector that follows a uniform distribution in the interval [0,1]. A random number that follows a uniform distribution within the range [0,1]. To meet Conditional random vector Returned index S47: For each foal ,if The formula for updating the foal grazing behavior is: Otherwise, the foal will stray from the herd and be moved to the stray archive. Archive in, This indicates the current location of the foal in the herd. For the horse herd The position of the stallion. R It is a uniformly random number within the range Internally generated, Let the value of pi be 3.14. A random number that follows a uniform distribution within the range [0,1]. S48: Yes Archive The foals in the herd mate, and the position of the offspring foals is obtained through a double-outbreak segmented multi-point crossover mechanism. The updated offspring foals are then assigned to herds with vacant positions that are different from their original positions. After that, the herd is cleared. Archive S49: For each stallion If It will perform a Levi flight; otherwise, its position will be updated according to the following formula: , in, The value is a random number uniformly distributed within the range [0,1]. If the updated stallion's position dominates the original position, then... S410: Update the fitness of all individuals, perform conflict detection and remediation, and obtain the final fitness. S411: For each foal in each herd, conduct herd leadership elections. S412: ,if Otherwise, return to step S44; otherwise, skip to step S413. S413: Perform a non-dominated ranking on all individuals to obtain ( Output Compared with the prior art, the present invention has the following advantages: (1) For the task sequencing, equipment scheduling and storage location allocation problems of cross-level shuttle storage system, an integrated optimization mathematical model is established with the goal of minimizing the total operation time and maximizing the overall stability of the rack group, and an improved Wild Horse optimization algorithm is designed to solve the problem.

[0032] (2) Design encoding and decoding methods for the task sorting segment, elevator selection segment, shuttle selection segment, and cargo location allocation segment respectively. Introduce non-dominated sorting method and Levy flight strategy, and design a double outlier segment multi-point crossover mechanism to enhance the algorithm's global search and ability to escape local optima, thereby balancing the algorithm's exploration and utilization and improving the algorithm's performance.

[0033] (3) Analyze the conflict problem caused by the cross-layer execution of tasks by the equipment, design a conflict detection and correction strategy, and integrate it into the algorithm to avoid the conflict problem in the operation of the equipment in the obtained solution set.

[0034] (4) The proposed method is used to solve the mathematical model of the cross-level shuttle access system established above, so as to obtain the integrated optimization scheme of task sequencing, equipment scheduling and cargo location allocation. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the encoding and decoding process of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0036] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0037] Example The cross-level shuttle system is configured as follows: each aisle consists of two rows of high-density racking. Two combined elevators are located at the beginning of each aisle. For single-instruction-cycle outbound operations, one elevator functions as a shuttle elevator, and the other as a cargo elevator. Shuttles and cargo boxes can be transported vertically via the combined elevators. Each aisle is equipped with several shuttles, which can move horizontally along tracks on each rack level. Once a task is assigned to an aisle, the shuttle elevator and cargo elevator are determined. A buffer zone is provided between each rack level and the elevators to buffer cargo boxes. Each type of goods has multiple storage locations on the racks in different aisles. Orders received by the system are divided into different batches, each batch containing a defined number of orders. The distribution of outbound goods across the entire racking system for each batch of orders is known.

[0038] This application proposes a scheduling decision method for a cross-level shuttle access system, including the following steps: S1: A multi-objective integrated optimization mathematical model for a cross-level shuttle access system is established with the objectives of minimizing the total outbound operation time and maximizing the overall stability of the shelving, as detailed below: The objective functions of the multi-objective ensemble optimization mathematical model include: total outbound operation time and overall shelf stability. The total outbound operation time function is: in For alleyway The completion time of the task queue. Number the lanes For alleyway The length of the task queue. The calculation is as follows: in, Indicates cargo hoist Execution tunnel The last task The starting time, Indicates cargo hoist Execution tunnel The last task The time taken. Indicates a lane Equipped with a cargo lift. Indicates cargo hoist Execution tunnel The task queue r Task The starting time, The calculation is as follows: in, Indicates shuttle Execute the task The moment of arrival at the outbound storage location level Indicates a lane Equipped with j Taiwan shuttle bus, Indicates task Located in the shuttle The task queue One task, express Execute the task The end moment.

[0039] For non-cross-layer tasks ,in, express Execute the task The starting moment. The calculation formula is: in, express Execute the task The end moment.

