Task scheduling method and system for dense warehousing system

Abstract

The invention discloses a task scheduling method and system for a dense warehousing system, and the method comprises the steps: building an elevator task distribution model, a shuttle vehicle selection model and a warehouse-in and warehouse-out task time mathematical model according to the characteristics of warehousing operation, and preliminarily determining a task distribution scheme of an elevator through employing a heuristic rule on this basis, an improved A * algorithm is used for carrying out path planning on the shuttle vehicle, the time needed by the shuttle vehicle for completing a task is calculated, it is ensured that the elevator and the shuttle vehicle start to run, the scheme is further optimized through an improved PSO-SA algorithm, and the algorithm combines particle swarm optimization and a simulated annealing algorithm, so that the algorithm is optimized. A specific scheduling result of cooperative task execution of the elevator and the shuttle vehicle can be obtained, a scheduling scheme which enables the cooperative work effect of the whole system to be optimal is finally obtained, and maximization of the work efficiency is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of logistics, and more specifically, relates to a task scheduling method and system for an intensive warehousing system. Background Art

[0002] An intensive warehousing system is a new type of automated logistics storage system that combines high-density stereoscopic shelves, shuttle cars, elevators, conveyor belts, automatic barcode recognition systems, and a scheduling control system. It mainly includes the core equipment, namely four-way shuttle cars, vertical elevators, and corresponding shelves. The four-way shuttle car has a compact structure, a relatively fast running speed, and uses a wireless communication method. Multiple shuttle cars and elevators can work together to meet the application requirements of various scenarios. In an intensive warehousing environment, the four-way shuttle car can move in four directions on the shelf track and perform layer-changing operations in the shelves with the help of the elevator, greatly improving the flexibility of the automated warehousing layout and in-warehouse operations.

[0003] According to the functions and composition structures of the four-way shuttle car-based intensive warehousing system, it can be divided into three levels: the management level, the control level, and the equipment level. Among them, the management level is responsible for managing and scheduling various warehousing operations, providing data storage and information system support services for the warehousing system, and is the core part of the entire system; the control level includes auxiliary devices such as PLC, RFID, and sensors, which are responsible for controlling warehousing equipment and collecting and feedbacking information, playing a bridging role between the management level and the equipment level; the equipment level is composed of actual equipment such as shelves, shuttle cars, and elevators, which is the main part for performing inbound and outbound operations. The entire warehousing system processes the data fed back by the control level through the management level to realize the management and scheduling of tasks in the warehouse, and commands the equipment level to complete various operations through the control level.

[0004] In an intensive warehouse, equipment scheduling mainly focuses on four-way shuttle cars and elevators. These two types of equipment are the keys to achieving efficient logistics and can efficiently complete the handling and storage of items in a complex warehousing environment. The elevator realizes the handling of goods in the z-axis direction, and the shuttle car is responsible for the movement in the x and y axes. The task scheduling method assigns the elevator and the shuttle car system to transport goods, coordinates the task queues among multiple devices, to solve conflicts and deadlock problems, and improve the execution efficiency of inbound and outbound tasks. Task scheduling is the key to affecting the intelligent level of an automated warehousing system. In the collaborative operation of the two, it is very likely that the task allocation of the elevator is unbalanced, and goods are piled up in the roadway corresponding to a certain elevator, which in turn leads to congestion of the shuttle car; or the task allocation of the elevator is unreasonable, the operation efficiency is low, and it cannot match the high-efficiency index of the four-way shuttle car, resulting in the shuttle car being idle most of the time and the overall efficiency being low. Summary of the Invention

[0005] The present invention provides a task scheduling method and system for an intensive warehousing system. Aiming at the complex characteristics of multi-device collaborative scheduling, non-uniform motion of logistics equipment, corresponding time requirements, and combined inbound and outbound order tasks in the intensive warehousing system, the key points are to optimize the collaborative scheduling of elevators and shuttles, optimize order allocation, and optimize the waiting time of goods during the entire operation cycle. A mathematical model including elevator task allocation, shuttle selection, and task execution time is established to comprehensively describe the equipment collaborative operation problem in the intensive warehousing system, and a heuristic rule and PSO-SA hybrid algorithm are used to solve the model, combining the intuitiveness of the rule and the search ability of the algorithm, improving the quality of the scheduling scheme and the solution efficiency.

[0006] The present invention is implemented by the following technical solutions:

[0007] A task scheduling method for an intensive warehousing system is proposed, including:

[0008] Construct an elevator and shuttle system scheduling model, including: an elevator task allocation model, a collaborative shuttle selection model, and an inbound and outbound task execution time mathematical model;

[0009] Determine the elevator task allocation scheme based on heuristic rules; and perform path planning for the shuttle based on the improved A* algorithm;

[0010] Optimize the task through the improved PSO-SA hybrid algorithm to obtain the collaborative scheduling result of the elevator and the shuttle.

