A method, system, device and medium for optimizing allocation of a stereoscopic warehouse location

Abstract

The application discloses a kind of stereoscopic warehouse goods location allocation optimization method, system, equipment and medium, it is related to intelligent warehousing and logistics scheduling technical field, including: in the process of equipment execution warehousing operation, real-time acquisition equipment actual track point and the track deviation amount between equipment pre-planned track, when the track deviation amount accumulates exceeds preset deviation threshold, with current actual equipment state parameter set as initial state, jump to step S2 and re-perform track pre-planning and goods location allocation.The application is by real-time acquisition equipment actual track point and the track deviation amount between pre-planned track, when cumulative deviation is over threshold, with current actual equipment state parameter set as initial state trigger re-planning, form closed loop deviation feedback mechanism, effectively guarantee the robustness and dynamic adaptability of system operation.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing and logistics scheduling technology, and in particular to a method, system, equipment and medium for optimizing the allocation of storage locations in an automated warehouse. Background Technology

[0002] As modern logistics rapidly evolves towards intelligence and intensification, the operational efficiency of automated storage and retrieval systems (AS / RS), as a core node in automated warehousing, directly determines the supply chain response speed and operating costs. In scenarios such as smart manufacturing, e-commerce fulfillment, and cold chain logistics, the characteristics of goods being diverse, small-batch, and high-frequency are becoming increasingly prominent, leading to high complexity in the coordinated scheduling of inbound and outbound equipment and the allocation of storage locations. Traditional storage location allocation mainly relies on static rules such as ABC partitioning based on turnover rate or fixed physical attribute matching. While simple to implement, it is difficult to dynamically respond to multi-dimensional coupled constraints such as equipment status, aisle traffic flow, and task timing, resulting in frequent empty runs of equipment, aisle congestion, and task waiting, which seriously restricts the further improvement of the throughput capacity of AS / RS.

[0003] Current research on the optimization of storage location allocation in automated warehouses mainly follows two technical routes. One is offline optimization methods based on mathematical programming or heuristic algorithms. By establishing a mixed-integer programming model and using metaheuristic solutions such as genetic algorithms and particle swarm optimization, a relatively optimal static solution can be obtained under a given batch of tasks. The other is rule-based online dynamic adjustment methods, which use priority rules or fuzzy logic to adjust equipment paths and storage location selection based on real-time status. Some studies also introduce multi-agent cooperative control to alleviate equipment resource competition. However, the above methods generally do not adequately consider spatiotemporal dynamic factors and fail to explicitly correlate and jointly optimize the continuous movement trajectory of equipment with the time window for allocating storage locations. In actual operation, it is difficult to balance global optimality and dynamic adaptability.

[0004] As the number of devices increases and the degree of lane sharing deepens, existing technologies have exposed several key bottlenecks. When allocating storage locations, devices are treated as instantaneously reachable ideals, ignoring the continuous temporal characteristics of their three-dimensional spatial trajectories and the interference relationships between multiple device trajectories in the spatiotemporal domain. This leads to frequent path conflicts or failures to arrive at storage locations on time during allocation, forcing the system to perform numerous online rescheduling operations and significantly increasing control overhead. Traditional screening only considers the physical attributes of storage locations and the static matching of goods, failing to incorporate allocable time windows as pre-constraints. This results in physically available storage locations failing to be filled due to devices not being able to reach them within the specified time limit, causing task timeouts and inventory backlogs. More critically, existing replanning mechanisms mostly rely on periodic global recalculation or event-triggered local repairs, lacking the ability to finely perceive and close-loop correct the accumulated deviations between the actual execution trajectory of the devices and the pre-planned trajectory. This makes it difficult to intervene in time when deviations first appear, leading to a chain reaction of spatiotemporal baseline shifts in subsequent tasks, ultimately affecting the system's operational stability and the precise control of work rhythm.

[0005] Therefore, it is essential to invent a method, system, equipment, and medium for optimizing the allocation of storage locations in automated warehouses to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, equipment, and medium for optimizing the allocation of storage locations in automated warehouses, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for optimizing the allocation of storage locations in an automated warehouse, characterized by comprising the following steps:

[0008] S1: Obtain the current location status tensor of the automated warehouse, the cargo attribute vector of the goods to be put into the warehouse, and the current equipment status parameter set of the inbound and outbound equipment. The current location status tensor includes the occupancy flag and remaining occupancy time of each location. The current equipment status parameter set includes the position coordinates, running speed, and remaining execution time of the current task of each equipment.

[0009] S2: Using the current equipment status parameter set as input, pre-plan the spatiotemporal motion trajectory of each inbound / outbound equipment in the preset prediction time domain, and generate the equipment pre-planned trajectory for each equipment. The equipment pre-planned trajectory includes a sequence of trajectory points, and each trajectory point is associated with a timestamp and spatial coordinates.

[0010] S3: Based on the intersection and union of the pre-planned trajectories of all equipment in the spatiotemporal coordinate system, calculate the set of passable time windows for each spatiotemporal grid unit, and reverse-map the set of passable time windows to each location of the automated warehouse to generate the set of allocable time windows for each location.