[0040] For cross-level outbound tasks , With shuttle elevator Execute the task The start time Consistent, Indicates a lane Equipped with a shuttle lift, Indicates task Located in the shuttle elevator The task queue One task, Indicates shuttle lift Execute the task The time it takes up. The calculation formula is: in, Indicates shuttle lift Execute the task The end moment.

[0041] The calculation formula is: in, and These represent the travel time of the shuttle elevator from its current floor to the floor where the shuttle is located, and the travel time from the floor where the shuttle is located to the floor where the target cargo location is located, respectively. This indicates the time it takes for the shuttle car lift to load the shuttle car. This indicates the time it takes for the shuttle car to be unloaded by the shuttle car elevator.

[0042] In the calculation formula Indicates cargo hoist Execution tunnel The r Task The time taken The calculation is as follows: in, and These represent the travel time of the cargo hoist from the I / O point of the aisle to the floor where the target cargo location is located, and the travel time from the floor where the target cargo location is located back to the I / O point of the aisle, respectively. This indicates the time it takes for the cargo elevator to load cargo. This indicates the time it takes for the cargo elevator to unload cargo.

[0043] For shuttle Execute the task The time occupied for non-cross-layer outbound tasks is calculated as follows: This indicates the one-way travel time of the shuttle vehicle in the horizontal direction between the I / O point on its current floor and the target storage location. Indicates the time it takes for the shuttle to load cargo containers. This indicates the time it takes for the shuttle to unload the cargo container.

[0044] For cross-level outbound tasks The calculation is as follows: Preferably, the tunnel The overall stability function of the shelving group is: in, Indicates a lane The structural stability of the shelving unit Indicates a lane The overturning resistance of the shelving unit and The weights for the structural stability and anti-tipping properties of the shelving group are respectively, and they meet the following requirements. .

[0045] The travel time of the vertical elevator from shelf z1 to shelf z2. The calculation is as follows: , in, Z represents the height of the shelving unit, and Z represents the total number of shelves. To increase the acceleration of the booster, This refers to the maximum travel speed of the hoist.

[0046] The travel time of the horizontal shuttle from column y1 to column y2 of the shelf. The calculation is as follows: , in, Indicates the width of the shelf unit. C This represents the total number of columns on the shelf. For the acceleration of the shuttle, This represents the maximum speed of the shuttle.

[0047] Maximizing the overall stability of the shelving system is expressed as: in, Indicates a lane The overall stability of the shelving unit The larger the value, the worse the overall stability of the shelving unit.

[0048] The overall stability function of the shelving group is: in, Indicates a lane The structural stability of the shelving unit Indicates a lane The overturning resistance of the shelving unit and The weights for the structural stability and anti-tipping properties of the shelving group are respectively, and they meet the following requirements. .

[0049] The structural stability function of the shelving group is: in, Z The total number of shelves. C This represents the total number of columns on the shelf. x Indicates the shelf number. y Indicates the shelf row number. z Indicates the shelf level number. Indicates the occupancy status of the storage space. If the storage space is occupied, take 1; otherwise, take 0. Indicates stored in The quality of the goods at the storage location. This indicates the height of the shelving unit.

[0050] The overturning resistance function of the shelving group is: The constraints of the established mathematical model are: in, For the task In the alley The total number of storage locations in the alleyway The first in n One storage location was selected as the task.i When the goods are shipped out of the warehouse, Take 1; otherwise, Take 0. I This indicates the total number of tasks in the batch. J This indicates the total number of shuttle cars equipped in each alleyway. (When an order task...) alleyway The first in j Taiwan shuttle During execution, Take 1; otherwise, Set to 0. When the task... Cargo elevator During execution, Take 1; otherwise, Take 0. Indicates shuttle The length of the task queue. For alleyway The length of the task queue.