[0011] A task scheduling system for an intensive warehousing system is proposed, including:

[0012] A model construction unit for constructing an elevator and shuttle system scheduling model, including: an elevator task allocation model, a collaborative shuttle selection model, and an inbound and outbound task execution time mathematical model;

[0013] An elevator task allocation unit for determining the elevator task allocation scheme based on heuristic rules;

[0014] A shuttle path planning unit for performing path planning for the shuttle based on the improved A* algorithm and calculating the time required for the shuttle to complete the task;

[0015] A collaborative scheduling unit for optimizing the task through the improved PSO-SA hybrid algorithm to obtain the collaborative scheduling result of the elevator and the shuttle.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the task scheduling method and system of the intensive warehousing system proposed by the present invention, first, according to the characteristics of warehousing operations, a hoist task allocation model, a shuttle car selection model, and a mathematical model of the inbound and outbound task time are established. On this basis, a heuristic rule is used to preliminarily determine the task allocation plan of the hoist, and an improved A* algorithm is used to perform path planning for the shuttle car to ensure that the hoist and the shuttle car start running. Further, the task is optimized by an improved PSO-SA algorithm, which combines the particle swarm optimization and the simulated annealing algorithm, and can obtain the specific scheduling result of the hoist and the shuttle car cooperating to execute the task. Finally, a scheduling plan that makes the collaborative working effect of the entire system the best is obtained to ensure the maximization of work efficiency.

[0017] After reading the detailed description of the embodiments of the present invention in conjunction with the accompanying drawings, other features and advantages of the present invention will become clearer. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 Schematic diagram of the task route of the task scheduling method for the intensive warehousing system proposed by the present invention;

[0020] Figure 2 Example of map modeling of the intensive warehousing system using a grid map in the present invention;

[0021] Figure 3 Schematic diagram of the information of a grid storage location in the implementation of the present invention;

[0022] Figure 4 Schematic diagram of the environmental obstacle information of a loaded shuttle car in the embodiment of the present invention;

[0023] Figure 5 Schematic diagram of the environmental obstacle information of an empty shuttle car in the embodiment of the present invention;

[0024] Figure 6 Schematic diagram of the execution steps of the task scheduling method for the intensive warehousing system proposed;

[0025] Figure 7 Schematic diagram of the steps for determining the hoist task allocation plan based on heuristic rules in the method of the present invention;

[0026] Figure 8 Schematic diagram of the steps of the improved A* algorithm in the present invention;

[0027] Figure 9 It is a schematic diagram of the steps for determining the collaborative scheduling scheme of the elevator and the shuttle vehicle based on the PSO-SA algorithm in the method of the present invention. Specific implementation manner

[0028] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0029] In order to achieve the efficient collaborative operation of multiple elevators and multiple shuttle vehicles in a dense storage system, the present invention proposes a task scheduling method for a dense storage system. First, an elevator task allocation model, a shuttle vehicle selection model, and an inbound and outbound task time model are established. On this basis, a heuristic rule is used to preliminarily determine the task allocation scheme of the elevator, and an improved A* algorithm is used to perform path planning for the shuttle vehicle to ensure that the elevator and the shuttle vehicle can start running. Further, the preliminary task allocation scheme is optimized by an improved PSO-SA algorithm. Through the method of the present invention, the specific scheduling results of the elevator and shuttle vehicle system for executing tasks can be obtained, and finally a scheduling scheme that makes the collaborative work effect of the entire system the best can be obtained. The technical route is schematically shown as Figure 1 shown.

[0030] In the automated dense storage with the cooperation of multiple shuttle vehicles and multiple elevators, the expression of the operation environment is the basis for studying scheduling technology. After obtaining the environmental information and establishing the corresponding map model, such as information on order tasks, shelf storage status, elevator operation status, and shuttle vehicle operation path, etc., on this basis, the algorithm can be called and the data can be processed and analyzed.

[0031] To illustrate the scheduling method proposed by the present invention, in the embodiments of the present invention, a warehousing system scheduling model is first established according to the actual warehousing scheduling environment. Considering the characteristics of complex constraints, dynamic uncertainty, and multi-objectivity of the shelf structure and equipment operations in the actual scheduling process, in a dense warehousing, the shuttle runs in four directions along the current-layer track, and the sizes of the shuttle, track, and storage location correspond to the size of the pallet goods. From a simplified perspective, a single-layer warehouse in a four-way shuttle dense warehousing is grid-shaped. After dividing the grid into the size of a single storage location within the warehousing layer, the grid modeling method can accurately describe information such as the position of the shuttle and the storage state of the storage location. According to the layout characteristics of the dense warehousing system, the goods are stored in units of pallets according to the storage location, and the four-way shuttle runs along the longitudinal and transverse tracks distributed in the warehouse. Exactly one shuttle corresponds to the size of one storage location. Taking a single storage location as the minimum resolution unit, a grid map is used to perform map modeling on the four-way shuttle dense warehousing system, and the obtained model is as Figure 2 shown.