[0011] S4: Match the cargo attribute vector with the current storage location state tensor. Under the constraint of the allocable time window set, select a set of candidate storage locations for each cargo to be put into storage that meet the cargo attribute matching and have a non-empty time window. Use the minimum spatial deviation between the storage location center coordinates in the candidate storage location set and the pre-planned trajectory of the corresponding equipment as the objective function to determine the target storage location and the corresponding target equipment.

[0012] S5: Based on the target storage location and the target equipment, insert a dwell trajectory point on the pre-planned trajectory of the target equipment. The dwell time of the dwell trajectory point is equal to the operation time required to transport the goods from the inlet to the target storage location. Update the pre-planned trajectory of the equipment to obtain the updated equipment execution trajectory. Control the equipment to perform the inlet operation according to the equipment execution trajectory.

[0013] S6: During the process of the equipment performing the warehousing operation, the trajectory deviation between the actual trajectory point of the equipment and the pre-planned trajectory of the equipment is collected in real time. When the trajectory deviation exceeds the preset deviation threshold, the current actual equipment status parameter set is used as the initial state, and the process jumps to step S2 to re-plan the trajectory and allocate the storage location.

[0014] Preferably, the pre-planning of the spatiotemporal motion trajectory in S2 further includes: discretizing the preset prediction time domain into equally spaced time steps; within each time step, using the current position of the equipment as the starting point and the entrances of each lane of the automated warehouse as the ending point, generating multiple candidate trajectories using a dynamic programming method; calculating the weighted sum of the travel time and energy consumption of each candidate trajectory; and selecting the candidate trajectory with the smallest weighted sum as the pre-planned trajectory of the equipment; when multiple devices share the same lane, adding anti-collision time interval constraints during the candidate trajectory generation process, so that the time interval between the trajectory points of different devices on the same spatiotemporal coordinate is greater than the passage time corresponding to the safe distance of the equipment.

[0015] Preferably, generating the set of allocable time windows for each storage location in S3 includes: mapping each storage location to its corresponding shelf column, layer, and depth; calculating the detour time required for the equipment to move from the trajectory point to the storage location's retrieval point based on the spatial distance between the trajectory point in the equipment's pre-planned trajectory and the retrieval point at the front of the storage location; marking the timestamp intervals corresponding to trajectory points whose detour time is less than the equipment's remaining task buffer time as the allocable time windows for that storage location; when multiple equipment pre-planned trajectories can all reach the same storage location, merging and arranging the allocable time windows corresponding to each equipment in chronological order, and using the merged time windows as the set of allocable time windows for that storage location.

[0016] Preferably, the screening of the candidate storage location set in S4 further includes: extracting three dimensions of features from the cargo attribute vector: cargo volume, cargo weight, and cargo turnover frequency; using the volume capacity, load-bearing limit, and aisle location of each available storage location in the current storage location state tensor as the constraint boundaries of the corresponding dimensions; and removing storage locations that do not meet any dimension constraint to obtain a preliminary storage location set; for each storage location in the preliminary storage location set, checking whether its allocable time window set contains an available time window that is greater than the cargo entry deadline; if it does, then retaining the storage location in the candidate storage location set.

[0017] Preferably, determining the target storage location and target equipment in step S4 further includes: for each storage location in the candidate storage location set, calculating the shortest spatial deviation from the storage location to the current pre-planned trajectory of each piece of equipment. The spatial deviation is calculated by: traversing each trajectory point on the pre-planned trajectory of the equipment, calculating the Euclidean distance between the coordinates of the trajectory point and the coordinates of the storage location's pickup point, and taking the minimum Euclidean distance among all trajectory points as the spatial deviation of the equipment from the storage location; forming a deviation matrix from the spatial deviations of each equipment to each storage location, and using the Hungarian algorithm to perform bipartite graph matching on the deviation matrix to obtain the storage location-equipment pair that minimizes the total deviation as the target storage location and target equipment.

[0018] Preferably, the insertion of the dwell trajectory point in S5 includes: obtaining the trajectory point with the smallest spatial deviation from the target storage location in the current pre-planned trajectory of the target equipment as the insertion anchor point; shifting the timestamps of all trajectory points after the insertion anchor point backward by the operation duration; inserting the dwell trajectory point at the position immediately adjacent to the insertion anchor point; setting the spatial coordinates of the dwell trajectory point to the picking point coordinates of the target storage location; and setting the dwell duration attribute of the dwell trajectory point to the operation duration; after the insertion is completed, restoring the spatial coordinates of the next trajectory point after the dwell trajectory point to the coordinates of the first trajectory point after the original insertion anchor point, thus forming the updated equipment execution trajectory.

[0019] Preferably, the real-time acquisition of trajectory deviation in S6 further includes: within each control cycle of the device, acquiring the actual position coordinates of the device and the expected position coordinates corresponding to the current timestamp in the pre-planned trajectory of the device, calculating the vector difference between the actual position coordinates and the expected position coordinates as the single-step deviation; taking the vector and magnitude of all single-step deviations within a preset time window as the cumulative trajectory deviation; when the cumulative trajectory deviation exceeds a preset deviation threshold, triggering a replanning flag, and replacing the position coordinates in the current actual device status parameter set with the actual position coordinates, and replacing the remaining execution time of the current task with the remaining time re-estimated based on the actual trajectory.