[0051] S2: Design an improved multi-objective Mustang optimization algorithm, specifically including: S21: Design encoding and decoding methods for the task sequencing segment, elevator selection segment, shuttle selection segment, and cargo location allocation segment, as detailed below: To solve the task scheduling and storage location allocation problem, each solution includes four segments: task sequencing, elevator selection, shuttle selection, and storage location allocation. The length of each segment's code equals the total number of tasks in the batch order. The elevator and shuttle selection segments constitute the equipment selection segment. The task sequencing segment uses random key encoding, while the equipment selection and storage location allocation segments use real number encoding. Codes for the same sequential position in the four segments are grouped together and change along with the task sequencing. The random key value range for the task sequencing segment is... Tasks with smaller random keys are ranked higher. The encoding value range for the elevator selection segment is... The range of coded values ​​for the shuttle selection segment is: .in, For the number of lanes, The number of shuttle cars assigned to each lane. The code for the equipment selection segment is rounded down to obtain the decoded result. For example, if a task's code 3.42 in the hoist selection segment is rounded down to 3, then the task will be retrieved by hoist #3 and added to the queue of lane #3. If the code 2.77 in the shuttle car selection segment is rounded down to 2, then the task will be handled by shuttle car #2 in lane #3. The range of the code values ​​for the location allocation segment is... ,in, This represents the number of storage locations for this task in the selected aisle. The encoded value is rounded down and decoded. For example, 4.09 rounded down to 4 indicates that the outbound storage location assigned to this task is storage location #4 in the storage location set of the selected aisle.

[0052] S22: Introducing a non-dominated ranking method, redefining the selection methods for stallions, foals, and habitats, and modifying the herd leadership election behavior, as follows: For the initialization of the horse herd, all individuals in the population are sorted in a non-dominated order to divide them into different sets. Stallions are preferred from The remaining stallions are randomly selected from the individuals in the remaining set. The remaining members are randomly assigned to stallions as foals, forming different herds. Afterwards, stallions within the herds are selected based on fitness. A non-dominated sort is performed on all stallions to obtain (…). ), randomly from Choose one stallion as the habitat. The formula for the herd's election leadership behavior is: in, This indicates the current location of the foal in the herd. For the horse herd The position of the stallion.

[0053] S23: Design a double outlier segmentation and multi-point crossover mechanism to improve the mating and reproduction behavior of horse herds and increase the algorithm's global search capability. Details are as follows: Mature foals are separated from their herds and mated with other mature foals from different herds, producing two offspring. These offspring are then randomly assigned to two different herds. The specific steps of this segmented multi-point crossover are as follows: S231: Randomly select member 1 from the task sorting segment. The code is swapped with the code at the same position as member 2. .in, This represents the number of orders.

[0054] S232: The elevator and shuttle car selection segment randomly selects member 1. The adjacent double code is swapped with the adjacent double code of member 2. .

[0055] S233: Randomly select member 1 for the cargo location allocation section. Each neighboring three-bit code, and member 2 is randomized. Three neighboring bits are swapped. .

[0056] S24: Introduce the Levy flight strategy to improve the herd guidance behavior and increase the algorithm's ability to escape local optima.

[0057] Levi's flight strategy includes: performing a non-dominated sort on all stallions to obtain the Pareto set ( ),from One stallion is randomly selected as... Instead of moving towards its habitat, it opted for Levi flight. The details of Levi flight are as follows: in Indicates the step size scaling factor. d Dimensions representing an individual's location and It follows a standard normal distribution and is usually set as follows: , . Indicates stallion The new location after the move.

[0058] S3: Analyze the device conflict problem during the system's task execution process, propose a conflict detection and correction strategy, and integrate it into the algorithm of step S2; Record the starting level of the shuttle hoist and the shuttle car itself, as well as the level where they remain after each task execution. Perform conflict detection on the decoded individuals. For all shuttle cars in the same roadway, conflicts may occur in the following two situations for tasks of the same execution order: 1) The target cargo level of shuttle car A is occupied by shuttle car B (…). ); 2) At least two shuttles have the same target cargo location layer. Conflict detection is performed on a per-lane basis. The layer of the target cargo location is obtained from the assigned cargo location number. A conflict is determined based on the target cargo location layer and the shuttle's current layer. For detected conflicts, the individual code is corrected by changing the target cargo location number, thus avoiding the conflict. For individuals that cannot be corrected using this method, a fitness penalty is applied.