[0032] Then, the grid model is improved by combining the topology method, and the map is digitally stored using grid attributes and grid communication information. The former includes the category, position coordinates, and occupancy status of grid nodes, etc., and the latter is the direction information allowed for the shuttle to travel at this node.

[0033] Grid position information and type: In the grid map, not only the grid coordinates need to be recorded, but also information such as grid category, storage location occupancy status, storage location and aisle numbers, etc. needs to be included. Grid passage information: According to different grid categories, different passage direction information is set.

[0034] One of the main reasons for the high-efficiency operation of the dense warehousing shuttle is that the running track of the shuttle is located below the storage location. Whether the storage location is idle or not, the idle shuttle can pass through the track below the storage location, but the loaded shuttle can only pass at the idle storage location. Therefore, in order to improve the running efficiency of the shuttle and reduce congestion, it is necessary to obtain map information containing different obstacle information according to the inventory status of the storage location and the loading status of the shuttle, as shown in the passage status of different category nodes in the following table:

[0035] Table 1 Passage status of different category nodes

[0036]

[0037] Due to the limitations of the walking direction of the warehousing shelf and the walking characteristics of the shuttle, when there are obstacles in the aisle, this aisle cannot be passed. For example, when the aisle is not full of goods, the loaded shuttle cannot pass through the current aisle. However, when searching for a path, the idle storage locations in the current aisle will still be searched, resulting in redundant consumption of computing and storage space. At the same time, due to the limitations of the walking direction on the storage location, to avoid shuttle deadlocks, only a single shuttle is allowed to pass in the aisle.

[0038] InFigure 3 In the storage environment shown, for the load shuttle vehicle, when a lane is occupied or there is a shuttle vehicle in the lane, as Figure 4 shown, set the grid status corresponding to the entire lane as an obstacle, prohibiting other shuttle vehicles from passing, and reducing unnecessary node searches. For the empty-load shuttle vehicle, the shuttle vehicle can pass through the track under the stored cargo location. At this time, the obstacle nodes in the environment are as Figure 5 shown.

[0039] Suppose in a dense storage environment, after order placement and cargo location allocation, there are a total of k order tasks in the system that require outbound and inbound operations. The task coordinate set is T = {(x i , y i , z i ), i = 1, 2,..., k}. For the outbound task, this coordinate indicates which cargo location the shuttle vehicle should pick up the goods from. For the inbound task, this coordinate indicates which cargo location the shuttle vehicle should transport the goods to.

[0040] The number of shuttle vehicles in the warehouse system is N, and the number of elevators is M. The goal of building the model is to reasonably distribute and sort the inbound and outbound tasks in the system orders, fully mobilize the available elevators and shuttle vehicles to complete all inbound and outbound tasks in the shortest time, so as to ensure the maximization of the system working efficiency.

[0041] To achieve the scheduling optimization of the multi-elevator and shuttle vehicle system in a dense storage environment, it is necessary to establish a system scheduling model, including an elevator task allocation model, a collaborative shuttle vehicle selection model, and an elevator and inbound / outbound system operation time model.

[0042] The model satisfies the following assumptions:

[0043] (1) Consider that the elevator and the shuttle vehicle have certain acceleration and deceleration.

[0044] (2) Do not consider the loading time of the elevator and the shuttle vehicle.

[0045] (3) Do not consider the congestion interference between vehicles when multiple shuttle vehicles execute tasks in parallel, and simplify the model.

[0046] I. Build an elevator task allocation model.

[0047] The elevator operation task allocation model needs to comprehensively consider the balance of elevator tasks and the maximization of system efficiency, reasonably allocate k tasks to M elevator devices, so that the elevator operation tasks are overall balanced and the time required to complete the operation tasks is the least. Considering the acceleration and deceleration related data of the elevators and shuttle vehicles provided by the enterprise, the model goal not only minimizes the distance but also considers time, including acceleration, uniform motion, and deceleration time.

[0048] The calculation of the task time is based on the specific distance D ij (i.e., the distance from the current position of the elevator to the task cargo floor) to determine whether the maximum speed can be reached. If the distance D ij is less than the critical distance then only acceleration and deceleration are possible; otherwise, acceleration, constant speed, and deceleration are possible; where V is the maximum speed, a is the acceleration, and d is the deceleration.

[0049] (1) The case of accelerating to the maximum speed: D ij ≥D critical When the task distance D ij is greater than the critical distance, the time calculation is as follows:

[0050] Acceleration phase time:

[0051] Deceleration phase time:

[0052] Constant speed phase time:

[0053] (2) The case where the maximum speed cannot be reached: D ij <D critical , when the distance is insufficient, calculate the critical speed

[0054] The acceleration, constant speed, and deceleration times are respectively:

[0055] The total time is:

[0056] The objective function of the elevator task assignment is to minimize the total time:

[0057] where, T Lj represents the total task time of the jth elevator.

[0058] The above variables are explained in Table 2 below:

[0059] Table 2: Explanation of variables in the elevator task assignment model

[0060]

[0061]

[0062] II. Construct a collaborative shuttle vehicle selection model.