[0020] A system for optimizing the allocation of storage locations in an automated warehouse includes the following modules:

[0021] The data acquisition module acquires the current location status tensor of the automated warehouse, the cargo attribute vector of the goods to be received, and the current equipment status parameter set of the inbound and outbound equipment.

[0022] The trajectory pre-planning module takes the current equipment status parameter set as input and pre-plans the spatiotemporal motion trajectory of each inbound / outbound equipment in the preset prediction time domain, generating the equipment pre-planned trajectory for each equipment.

[0023] The time window reverse mapping module calculates the passable time window set for each spatiotemporal grid unit based on the pre-planned trajectory of all equipment, and reverse maps the passable time window set to each cargo location to generate an allocable time window set for each cargo location.

[0024] The cargo location screening and matching module matches the cargo attribute vector with the current cargo location state tensor, filters the candidate cargo location set under the constraints of the allocable time window set, and determines the target cargo location and the corresponding target equipment with the goal of minimizing spatial deviation.

[0025] The trajectory update and execution module inserts a stop trajectory point on the pre-planned trajectory of the target device to obtain the updated device execution trajectory, and controls the device to perform the warehousing operation according to the device execution trajectory;

[0026] The deviation monitoring and replanning module collects the trajectory deviation between the actual trajectory points of the device and the pre-planned trajectory of the device in real time. When the cumulative trajectory deviation exceeds the preset deviation threshold, the trajectory preplanning unit is triggered to restart execution with the current actual device state parameter set.

[0027] An electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for optimizing the allocation of storage locations in an automated warehouse as described above.

[0028] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for optimizing the allocation of storage locations in an automated warehouse as described above.

[0029] The technical effects and advantages of this invention are as follows:

[0030] 1. This invention obtains the current storage location status tensor, cargo attribute vector, and equipment status parameter set of the automated warehouse, and integrates the remaining storage location occupancy time and equipment operating status into the input data model, providing multi-dimensional and accurate initial conditions for subsequent dynamic spatiotemporal planning;

[0031] 2. This invention pre-plans the spatiotemporal motion trajectory of each inbound / outbound device within a preset prediction time domain, and adds anti-collision time interval constraints when multiple devices share a lane, generating a pre-planned trajectory of the device containing timestamps and spatial coordinates, thereby achieving forward-looking coordination and conflict avoidance of device movement in the spatiotemporal domain.

[0032] 3. This invention calculates the set of passable time windows of spatiotemporal grid cells and maps them inversely to each storage location to generate a set of allocable time windows for each storage location. It explicitly associates the reachability time of the equipment with the physical location of the storage location, providing precise time window constraints for storage location selection.

[0033] 4. This invention performs multi-dimensional matching of cargo attributes and cargo location status under the constraint of allocable time window set, and determines the target cargo location and target equipment with the goal of minimizing the spatial deviation between the cargo location center coordinates and the pre-planned trajectory of the equipment, taking into account both cargo adaptability and equipment path deviation optimization.

[0034] 5. This invention achieves smooth and conflict-free embedding of task insertion into the original spatiotemporal trajectory by inserting a dwell trajectory point into the pre-planned trajectory of the target equipment and setting the dwell time to be equal to the warehousing operation time, updating and generating the equipment execution trajectory, and controlling the equipment operation according to the trajectory.

[0035] 6. This invention collects the trajectory deviation between the actual trajectory points of the device and the pre-planned trajectory in real time. When the cumulative deviation exceeds the threshold, it triggers replanning with the current actual device state parameter set as the initial state, forming a closed-loop deviation feedback mechanism, which effectively ensures the robustness and dynamic adaptability of the system operation. Attached Figure Description

[0036] Figure 1 This is a flowchart illustrating the overall method of the present invention;

[0037] Figure 2 This is a flowchart of the cargo location screening and matching decision-making process of the present invention;

[0038] Figure 3 This is a flowchart of the deviation monitoring and replanning triggering mechanism of the present invention;

[0039] Figure 4 This is a system module architecture diagram of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example 1

[0042] This embodiment provides a method for optimizing the allocation of storage locations in an automated warehouse. This method can be implemented in a warehouse management system (WMS) or warehouse control system (WCS), such as... Figure 1 As shown, the specific steps include S1 to S6.

[0043] S1: Obtain the current location status tensor of the automated warehouse, the cargo attribute vector of the goods to be put into the warehouse, and the current equipment status parameter set of the inbound and outbound equipment. The current location status tensor includes the occupancy flag and remaining occupancy time of each location. The current equipment status parameter set includes the position coordinates, running speed, and remaining execution time of the current task of each equipment.