[0059] The specific steps of the conflict detection and correction strategy are as follows: Three shuttle cars, C11, C12, and C13, are installed in tunnel #1. S31: Calculate the number of detection cycles max_cycle = Number of tasks in the longest shuttle task queue; initialize conflict flags. with = 0, the floor where shuttles C11, C12, and C13 stop. ST 1. ST 2. ST 3; Order cycle 1 S32: Get the current C11, C12, C13 cycle Target storage level OT 1, OT 2, OT 3 S33: If OT 1 == ST 2 or OT 1 == ST 3. Then, retrieve the remaining storage locations in the same aisle where C11 is currently performing a task, randomly select one of these storage locations to replace the original one, and update the settings. OT 1; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let with = 1, skip to step S37 S34: If OT 2 == ST 1 or OT 2 == ST 3 or OT 2 == OT 1. Then, retrieve the remaining storage locations in the same aisle where C12 is currently executing a task, randomly select one of these storage locations to replace the original storage location, and update the settings. OT 2; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let with = 1, skip to step S37 S35: OT 3 == ST 1 or OT 3 == ST 2 or OT 3 == OT 1 or OT 3 == OT 2. Then, retrieve the remaining storage locations in the same aisle where C13 is currently performing a task, randomly select one of these storage locations to replace the original storage location, and update the settings. OT 3; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let with = 1, skip to step S37 S36: ST 1 OT 1, ST 2 OT 2, ST 3 OT 3; Order cycle cycle + 1. If cycle max_ cycle Return to step S32 S37: End, Output with and the revised cargo allocation section S4: Solve the multi-objective integrated optimization model from step S1 using the methods in steps S2 and S3 to obtain the integrated optimization scheme for task sequencing, equipment scheduling, and cargo location allocation. The specific steps are as follows: S41: Set population size N stallion ratio P.S. Crossover probability PC Number of iterations S42: Randomly initialize the population and calculate the fitness function for each individual. , And perform conflict detection and correction to obtain with ,if with = 1, then ,in yes Penalty coefficient S43: Perform a non-dominated ranking of all individuals based on their fitness, and obtain the Pareto set of all individuals. Select stallions and assign foals to form a herd. S44: Perform a non-dominated sort on all stallions to obtain the Pareto set of stallions. Randomly from Choose a location for a stallion as its habitat. WH ,from One stallion is randomly selected as... S45: Calculation ,in The coefficients decrease linearly from 1 to 0. The maximum number of iterations for the algorithm S46: For each herd Calculate the adaptive mechanism vector in, , , It is a vector consisting of 0s and 1s, with the same dimension as the variable. and Let [a] be a random vector that follows a uniform distribution in the interval [0,1]. A random number that follows a uniform distribution within the range [0,1]. To meet Conditional random vector Returned index S47: For each foal ,if The formula for updating the foal grazing behavior is: Otherwise, the foal will stray from the herd and be moved to the stray archive. Archive in, This indicates the current location of the foal in the herd. For the horse herd The position of the stallion. R It is a uniformly random number within the range Internally generated, Let the value of pi be 3.14. A random number that follows a uniform distribution within the range [0,1]. S48: Yes Archive The foals in the herd mate, and the position of the offspring foals is obtained through a double-outbreak segmented multi-point crossover mechanism. The updated offspring foals are then assigned to herds with vacant positions that are different from their original positions. After that, the herd is cleared. Archive S49: For each stallion If It will perform a Levi flight; otherwise, its position will be updated according to the following formula: , in, The value is a random number uniformly distributed within the range [0,1]. If the updated stallion's position dominates the original position, then... S410: Update the fitness of all individuals, perform conflict detection and remediation, and obtain the final fitness. S411: For each foal in each herd, conduct herd leadership elections. S412: ,if Otherwise, return to step S44; otherwise, skip to step S413. S413: Perform a non-dominated ranking on all individuals to obtain ( Output In this embodiment, to test the effectiveness of the model and algorithm, test cases are required. A total of 20 test cases were generated, including 7 combinations of task sizes and 4 combinations of the number of aisles. All cases have 3 multi-level shuttles per aisle. Each shelf has 20 layers and 40 columns. The type, location, and quality information of the goods stored in each aisle are known, as shown in Table 1.

[0060] Table 1 Since the choice of parameters directly affects the performance of the algorithm, it is necessary to set the optimal algorithm parameters to ensure the quality of the obtained solution. The improved multi-objective Mustang optimization algorithm includes two important parameters: (1) stallion ratio PS; (2) crossover probability PC. The Taguchi method was used to conduct parameter selection experiments, and the optimal performance parameter set was obtained: PS=0.25, PC=0.25. To verify the effectiveness of the algorithm, the algorithm of this application (IMOWHO) was compared with the classic multi-objective algorithm - multi-objective particle swarm optimization (MOPSO). The 20 cases in Table 1 were used as comparative experiments. The population size of all experiments was uniformly set to 300, the algorithm running time was set to task seconds, and each experimental case was run 20 times. The average C metric is shown in Table 2.