[0063] During the operation of the task, if there is no shuttle car on the current goods location layer, the system will select the existing shuttle car according to the current positions of the elevator and the shuttle car, the task execution status of the shuttle car, and the task time consumption of each shuttle car. Finally, the consideration of the acceleration and deceleration process is also required.

[0064] To optimize the cooperation between the elevator and the shuttle car, the task can be divided into two working modes: single-car mode and multi-car mode.

[0065] In the single-car mode, the elevator cooperates with a shuttle car. The key lies in selecting the shuttle car that is closest to the current task target (the entrance of the goods shelf layer) and can reach it in the shortest time. This requires calculating the scheduling time of the shuttle car from its current goods location to the target position and considering the completion time of the previous task of the shuttle car to accurately predict its availability. In this module of the present invention, this mode is adopted to establish a basic model.

[0066] In the multi-car mode, the elevator can schedule multiple shuttle cars simultaneously. This requires a comprehensive comparison of the scheduling times of all shuttle cars to determine the optimal combination to maximize the collaborative efficiency of the task. For this purpose, the scheduling coordination between shuttle cars and possible path conflicts need to be considered.

[0067] After comprehensively considering the above factors, select the shuttle car that can minimize the operation time in both modes, cooperate with the elevator to complete the inbound and outbound tasks, thereby improving the overall operation efficiency. The ultimate goal is to make the entire system efficient and flexible when executing tasks through an intelligent scheduling algorithm.

[0068] In the dense storage system, for the j-th elevator, let the node where its current position is located at the current moment be P j , and its coordinates are recorded as (P jx , P jy , P jz ). P ji represents the node of the i-th task target of the j-th elevator at the entrance of the goods shelf layer, and its coordinates are (P jix , P jiy , P jiz ). S ji represents the start time of the j-th elevator to execute the i-th task. The total number of shuttle cars is N, and C k represents the k-th shuttle car. P k represents the position of the entrance and exit of the k-th shuttle car on the current goods location layer of the j-th elevator, and its coordinates are (P kx , P ky , P kz ). P C(i-1) , E k(i-1)They respectively represent the coordinates and time when a task of the k-th shuttle vehicle is completed. For the i-th task performed by the j-th elevator, the task objective is to select a shuttle vehicle to execute the task and minimize the collaborative task completion time.

[0069] When the elevator and the shuttle vehicle collaborate to complete the warehousing task, first calculate the scheduling time of the shuttle vehicle, and then select the shuttle vehicle for collaborative operation. However, since there are the following two working modes when the elevator and the shuttle vehicle collaborate, it is necessary to consider the shuttle vehicle with the minimum operation time under these two working modes separately, and let it collaborate with the elevator to complete the warehousing and outbound tasks.

[0070] Mode 1: Load the goods first and then the vehicle.

[0071] In this scenario, the elevator first transports the goods to the storage location layer where the shuttle vehicle is located, and then transports the goods and the shuttle vehicle together to the target storage location layer. Based on this, the minimum call time t1 required for the j-th elevator to execute the i-th task and call the shuttle vehicle for collaborative operation is expressed as:

[0072]

[0073] Mode 2: Load the vehicle first and then the goods.

[0074] The elevator first picks up the collaborative shuttle vehicle and then goes to the inbound and outbound layer to pick up the goods and transports them to the target storage location layer. In this mode, the minimum call time t2 required for the j-th elevator to execute the i-th task and call the shuttle vehicle for collaborative operation is expressed as:

[0075]

[0076]

[0077] In the formula, t′1 and t′2 represent the waiting time at position P after the elevator arrives at the storage location layer where the k-th shuttle vehicle is located; V represents the running speed of the elevator; D(P k position), D(P L represent the walking distances of the elevator and the shuttle vehicle between the shelves at positions P L (P i ,P j ) and D C (P i ,P j ) respectively; D i is the critical distance required for the elevator to reach the maximum speed, and V j 、V′ critical are the peak speeds when the maximum speed cannot be reached in the two modes, as shown in the following formula: peak 、V′ peak are the peak speeds when the maximum speed cannot be reached in the two modes, as follows:

[0078] D L (P i ,Pj ) = |P iz -P jz |;

[0079] D C (P i , P j ) = |P ix -P jx | + |P iy -P jy |;

[0080]

[0081] During the collaborative outbound operation, only the nearest idle shuttle car can be transported to the target layer of the goods location by the elevator. After the shuttle car picks up the goods from the target goods location, the outbound task is completed by the elevator.

[0082]

[0083] In the formula, t′3 represents the waiting time at the position of P after the elevator reaches the goods location layer where the kth shuttle car is located. k Position waiting time.

[0084] The explanations of the above variables are shown in Table III below:

[0085] Table III: Explanation of variables in the collaborative shuttle car selection model

[0086]

[0087]

[0088]

[0089] III. Establish a mathematical model for the execution time of inbound and outbound tasks.