[0044] In this step, the system acquires three types of structured data in real time through a sensor network, PLC controller, and upper-level database interface deployed at the automated warehouse:

[0045] Current location status tensor: Defined as a three-dimensional data structure S, with dimensions R rows × C columns × L layers. Each element in this tensor is a tuple (Flag, T). remain ), where Flag takes a value of 0 or 1, indicating the occupancy flag (1 for occupied, 0 for idle); T remain This represents the remaining occupied time in seconds; the value for an idle storage space is 0.

[0046] Cargo attribute vector: For each cargo unit to be received into the warehouse, extract its feature vector V. cargo =[v, w, f], where v represents the volume of goods (cubic meters), w represents the weight of goods (kilograms), and f represents the turnover frequency of goods (times / day) calculated based on historical order data.

[0047] Current equipment status parameter set: For each piece of inbound / outbound equipment (such as stacker crane, shuttle car, hoist), its status parameter set D is collected. i ={P i (x, y, z), Vel i T task，remain} represent the device's current three-dimensional position coordinates, current running speed vector, and remaining execution time (in seconds) of the current task, respectively.

[0048] In the current location status tensor, each location element is also associated with a location static attribute table. This attribute table uses the location ID as the primary key and is pre-stored in the automated warehouse's configuration database. It contains the following fields: location volume capacity V. slot (Unit: cubic meters), Maximum load-bearing capacity of the cargo space (W) slot (Unit: kg) Three-dimensional coordinates P of the exit of the aisle where the cargo location is located. exit (x) exit y exit , z exit During constraint filtering in step S4, the system queries the static attribute table of the storage location in real time based on the available storage location ID in the current storage location status tensor to obtain the constraint boundaries of the corresponding dimension.

[0049] S2: Using the current equipment status parameter set as input, pre-plan the spatiotemporal motion trajectory of each inbound / outbound equipment in the preset prediction time domain, and generate the equipment pre-planned trajectory for each equipment. The equipment pre-planned trajectory includes a sequence of trajectory points, and each trajectory point is associated with a timestamp and spatial coordinates.

[0050] Using the current device state parameter set obtained in step S1 as the initial state input, this step performs forward planning for the device motion within a preset prediction time domain (e.g., the next 120 seconds).

[0051] The specific planning algorithm details are as follows:

[0052] The preset prediction time domain is discretized into equally spaced time steps Δt (e.g., Δt = 0.5 seconds). Within each time step, multiple candidate trajectories are generated using dynamic programming, starting from the current position of the equipment and ending at the entrances of each lane or critical path node of the automated warehouse.

[0053] In this embodiment, the dynamic programming method used in step S2 to pre-plan the spatiotemporal motion trajectory specifically includes the following state definitions and recursive logic:

[0054] State definition: The time step number within the preset prediction time domain is denoted as stage k, and the state variable is defined as S. k =(x k y k , z k v k ), where (x k y k , z k Let v be the spatial coordinates of the device at the end of the k-th time step. k This is the instantaneous velocity scalar at this moment.

[0055] Decision variables: Define decision variable u k Its value range is the set {acceleration, constant speed, deceleration, left turn, right turn, straight ahead}.

[0056] State transition equation: Based on the operating speed and maximum acceleration / deceleration constraints in the set of equipment state parameters, a kinematic equation is established, expressed as (x k+1 y k+1 , z k+1 v k+1 )=f(S k u k ).

[0057] Cost function: When calculating the weighted sum of travel time and energy consumption for each candidate trajectory, the following specific calculation method is used:

[0058] Cost(Traj) = [α×Δt+β×(m×a k ×v k ×Δt)]

[0059] Where α and β are preset weighting coefficients, α+β=1, α and β are between 0 and 1, Δt is the time step, m is the sum of the equipment's self-weight and rated load, and a k To determine based on adjacent time v k The absolute value of acceleration calculated from the change.

[0060] When multiple inbound / outbound devices exist in the automated storage and retrieval system (AS / RS), the devices are prioritized and pre-planned sequentially. The prioritization rule is: devices with shorter remaining execution time for the current task have higher priority; if they are the same, they are ordered by device ID in ascending order. During the sequential planning process, the planned device trajectories are locked and stored in the trajectory set. When planning the next device, a collision constraint check is added to the state transition of the dynamic programming.

[0061] For each candidate state (x, y, z, t), if there exists a planned device trajectory point (x... locked y locked , z locked , t locked )satisfy:

[0062] (1) Time difference condition: |t–t locked | <T safe ;

[0063] (2) Spatial distance condition: Euclidean distance √[(xx locked ) 2 +(yy locked ) 2 +(zz locked ) 2 ] <D safe ,

[0064] The state transition cost is then set to infinity (INF), thus eliminating the candidate trajectory during the dynamic programming backtracking process.

[0065] Among them, D safe Assuming a safe distance for the equipment, we take twice the safe radius. In this embodiment, the safe radius is set to 0.25m, therefore D... safe =0.5m; T safe =D safe / V max +0.5s (safety time margin), V max This is the maximum speed of the equipment.

[0066] This sequential planning method avoids circular dependencies, ensuring that the pre-planned trajectory of each device can be calculated independently and is globally collision-free.