[0061] Table 2 As shown in Table 2, in the 20 sets of experiments, the Pareto solution obtained by the method proposed in this application dominates most of the solutions for MOPSO. Therefore, this method can more effectively solve the task sequencing, equipment scheduling, and storage location allocation problems of cross-level shuttle access systems and obtain better solutions.

[0062] To verify the effectiveness of the model and algorithm, a large-scale warehousing operation center is used as a real-world case. This warehousing center contains 4 rows of aisles, 8 automated storage and retrieval systems (AS / RS), each aisle is equipped with 3 shuttle cars capable of crossing multiple levels, and each rack has 20 layers and 40 columns. The batch order quantity is 50, and the warehouse location information is known. The parameters of the warehouse and equipment are shown in Table 3. The objective value of the solution obtained using the method proposed in this application is shown in Table 4.

[0063] Table 3 Table 4 As shown in Table 4, the operation time in the obtained Pareto solution set falls within the range of 331.795-479.221, and the overall stability of the racking system falls within the range of 8.514-33.378, indicating that both objectives have significant optimization space and potential. Therefore, the model and algorithm proposed in this application can effectively solve the integrated optimization problem of task sequencing, equipment scheduling, and storage location allocation in a cross-level shuttle access system.

[0064] In summary, this application proposes a scheduling decision-making method for a cross-level shuttle access system. An integrated optimization mathematical model for the system is established with the objectives of minimizing total outbound operation time and maximizing overall rack stability. An improved multi-objective Mustang optimization algorithm is designed, incorporating a non-dominated sorting method and a Lévy flight strategy, and employing a double outlier segmentation multi-point crossover mechanism to enhance algorithm performance. A conflict detection and correction strategy is proposed to address equipment conflicts during system task execution. The designed algorithm is used to solve problem cases of different scales and compared with the MOPSO algorithm for solving the same problem. Finally, a practical case is used to verify the effectiveness of the proposed integrated optimization model and the improved multi-objective Mustang optimization algorithm.

Claims

1. A scheduling decision method for a cross-level shuttle access system, characterized in that, Includes the following steps: S1: Establish a multi-objective integrated optimization mathematical model for a cross-level shuttle access system with the goals of minimizing total outbound operation time and maximizing overall shelf stability; S2: An improved multi-objective Mustang optimization algorithm is adopted, combined with conflict detection and correction strategies, to solve the multi-objective integrated optimization mathematical model of the cross-level shuttle access system in order to obtain an integrated optimization scheme for task sequencing, equipment scheduling and cargo location allocation. The conflict detection and correction strategy includes: recording the starting layer and the dwell layer of the shuttle hoist and shuttle after each task execution; performing conflict detection on the decoded individuals; for all shuttles equipped in the same roadway, for tasks of the same execution order, the following two situations may lead to conflict: 1) Shuttle B is present at the target cargo location layer of shuttle A's task. ;2) There are at least two shuttles with the same target cargo location layer; conflict detection is performed on a lane-by-lane basis; the layer where the task cargo location is located is obtained from the assigned cargo location number, and it is determined whether a conflict will occur based on the layer where the task cargo location is located and the layer where the shuttle is currently located; for detected conflicts, the individual code is corrected by changing the task cargo location number, thereby avoiding the conflict; for individuals that cannot be corrected by this method, a fitness penalty is applied. The complete steps of the solution method are as follows: S21: Set population size N stallion ratio PS Crossover probability PC Number of iterations ; S22: Randomly initialize the population and calculate the fitness function for each individual. , Conflict detection and correction are performed to obtain conflict indicators. con ,if con = 1, then ,in PE This is the penalty coefficient; S23: Perform a non-dominated ranking of all individuals based on their fitness, and obtain the Pareto set of all individuals. Select stallions and assign foals to form a herd; S24: Perform a non-dominated sort on all stallions to obtain the Pareto set of stallions. Randomly from Choose a location for a stallion as its habitat. WH ,from One stallion is randomly selected as... ; S25: Calculation ,in The coefficients decrease linearly from 1 to 0. This represents the maximum number of iterations for the algorithm. S26: For each herd Calculate the adaptive mechanism vector ; in, , , It is a vector consisting of 0s and 1s, with the same dimension as the variable. and Let [a] be a random vector that follows a uniform distribution in the interval [0,1]. A random number that follows a uniform distribution within the range [0,1]. To meet Conditional random vector The returned index; S27: For each foal ,if The formula for updating the foal grazing behavior is: Otherwise, the foal will stray from the herd and be moved to the stray archive. Archive; Among them, This indicates the current location of the foal in the herd. For the horse herd The position of the stallion; R It is a uniformly random number within the range Internally generated, Let the value of pi be 3.