[0090] (1) Inbound task, the inbound model is: Min Max{C mk , T mk} (1)

[0091]

[0092] T mk = T m,k-1 + mt mk (5)

[0093]

[0094] C mk = L mk + ct mk (7)

[0095]

[0096] (2) Outbound task, the outbound model is:

[0097] Min Max{T mk} (9)

[0098]

[0099] As mentioned above, the objective function (1) represents minimizing the maximum task completion time for the inbound task; the constraint (2) restricts that a task can be assigned to at most one elevator;

[0100] The constraint (3) restricts that each elevator is assigned K tasks;

[0101] Equation (4) calculates the time required for the elevator to complete one inbound task. The time is divided into (initial position z m to l mk floor), (l mk floor to the target floor z i ), (target floor z i to the initial position z m ) into three parts.

[0102] It also includes the running time of the elevator between two floors calculation and l mk floor calculation;

[0103] Equation (5) represents the update of the task time window for the elevator to complete the task;

[0104] Equation (6) represents the update of the arrival time window of the elevator at the target floor;

[0105] Equation (7) represents the update of the task time window for the shuttle vehicle to complete the task, where is the time required for the shuttle vehicle to complete one task. For the inbound task, the shuttle vehicle first delivers the goods to the target position and then returns to the buffer area. For the outbound task, the shuttle vehicle first departs from the buffer area to pick up the goods and then delivers them to the elevator. The grid map adopted in the present invention can obtain the path through the improved A* algorithm proposed below, and then divide it by the average speed of the shuttle vehicle to obtain this time;

[0106] Equation (8) represents the update of the shuttle vehicle quantity distribution set when executing the inbound task;

[0107] The objective function (9) represents minimizing the maximum task completion time for the outbound task;

[0108] The constraint (10) restricts that a task can be assigned to at most one elevator;

[0109] The constraint (11) limits that each elevator is assigned to K tasks;

[0110] Equation (12) calculates the time required for the elevator to complete an outbound task, and divides the time into (

[0111] the time for the elevator to travel from the initial position to pick up the goods) and (

[0112] the time to return to the initial position after picking up the goods) into two parts;

[0113] Equation (13) represents the update of the task time window for the elevator to complete the task;

[0114] Equation (14) represents the update of the distribution set of the number of shuttle cars when performing the outbound task.

[0115] As mentioned above, the parameter explanations are shown in Table IV:

[0116] Table IV Parameter Explanation

[0117]

[0118]

[0119] In summary, according to the above equations, the time required for the elevator and the shuttle car to perform tasks under different scheduling strategies can be calculated when multiple elevators and shuttle cars are coordinated and scheduled, so as to complete the calculation of the total time for all tasks.

[0120] IV. Construct a collaborative scheduling algorithm for elevators and shuttle cars.

[0121] 1. Create a method for allocating elevator operation tasks based on heuristic rules.

[0122] When allocating tasks for multiple elevators, it is necessary to comprehensively consider the status of the elevator, the distance between the current position and the task position, and the influence of the positional relationship between task storage locations. To balance the task volume of each elevator and shorten the driving distance for the elevator to complete the task, based on the established elevator task allocation model above and considering the influence of the above factors, the following heuristic rules for elevator task allocation are designed to ensure the balanced allocation of tasks between each elevator and shuttle car, so as to improve the task execution efficiency.

[0123] Rule 1: The task is preferentially assigned to the elevator closer to the target storage location of the task.

[0124] Real-time track the working status of the elevator, including the task volume and real-time position coordinates, and assign the task to the elevator closest to the target storage location of the goods and with a smaller task volume, reducing the time for the shuttle car to transport the goods, so as to improve the overall task operation efficiency.

[0125] Rule 2: Tasks are preferentially assigned to the elevator with the least amount of tasks.

[0126] Based on Rule 1, if the amount of tasks assigned to the elevator closest to the inbound and outbound tasks exceeds the threshold, the task is assigned to the elevator with the least amount of tasks.

[0127] Rule 3: Tasks of the same layer and different types at the target storage location are preferentially assigned to the same elevator.

[0128] If the task target storage locations are on the same layer and the task types are different, they are assigned to the same elevator, which can improve the operation efficiency of the elevator and the shuttle vehicle.

[0129] Based on the above heuristic rules, the process of the elevator task assignment method is combined with Figure 7 as shown, including:

[0130] Step 1: After a new task order is generated, select task T from the task set and assign an elevator to it.

[0131] Step 2: Determine whether there are tasks on the same layer and of different types as T among the tasks already assigned to the current elevators. If so, preferentially assign task T to this elevator to reduce the equipment layer change time; if not, assign T to the elevator closer to its task storage location.

[0132] Step 3: Determine whether the difference in the amount of tasks between the elevator to which the task is to be assigned in Step 2 and the other elevators exceeds the threshold. If it exceeds the threshold, directly assign T to the elevator with the least amount of tasks to balance the workload of the warehousing equipment; if it does not exceed, proceed according to the plan in Step 2.