[0067] For each candidate trajectory Traj k Calculate its cost function Cost(Traj) k )=α×T travel +β×E consumption T travel E represents the travel time (in seconds). consumption To estimate energy consumption (calculated based on acceleration and velocity curve integrals, in joules), α and β are preset weighting coefficients. The candidate trajectory with the smallest weighted sum is selected as the pre-planned trajectory for the device.

[0068] In multi-device collaborative scenarios, to prevent collisions in shared lanes, this step introduces an anti-collision time interval constraint during candidate trajectory generation. Specifically, the constraint is: for any two devices i and j, the timestamp t of their trajectory points passing through the same spatial coordinate point (x, y, z) must be consistent. i and t j The condition |t must be satisfied i -t j |>D safe / Vel avg D safe For equipment safety distance (meters), Vel avg The average passing speed (m / s).

[0069] The generated pre-planned equipment trajectories are in the form of a time-ordered sequence: Traj plan ={(t1,x1,y1,z1), (t2,x2,y2,z2),…}.

[0070] S3: Based on the intersection and union of the pre-planned trajectories of all equipment in the spatiotemporal coordinate system, calculate the set of passable time windows for each spatiotemporal grid unit, and then map the set of passable time windows back onto the location of each storage location in the automated warehouse to generate the set of allocable time windows for each storage location. Specifically:

[0071] Rasterization parameters: The physical space of the automated storage and retrieval system (AS / RS) is divided into uniform spatiotemporal grid units with a spatial resolution of Δs = 0.2m × 0.2m × 0.2m (not greater than 1 / 10 of the equipment safety distance) and a temporal resolution of Δt = 0.1 seconds (consistent with the control cycle). The origin of the spatiotemporal coordinate system is taken as the ground point at the southwest corner of the AS / RS.

[0072] Occupancy marker: Map the pre-planned trajectory point sequence of each device to a grid cell: If the timestamp of a trajectory point falls within the interval [k·Δt, (k+1)·Δt) and the spatial coordinates fall within a certain grid cell, then mark that the spatiotemporal grid cell is occupied within that time slice.

[0073] Passable time window calculation: For spatiotemporal grid cells that are not occupied by any device trajectories, their continuous idle time intervals are merged to obtain the passable time window set W. passable ={[t start , t end ],…}.

[0074] Reverse mapping to storage location: For each storage location j, take the coordinates P of its front-end pickup point. pick，j Traverse all spatiotemporal grid cells. If the spatial center point of a certain grid cell is parallel to P... pick，j If the Euclidean distance is less than 0.5m, then the set of passable time windows for that grid cell is taken as the initial set of allocable time windows for that storage location. Then, the time windows are shrunk according to the travel time of the equipment from the grid cell to the storage location pickup point (calculated according to the Manhattan distance formula in Example 1) to obtain the final set of allocable time windows for each storage location.

[0075] In step S3, the detour time required for the device to move from the trajectory point to the pickup point of the storage location is calculated, which is specifically achieved through the following mathematical model:

[0076] Let P be a point on the pre-planned trajectory of the equipment. traj (x t y t , z t The pickup point for the target location is P. pick (x p y p , z p ).

[0077] Step 1: Calculate the Manhattan distance. Considering the operating characteristics of the automated warehouse tunnel (equipment moves independently along the X / Y / Z axes), the effective travel distance D... eff The calculation formula is:

[0078] D eff =|x t -x p |+|y t -y p |+|z t -z p |

[0079] Step 2: Calculate the detour time. Based on the operating speed from the current equipment status parameter set and the preset acceleration / deceleration characteristic curve, the detour time T is calculated. detour The calculation formula is:

[0080] When D eff Greater than the minimum start distance D min At that time, T detour =(D eff / Vel max )+Tacc，dec ;

[0081] When the trajectory point P traj Located exactly at pickup point P pick When working on the front end, T detour =0.

[0082] Among them, Vel max T is the maximum operating speed of the equipment. acc，dec This is a preset fixed start / stop time compensation constant.

[0083] Step 3: Mark the time window. Only if condition T is met. detour ≤T buffer At that time, the trajectory point P traj The associated timestamp range is extended backward by T. detour Generate an allocatable time window for that storage location.

[0084] It is important to know that the remaining task buffer time T of the device is... buffer =max(0,T) task ,remain-T base ), where T base 'remain' represents the remaining execution time of the current task collected in step S1, T. base The baseline time required for the equipment to complete currently assigned but not yet executed inbound / outbound operations (calculated by the sum of the travel times of unexecuted trajectory segments in the equipment's current pre-planned trajectory). For idle equipment with no current tasks, T... task , remain=0, at this time T buffer Take the default value T idle，buffer In this embodiment, T idle，buffer =10 seconds. This buffer time represents the acceptable detour pickup time margin for the equipment without delaying existing tasks.

[0085] S4: Match the cargo attribute vector with the current storage location state tensor. Under the constraint of the allocable time window set, select a set of candidate storage locations for each cargo to be put into storage that meet the cargo attribute matching and have a non-empty time window. Use the minimum spatial deviation between the storage location center coordinates in the candidate storage location set and the pre-planned trajectory of the corresponding equipment as the objective function to determine the target storage location and the corresponding target equipment.