14. A random number that follows a uniform distribution within the range [0,1]. S28: Yes Archive The foals in the herd mate, and the position of the offspring foals is obtained through a double-outbreak segmented multi-point crossover mechanism. The updated offspring foals are then assigned to herds with vacant positions that are different from their original positions. After that, the herd is cleared. Archive; S29: For each stallion If It will perform a Levi flight; otherwise, its position will be updated according to the following formula: ， in, The result is a random number that follows a uniform distribution within the range [0,1]; if the updated stallion's position dominates the original position, then... ; S210: Update the fitness of all individuals, perform conflict detection and repair, and obtain the final fitness; S211: For each foal in each herd, conduct herd leadership elections; S212: ,if Otherwise, return to step S24; or skip to step S213. S213: Perform a non-dominated ranking on all individuals to obtain ( Output .

2. The scheduling decision method for a cross-level shuttle access system according to claim 1, characterized in that, The objective functions of the multi-objective integrated optimization mathematical model include: total outbound operation time and overall shelf stability; the total outbound operation time function is: ； in For alleyway The completion time of the task queue. Number the lanes For alleyway Task queue length; The maximum overall stability of the shelving is expressed as: ； in, Indicates a lane The overall stability of the shelving unit The larger the value, the worse the overall stability of the shelving unit. In the total outbound operation time function The calculation is as follows: ； in, Indicates cargo hoist Execution tunnel The last task The starting time, Indicates cargo hoist Execution tunnel The last task The time taken; Indicates a lane Equipped with a cargo lift; The Calculation in progress Indicates cargo hoist Execution tunnel The task queue r Task The starting time, The calculation is as follows: ； in, Indicates shuttle Execute the task The moment of arrival at the outbound storage location level Indicates a lane Equipped with j Taiwan shuttle bus, Indicates task Located in the shuttle The task queue One task, express Execute the task The end time; For non-cross-layer tasks ,in, express Execute the task The starting time; The calculation formula is: ； in, express Execute the task The end time; For cross-level outbound tasks , With shuttle elevator Execute the task The start time Consistent, Indicates a lane Equipped with a shuttle lift, Indicates task Located in the shuttle elevator The task queue One task, Indicates shuttle lift Execute the task The time it takes; The calculation formula is: ； in, Indicates shuttle lift Execute the task The end time; The calculation formula is: ； in, and These represent the travel time of the shuttle elevator from its current floor to the floor where the shuttle is located, and the travel time from the floor where the shuttle is located to the floor where the target storage location is located, respectively. This indicates the time it takes for the shuttle car lift to load the shuttle car. This indicates the time it takes for the shuttle car elevator to unload the shuttle car; The Calculation in progress Indicates cargo hoist Execution tunnel The first r Task The time taken The calculation is as follows: ； in, and These represent the travel time of the cargo hoist from the I / O point of the aisle to the floor where the target cargo location is located, and the travel time from the floor where the target cargo location is located back to the I / O point of the aisle, respectively. This indicates the time it takes for the cargo elevator to load cargo. Indicates the time it takes for the cargo elevator to unload cargo; For shuttle Execute the task The time occupied for non-cross-layer outbound tasks is calculated as follows: ； This indicates the one-way travel time of the shuttle vehicle in the horizontal direction between the I / O point on its current floor and the target storage location. Indicates the time it takes for the shuttle to load cargo containers. Indicates the time it takes for the shuttle to unload the cargo box; For cross-level outbound tasks The calculation is as follows: 。 3. The scheduling decision method for a cross-level shuttle access system according to claim 2, characterized in that, The alley The overall stability function of the shelving group is: ； in, Indicates a lane The structural stability of the shelving unit Indicates a lane The overturning resistance of the shelving unit and The weights for the structural stability and anti-tipping properties of the shelving group are respectively, and they meet the following requirements. ; The structural stability function of the shelving group is: ； in, Z The total number of shelves. C This represents the total number of columns on the shelf. x Indicates the shelf number. y Indicates the shelf row number. z Indicates the shelf level number. Indicates the occupancy status of the storage space. If the storage location is occupied, return 1; otherwise, return 0. Indicates stored The quality of the goods at the storage location. Indicates the height of the shelf unit; The overturning resistance function of the shelving group is: 。 4. The scheduling decision method for a cross-level shuttle access system according to claim 3, characterized in that, The improved multi-objective Mustang optimization algorithm in step S2 specifically includes: S201: Design encoding and decoding methods for the task sequencing segment, elevator selection segment, shuttle selection segment, and cargo location allocation segment respectively; S202: Introduces a non-dominant ranking method, redefines the selection methods for stallions, foals, and habitats, and modifies the herd leadership election behavior; S203: Design a double outcry segmentation multi-point crossover mechanism to improve the mating and reproduction behavior of horse herds and increase the global search capability of the algorithm; S204: Introducing the Levy flight strategy improves the herd guidance behavior and increases the algorithm's ability to escape local optima.