[0133] Step 4: Determine whether the task assignment is completed. If so, end the current process; if not, go to Step 1 and continue the elevator task assignment.

[0134] 2. Create a shuttle vehicle path planning method based on the improved A* algorithm.

[0135] The present invention searches for the path of the shuttle vehicle through the improved A* algorithm (and then dividing by the average speed of the shuttle vehicle can obtain the time ct required for the shuttle vehicle to complete the task mk ). During the search process, the starting point S is established as the parent node. By estimating the value of the adjacent node n, the one with the maximum value is selected as the next parent node until the target node G is searched. The A* algorithm maintains two lists through the heuristic function. The nodes to be detected during the path planning process are stored in the Openlist, while the nodes that have been detected are stored in the Closelist. The algorithm search process is:

[0136] f(n) = s(n) + g(n)

[0137] Where: f(n) represents the total moving distance through the current node; s(n) represents the moving distance from the initial point to the current node n; g(n) represents the estimated distance from node n to the target point.

[0138] When the A* algorithm searches for nodes, it selects the node with the smallest f(n) as the next node each time. In f(n), g(n) plays a decisive role in the search result. In the dense storage system, the four-way shuttle travels along the three axes x, y, and z. Therefore, in this paper, the Manhattan distance g(n) is selected to calculate the estimated distance from node n to the target point, which is expressed as the sum of the distances between two coordinate points along the axis directions.

[0139] g(n) = |x g - x n | + |y g - y n |

[0140] For specific reference, Figure 6 see the schematic diagram of the A* algorithm steps shown.

[0141] When the A* algorithm estimates the path cost according to the traditional heuristic function, it does not consider the time consumption caused by the shuttle turning. The standard A* algorithm has the problem of low search efficiency. Introducing a static weight into the heuristic function can effectively accelerate the algorithm search speed. However, when the search node is close to the target point, the static weight causes the heuristic function to be increasingly smaller than the actual minimum path cost, reducing the path diversity.

[0142] During actual operation, the shuttle needs to first decelerate to a stop at the turning point for a reversing operation and then accelerate to continue driving. First, the frequently occurring turning points will limit the time for the shuttle to operate at the maximum driving speed. Even in the straight-line distance between two turning points, the shuttle may not be able to achieve high-speed driving at all. Second, based on the characteristics of the shelf track in the dense storage, the inflection point must be at the track point. And the process of the shuttle changing speed at the inflection point is relatively slow, which not only increases the path time consumption but also easily causes vehicle waiting congestion. Therefore, too many inflection points in the path will seriously affect the operation efficiency of the shuttle.

[0143] The present invention reduces the number of turns by adding a turning penalty term to the heuristic function of the A* algorithm. When the shuttle performs path planning, it is known that the shuttle is at the current node now(x n , y n ), the previous node is pre(x p , y p ), and the candidate node is next(x nx , y nx)。As shown in the following formula, the turning cost is denoted as turnCost(n). If there is no turn at node n, the value is 0; if there is a turn, the time cost of the shuttle turning at node n is converted into the path cost of a storage grid, that is, the value is 1. As shown in the following formula, the turning cost is incorporated into the heuristic function:

[0144] f(n) = s(n) + g(n) + turnCost(n)

[0145] In order to balance the search efficiency of the algorithm and the diversity of path individuals, the present invention also proposes a dynamic weight balancing strategy to improve the heuristic function. The basic idea of the dynamic weight balancing strategy is: introducing a weight factor to accelerate the search efficiency of the algorithm, and making the weight factor dynamic according to the relative positions of the node, the starting point and the target point; according to the different tasks of the shuttle, the weight factor can be adjusted adaptively; and through the decay property of the exponential function, the effect of balancing the search efficiency and the diversity of search paths is achieved.

[0146] On the basis of the original formula, a dynamic weight factor is introduced to improve g(n), and the value of the weight is adjusted adaptively according to the distance between the current node and the target point. At the initial stage of the search, the weight value is large, and g(n) in the heuristic function accounts for a relatively large proportion, accelerating the search speed of the algorithm towards the end point g; when the search node gradually approaches the target node, the weight factor gradually decreases, and more nodes are expanded in the path search, increasing the diversity of path solutions. The improved heuristic function is as follows:

[0147] where x g , y g are the coordinates of the current node, x n , y n are the coordinates of the target point, and x s , y s are the coordinates of the starting point.

[0148] 3. Create a task scheduling method based on the PSO-SA hybrid algorithm.

[0149] The optimization goal of the vehicle task scheduling problem is to minimize the travel distance of the shuttle vehicle while minimizing the cargo access time, so as to improve the logistics efficiency and working efficiency of the warehouse. PSO (Particle Swarm Optimization Algorithm) can be used to solve this problem, but the traditional PSO is prone to falling into local optimal solutions and has a slow convergence speed; SA (Simulated Annealing Algorithm), as an algorithm for solving combinatorial optimization problems, searches for the global optimal solution of the problem by simulating the physical annealing process and continuously reducing the temperature. By hybridizing PSO and SA, the advantages of both algorithms can be fully utilized to improve the search ability and optimization effect of the algorithm. Specifically, PSO can be used for global search and local optimization, while SA can be used to jump out of local optimal solutions, thus avoiding the algorithm from falling into local optimal solutions and improving the convergence speed and optimization effect of the algorithm.