[0086] refer to Figure 2 This step involves optimally matching cargo attributes with storage space resources, and consists of two sub-steps: candidate set filtering and optimal matching.

[0087] Sub-step 1: Filtering the candidate storage location set.

[0088] Three features—cargo volume v, cargo weight w, and cargo turnover frequency f—are extracted from the cargo attribute vector. The attributes of each available cargo location in the current location state tensor are used as constraint boundaries.

[0089] Volume constraint: Cargo volume v is less than or equal to storage location volume V slot ;

[0090] Weight constraint: The weight of the goods w is less than or equal to the maximum load capacity W of the storage location. slot ;

[0091] Turnover rate matching: Prioritize allocating goods with high f values ​​to locations near the aisle exit (i.e., locations with less spatial deviation).

[0092] After removing locations that do not meet any of the constraints, a preliminary location set is obtained. Then, it is checked whether the allocatable time window set for each location in this set contains items beyond the goods receipt deadline. in The available windows retain qualified storage locations to form a candidate storage location set.

[0093] The deadline for goods to be received into the warehouse in Defined as: a timestamp (unit: seconds, relative to system midnight) parsed from the "Latest Inbound Completion Time" field in the inbound task order issued by the warehouse operation scheduling system. If not explicitly given in the task order, the default value: Deadline is used. in = Current system time + 3600 seconds (i.e., 1 hour). This deadline is used to constrain the timeliness of warehouse location allocation, ensuring that goods are received into the warehouse within the required time.

[0094] For each location in the initial location set, calculate its turnover rate matching score. f =f cargo ×(1 / (1+d exit )), where f cargo d represents the turnover frequency (times / day) of goods awaiting warehousing. exit This is the Manhattan distance (in meters) from the pickup point of the storage location to the nearest aisle exit. This score, after normalization, is used as an additional item in sub-step two of subsequent step S4, along with the spatial deviation. ij Perform a weighted summation to form the overall deviation Cost. ij =Deviation ij -λ·Score f λ is a preset weighting coefficient (λ ranges from 0 to 1, and in this embodiment, λ = 0.3). The Hungarian algorithm is used to match the overall deviation matrix to minimize the total overall deviation.

[0095] Sub-step 2: Determining the objective based on minimizing spatial deviation.

[0096] Define spatial deviation ij Let be the shortest Euclidean distance from the pickup point coordinates of storage location j to all trajectory points on the current pre-planned trajectory of device i. The calculation formula is as follows:

[0097] For each point k in the trajectory point sequence of device i, calculate the coordinates (x, y, k) of that point. k y k , z k ) and the coordinates of the pickup point (x) of location j j y j , z j The square root of the distance √[(x) k -x j ) 2 +(y k -y j ) 2 +(z k -z j ) 2 Then, take the minimum value among all distances as the Deviation. ij .

[0098] Construct a deviation matrix M (rows are devices, columns are storage locations) between all available devices and candidate storage locations. Since the device and storage location are assigned in a one-to-one manner, the Hungarian algorithm is used to solve the bipartite graph. The objective function is to minimize the sum of the total deviations, and the optimal (target device, target storage location) pairing result is output.

[0099] It should be noted that when constructing the deviation matrix and solving it using the Hungarian algorithm, there may be a discrepancy between the number of inbound / outbound devices and the number of elements in the candidate location set. To accommodate the input requirements of the Hungarian algorithm for the square matrix, this step further includes the following preprocessing logic:

[0100] Scenario 1: The number of devices M is greater than the number of candidate storage locations N. In this case, instead of directly adding a virtual storage location column to the deviation matrix, the following preprocessing logic is executed:

[0101] (1) Sort all M devices in ascending order of their remaining execution time for their current tasks;

[0102] (2) Select the top N devices after sorting to construct an N×N deviation matrix, and input it into the Hungarian algorithm to solve;

[0103] (3) The remaining (MN) equipment will not participate in this allocation of storage space, and will maintain their original pre-planned equipment trajectory, or will be put into idle standby status.

[0104] By employing the above logic, algorithmic deviations caused by the introduction of virtual storage locations and improper setting of penalty constant values ​​are fundamentally avoided, ensuring the dimensional adaptability of matrix operations.

[0105] Scenario 2: The number of candidate storage locations N is greater than the number of devices M. In this case, (NM) virtual devices are added, and the spatial deviation of the virtual devices is set to 0. In the algorithm output, the candidate storage locations matched to the virtual devices will enter the waiting pool, waiting for the next round of trajectory pre-planning cycle for allocation.

[0106] S5: Based on the target storage location and the target equipment, insert a dwell trajectory point on the pre-planned trajectory of the target equipment. The dwell time of the dwell trajectory point is equal to the operation time required to transport the goods from the inlet to the target storage location. Update the pre-planned trajectory of the equipment to obtain the updated equipment execution trajectory. Control the equipment to perform the inlet operation according to the equipment execution trajectory.

[0107] Based on the pairing results determined in step S4, the operation action is inserted into the pre-planned trajectory of the target equipment.