5. The scheduling decision method for a cross-level shuttle access system according to claim 4, characterized in that, In step S202, the selection method for stallions, foals, and habitats is as follows: For herd initialization, all individuals in the herd are divided into different sets ( Stallions are preferred from In the first set, the remaining stallions are randomly selected from the individuals in the remaining set; the remaining members are randomly assigned to stallions as foals, forming different herds; then, the stallions in the herds are selected according to fitness; a non-dominated sort is performed on all stallions to obtain ( ), randomly from Choose one stallion as the habitat; the formula for the herd's election leadership behavior is: ； in, This indicates the current location of the foal in the herd. For the horse herd The position of the stallion.

6. The scheduling decision method for a cross-level shuttle access system according to claim 4, characterized in that, The double-outbreak segmented multi-point crossover mechanism in step S203 includes: mature foals are separated from their original herds and mated with mature foals from different herds to produce two offspring, which are then randomly added to two other herds different from their original herds; the specific steps of the segmented multi-point crossover are as follows: S231: Randomly select member 1 from the task sorting segment. The code is swapped with the code at the same position as member 2. ;in, Number of orders; S232: The elevator and shuttle car selection segment randomly selects member 1. The adjacent double code is swapped with the adjacent double code of member 2. ; S233: Randomly select member 1 for the cargo location allocation section. Each neighboring three-bit code, and member 2 is randomized. Three neighboring bits are swapped. .

7. The scheduling decision method for a cross-level shuttle access system according to claim 4, characterized in that, The Levi flight strategy in step S204 includes: performing a non-dominated sort on all stallions to obtain a Pareto set. ),from One stallion is randomly selected as... Instead of moving towards its habitat, it flew Levi instead.

8. The scheduling decision method for a cross-level shuttle access system according to claim 1, characterized in that, The specific steps of the conflict detection and correction strategy are as follows: Three shuttle cars, C11, C12, and C13, are installed in tunnel #1. S31: Calculate the number of detection cycles max_cycle = Number of tasks in the longest shuttle task queue; initialize conflict flags. con = 0, the floor where shuttles C11, C12, and C13 stop. ST 1. ST 2. ST 3; Order cycle 1; symbol This indicates assignment; S32: Obtain the current status of shuttles C11, C12, and C13. cycle Target storage level OT 1, OT 2, OT 3; S33: If OT 1 == ST 2 or OT 1 == ST 3. Then, search for the remaining storage locations in the same aisle where shuttle C11 is currently performing a task, randomly select one of these storage locations to replace the original one, and update the data. OT 1; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let con = 1, skip to step S37; S34: If OT 2 == ST 1 or OT 2 == ST 3 or OT 2 == OT 1. Then, search for the remaining storage locations in the same aisle where shuttle C12 is currently performing a task, randomly select one of these storage locations to replace the original storage location, and update the target storage location layer. OT 2; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let con = 1, skip to step S37; S35: OT 3 == ST 1 or OT 3 == ST 2 or OT 3 == OT 1 or OT 3 == OT 2. Then, search for the remaining storage locations in the same aisle where shuttle C13 is currently performing a task, randomly select one of these storage locations to replace the original storage location, and update the target storage location layer. OT 3; Repeat this step until the condition is no longer met; If the remaining number of units for the task is 0, then let con = 1, skip to step S37; S36: ST 1 OT 1, ST 2 OT 2, ST 3 OT 3; Order cycle cycle +1, if cycle max_ cycle Return to step S32; S37: End, output conflict flag. con And the revised cargo location allocation section.

Images