[0150] The basic process of this algorithm is as follows:

[0151] Step 1: Regard the task sequence as a particle, and initialize the position and velocity of the particle. Determine the particle population size X and the dimension Y, and set the current iteration number and the maximum iteration number U.

[0152] Step 2: Calculate the fitness of each particle, and assign the fitness of the current particles to p i , and store the optimal fitness values of all particles in g.

[0153] Step 3: Judge whether the global optimal value has stagnated or reached the specified iteration number. If it is satisfied, go to 6.

[0154] Step 4: Perform the next generation of iterative calculation. If the maximum iteration number of PSO is reached, go to 6.

[0155] Step 5: According to the roulette wheel strategy, determine a substitute value g of the global optimum from each particle, and then update the velocity and position of each particle.

[0156] Step 6: Set the initial temperature and the ending temperature.

[0157] Step 7: At the current temperature, calculate the fitness values of each particle, and update the solution according to the Metropolis criterion.

[0158] Step 8: Judge whether the search stops according to whether the SA stop condition is met (the stop condition is generally the preset operation accuracy or the program iteration number). If it is not met, go to 7.

[0159] Step 9: Reduce the temperature and return to 7 until the optimal solution is obtained or the lowest temperature is reached.

[0160] The overall flowchart is as Figure 9 shown.

[0161] Based on the above, the present invention further provides a task scheduling system for a dense storage system, including:

[0162] A model construction unit for constructing a hoist and shuttle car system scheduling model, including: a hoist task allocation model, a collaborative shuttle car selection model, and a mathematical model for the execution time of inbound and outbound tasks;

[0163] A hoist task allocation unit for determining a hoist task allocation scheme based on heuristic rules;

[0164] A shuttle car path planning unit for performing path planning on the shuttle car based on the A* algorithm and calculating the time required for the shuttle car to complete the task;

[0165] A collaborative scheduling unit for optimizing the tasks through an improved PSO-SA hybrid algorithm to obtain a collaborative scheduling result of the hoist and the shuttle car.

[0166] The method for the system to perform task scheduling has been described in detail above and will not be elaborated here.

[0167] It should be noted that in the specific implementation process, the above control part can be implemented by a processor in hardware form executing computer execution instructions in software form stored in a memory, which will not be elaborated here, and the programs corresponding to the actions executed by the above control circuit can all be stored in the computer-readable storage medium of the system in software form, so as to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0168] The computer-readable storage medium mentioned above may include volatile memory, such as random access memory; it may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive; it may also include a combination of the above types of memory.

[0169] The processor mentioned above can also be a general term for multiple processing elements. For example, the processor can be a central processing unit, or it can be other general-purpose processors, digital signal processors, application-specific integrated circuits, field-programmable gate arrays, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and can also be a dedicated processor.

[0170] It should be pointed out that the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions, or substitutions made by those of ordinary skill in the art within the scope of the essence of the present invention should also fall within the protection scope of the present invention.

Claims

1. A task scheduling method for an intensive warehousing system, characterized in that, Including: Constructing a hoist and shuttle car system scheduling model, including: a hoist task allocation model, a collaborative shuttle car selection model, and a mathematical model for the execution time of inbound and outbound tasks; Determining the hoist task allocation plan based on heuristic rules; and, performing path planning for the shuttle car based on the improved A* algorithm to calculate the time required for the shuttle car to complete the task; Optimizing the task through the improved PSO-SA hybrid algorithm to obtain the collaborative scheduling result of the hoist and the shuttle car.

2. The task scheduling method of the intensive warehousing system according to claim 1, characterized in that, The constructed hoist task allocation model is: Denote the total task time of the j-th elevator; is the time of the acceleration phase, is the time of the constant-speed phase, is the time of the deceleration phase.

3. The task scheduling method of the intensive warehousing system according to claim 1, characterized in that The constructed collaborative shuttle car selection model is: Inbound operation of loading goods first and then the car: Inbound operation of loading the car first and then goods: Wherein, t′1 and t′2 represent the waiting time at position P after the elevator reaches the layer where the k-th shuttle vehicle is located k ; V L represents the running speed of the elevator; D L (P i , P j ) and D C (P i , P j ) respectively represent the walking distances of the elevator and the shuttle vehicle between the shelves at positions P i and P j ; D critical is the critical distance required for the elevator to reach the maximum speed, and V peak , V′ peak are the peak speeds when the maximum speed cannot be reached in two modes; Outbound operation: Wherein, t′3 represents the waiting time at the P k position after the elevator arrives at the cargo location floor where the k-th shuttle vehicle is located.