[0108] First, obtain the trajectory point with the smallest deviation from the target storage space in the pre-planned trajectory of the target equipment as the insertion anchor point P. anchor .

[0109] Then, the operation time T required to transport the goods from the inbound port to the target storage location is calculated. operation Homework duration T operation Calculate using the following formula:

[0110] T operation =T fork +T lift +T travel，detour ,

[0111] Among them, T fork =2×(L fork / v fork )+t fork，acc L fork Fork travel (equal to pallet depth, in meters), v fork The rated extension / retraction speed of the forks (m / s), t fork，acc This is a fixed time compensation for the start and stop of the forks (0.5 seconds in this embodiment);

[0112] T lift =|z target -z current | / v lift +t lift，acc , where z target Let z be the coordinate of the target storage level height. current v represents the current height of the loading platform of the equipment.lift To increase the rated speed of the mechanism, t lift，acc For start-stop compensation (0.3 seconds in this embodiment);

[0113] T travel，detour =D detour / v max D detour v is the total Manhattan distance required for the equipment to move from the insertion anchor point on the current pre-planned trajectory to the target pickup point and back to the anchor point. max This is the maximum speed at which the equipment operates horizontally.

[0114] Of the parameters mentioned above, L fork v fork v lift v max These are all inherent parameters of the equipment.

[0115] The specific trajectory update operation is as follows:

[0116] Insert anchor point P anchor The timestamps t of all subsequent trajectory points k Uniformly offset backward T operation .

[0117] In P anchor Then, insert a stopping trajectory point P at the immediately adjacent position. stay Its timestamp is t anchor With a tiny increment, the spatial coordinates are set to the coordinates of the pickup point (x) of the target location. target y target , z target ), and add a stay duration attribute T. operation .

[0118] After insertion, the spatial coordinates of subsequent trajectory points are connected to the original path logic to form an updated device execution trajectory. The controller then issues servo commands in real time according to this updated trajectory.

[0119] S6: During the process of the equipment performing the warehousing operation, the trajectory deviation between the actual trajectory point of the equipment and the pre-planned trajectory of the equipment is collected in real time. When the trajectory deviation exceeds the preset deviation threshold, the current actual equipment status parameter set is used as the initial state, and the process jumps to step S2 to re-plan the trajectory and allocate the storage location.

[0120] refer to Figure 3 During each control cycle of the equipment's warehousing operation, the system continuously monitors the trajectory tracking accuracy.

[0121] The data acquisition rules are as follows: Obtain the actual location coordinates P of the device. actualAnd the expected position coordinate P corresponding to the current timestamp in the device execution trajectory generated in step S5 expect Calculate the single-step deviation e vec =P actual -P expect (Three-dimensional vector difference).

[0122] Within a preset time window W (window length N control cycles, e.g., N=10), the actual position P is sampled in each control cycle. actual (n) and desired position P expect The sum of the three-dimensional Euclidean distances between (n). Let the single-step distance deviation of the nth control cycle be d(n) = ||P actual (n)-P expect (n)||, where ||·|| represents the Euclidean distance in three-dimensional space (i.e., the square root of the sum of squares), then the cumulative trajectory deviation E acc The calculation formula is:

[0123] E acc =d(1)+d(2)+…+d(N),

[0124] At the same time, the maximum single-step distance deviation E is defined. max It is the maximum value among the above N single-step distance deviations, that is:

[0125] E max =max{d(1),d(2),…,d(N)},

[0126] Set the cumulative deviation threshold δ acc (e.g., δ) acc =0.8 meters) and the maximum single-step deviation threshold δ max (e.g., δ) max =0.2 meters). When E acc >δ acc or E max >δ max At that time, the replanning flag is triggered.

[0127] Cumulative deviation threshold δ acc =min(0.1×L min，aisle 0.05×V max ×T control ×N), where L min，aisle The minimum clear width of the automated warehouse tunnel, in meters, is the safe passage width obtained by subtracting the width of the equipment body from the actual tunnel width; a typical value is 0.3m to 0.5m. max The maximum operating speed of the equipment (in m / s) is obtained from the parameters on the equipment nameplate; T controlThe position sampling period of the equipment control system (in seconds) is usually 0.02s to 0.1s, with a default value of 0.1s; N is the number of control cycles within the time window, and in this embodiment, N=10.

[0128] Maximum single-step deviation threshold δ max =0.5×δ acc / N, and not less than the equipment positioning repeatability (usually taken as 0.01 meters).

[0129] In practical applications, the threshold can be fine-tuned through offline simulation or on-site trial operation, with the adjustment range not exceeding ±20% of the calculated value.

[0130] After replanning is triggered, the position coordinates in the current actual device state parameter set are replaced with the actual position P at the current moment. actual The remaining execution time of the current task is replaced with the remaining time re-estimated based on the remaining distance traveled on the actual trajectory. The remaining distance is calculated as follows: from the current actual position P... actual Starting from the beginning, along the sequence of subsequent trajectory points in the original pre-planned trajectory that have not yet been executed, calculate and sum the Euclidean distances between adjacent points. Divide the remaining distance by the current average operating speed of the equipment to obtain the updated remaining execution time of the current task. Using this updated parameter set as the initial state, proceed to step S2.