4. The task scheduling method of the intensive warehousing system according to claim 1, characterized in that The constructed mathematical model for the execution time of inbound and outbound tasks includes an inbound model and an outbound model; The warehousing model is: Min Max{C mk ,T mk}, and the constraints include: Limit the assignment of a task to at most one elevator; Limit each elevator to be assigned to K tasks; Limit the time required for the elevator to complete one cycle without tasks, as the initial position z m to l mk floor time, for l mk floor to the target floor z i time, for the target floor z i to the initial position z m time; which also includes the running time of the elevator between two floors Calculate and l mk floor calculation: T mk = T m,k-1 + mt mk indicates that the hoist completes the update of the task time window; Indicates the update of the time window for the hoist to reach the target floor; C mk = L mk + ct mk Expression for updating the task time window of the shuttle vehicle, where is the time required for the shuttle vehicle to complete one task; Indicates the update of the distribution set of the number of shuttle cars when performing the warehousing task; The outbound model is MinMax{T mk}; The constraints include: Limit the assignment of a task to at most one elevator; Limit each elevator to be assigned to K tasks; Calculate the time required for the elevator to complete one outbound task, is the time for the elevator to travel from the initial position to pick up the goods, is the time to return to the accident location after picking up the goods; which also includes the running time of the elevator between two floors Calculate and l mk Calculation for the l-th floor: T mk = T m,k-1 + mt mk indicates the update of the task time window completed by the hoist; Indicates the update of the distribution set of the number of shuttle cars when performing the outbound task; The explanations of the above parameters are as follows:

5. The task scheduling method of the intensive warehousing system according to claim 4, characterized in that, The heuristic function in the improved A* algorithm is: f(n) = s(n) + e α · g(n) + turnCost(n); turnCost(n) is the turning cost, which is 0 when there is no turn at node n and 1 when there is a turn at node n; e α is the weight factor of the relative positions of the node with respect to the start point and the target point; x g y g are the coordinates of the current node, x n y n are the coordinates of the target point, x s y s are the coordinates of the start point.

6. The task scheduling method of the intensive warehousing system according to claim 4, characterized in that, The following model is used to calculate the operation time of the hoist and the shuttle car: Inbound task: tL ji = min(t1, t2); Outbound task:

7. The task scheduling method of the intensive warehousing system according to claim 1, characterized in that Determining the hoist task allocation plan based on heuristic rules, including: Step 1: After a new task order is generated, select task T from the task set and call the hoist task allocation model to allocate a hoist for it; Step 2: Determine whether there is a task on the same floor and of a different type from T among the tasks already allocated to the current hoists; if so, give priority to allocating task T to this hoist to reduce the equipment layer change time; if not, allocate T to the hoist closer to its task location; Step 3: Determine whether the difference in the task volume between the hoist to which the task is to be allocated in Step 2 and the other hoists exceeds the threshold; if it exceeds the threshold, directly allocate T to the hoist with the least task volume to balance the workload of the warehousing equipment; if it does not exceed, proceed according to the plan in Step 2; Step 4: Determine whether the task allocation is completed; if so, end the current process; if not, go to Step 1 and continue with the task allocation of the hoist.

8. The task scheduling method of the intensive warehousing system according to claim 1, characterized in that The improved PSO-SA hybrid algorithm includes: Step 1: Regard the task sequence as a particle, and initialize the position and velocity of the particle; determine the particle population size X and dimension Y, and set the current iteration number and the maximum iteration number U; Step 2: Calculate the fitness of each particle and assign the fitness of the current particles to p i , and store the optimal fitness values of all particles in g; Step 3: Determine whether the global optimal value is stagnant or reaches the specified iteration number, and if so, go to 6; Step 4: Perform the next-generation iterative calculation. If the maximum iteration number of PSO is reached, go to 6; Step 5: According to the roulette wheel gambling strategy, determine a substitution value g of the global optimum from each particle, and then update the velocity and position of each particle; Step 6: Set the initial temperature and the end temperature; Step 7: At the current temperature, calculate the fitness value of each particle and update the solution according to the Metropolis criterion; Step 8: Determine whether the search stops according to whether the SA stop condition (the stop condition is generally the pre-set operation accuracy or the program iteration number) is satisfied. If not, go to 7; Step 9: Cool down the temperature and return to 7 until the optimal solution is obtained or the lowest temperature is reached.

9. A task scheduling system for an intensive warehousing system, characterized in that, Including: A model construction unit for constructing a hoist and shuttle car system scheduling model, including: a hoist task allocation model, a collaborative shuttle car selection model, and a mathematical model for the execution time of inbound and outbound tasks; The elevator task allocation unit is used to determine the elevator task allocation plan based on heuristic rules; The shuttle path planning unit is used to perform path planning for the shuttle based on the improved A* algorithm and calculate the time required for the shuttle to complete the task; The cooperative scheduling unit is used to optimize the tasks through the improved PSO-SA hybrid algorithm to obtain the cooperative scheduling result of the elevator and the shuttle.

Images