[0131] Example 2

[0132] This embodiment provides a system for optimizing the allocation of storage locations in an automated warehouse, such as... Figure 4 As shown, the system adopts a modular design and is deployed on an edge computing server or cloud platform, specifically including the following functional modules:

[0133] Data acquisition module: It is equipped with an OPCUA client and a database connector, and is used to perform the data acquisition function described in step S1 of embodiment 1.

[0134] Trajectory pre-planning module: It has an embedded dynamic programming solver and collision detection algorithm library, and is used to perform the trajectory generation function described in step S2 of Example 1.

[0135] Time window reverse mapping module: contains a spatiotemporal data index structure, used to perform the calculation and mapping functions described in step S3 of embodiment 1.

[0136] Location screening and matching module: integrates a Hungarian algorithm solver to perform the constraint screening and optimization matching function described in step S4 of Example 1.

[0137] Track update and execution module: Communicates in real time with the underlying PLC or motion controller to execute the track insertion and distribution function described in step S5 of embodiment 1.

[0138] Deviation monitoring and replanning module: Equipped with a real-time state estimator, used to perform the deviation monitoring and closed-loop feedback function described in step S6 of Example 1.

[0139] Example 3

[0140] This embodiment provides an electronic device, including a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it specifically reads instructions from the memory to implement steps S1 to S6 of the automated warehouse location allocation optimization method described in Embodiment 1. This electronic device can be an industrial control computer, a server, or an embedded control unit.

[0141] Example 4

[0142] This embodiment provides a computer-readable storage medium storing computer program instructions. When these computer program instructions are executed by a processor, they trigger the processor to run the algorithm logic for optimizing the allocation of automated warehouse locations as described in Embodiment 1, including tensor calculation, trajectory search, time window calculation, and replanning logic.

[0143] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing the allocation of storage locations in an automated warehouse, characterized in that, Includes the following steps: S1: Obtain the current location status tensor of the automated warehouse, the cargo attribute vector of the goods to be put into the warehouse, and the current equipment status parameter set of the inbound and outbound equipment. The current location status tensor includes the occupancy flag and remaining occupancy time of each location. The current equipment status parameter set includes the position coordinates, running speed, and remaining execution time of the current task of each equipment. S2: Using the current equipment status parameter set as input, pre-plan the spatiotemporal motion trajectory of each inbound / outbound equipment in the preset prediction time domain, and generate the equipment pre-planned trajectory for each equipment. The equipment pre-planned trajectory includes a sequence of trajectory points, and each trajectory point is associated with a timestamp and spatial coordinates. S3: Based on the intersection and union of the pre-planned trajectories of all equipment in the spatiotemporal coordinate system, calculate the set of passable time windows for each spatiotemporal grid unit, and reverse-map the set of passable time windows to each location of the automated warehouse to generate the set of allocable time windows for each location. S4: Match the cargo attribute vector with the current storage location state tensor. Under the constraint of the allocable time window set, select a set of candidate storage locations for each cargo to be put into storage that meet the cargo attribute matching and have a non-empty time window. Use the minimum spatial deviation between the storage location center coordinates in the candidate storage location set and the pre-planned trajectory of the corresponding equipment as the objective function to determine the target storage location and the corresponding target equipment. S5: Based on the target storage location and the target equipment, insert a dwell trajectory point on the pre-planned trajectory of the target equipment. The dwell time of the dwell trajectory point is equal to the operation time required to transport the goods from the inlet to the target storage location. Update the pre-planned trajectory of the equipment to obtain the updated equipment execution trajectory. Control the equipment to perform the inlet operation according to the equipment execution trajectory. S6: During the process of the equipment performing the warehousing operation, the trajectory deviation between the actual trajectory point of the equipment and the pre-planned trajectory of the equipment is collected in real time. When the trajectory deviation exceeds the preset deviation threshold, the current actual equipment status parameter set is used as the initial state, and the process jumps to step S2 to re-plan the trajectory and allocate the storage location.

2. The method for optimizing the allocation of storage locations in an automated warehouse according to claim 1, characterized in that, The pre-planning of the spatiotemporal motion trajectory in S2 further includes: discretizing the preset prediction time domain into equally spaced time steps; within each time step, using the current position of the equipment as the starting point and the entrances of each lane of the automated warehouse as the ending point, generating multiple candidate trajectories using a dynamic programming method; calculating the weighted sum of the travel time and energy consumption of each candidate trajectory; and selecting the candidate trajectory with the smallest weighted sum as the pre-planned trajectory of the equipment; when multiple devices share the same lane, adding anti-collision time interval constraints during the candidate trajectory generation process, so that the time interval between the trajectory points of different devices on the same spatiotemporal coordinate is greater than the passage time corresponding to the safe distance of the equipment.

3. The method for optimizing the allocation of storage locations in an automated warehouse according to claim 1, characterized in that, The process of generating an allocable time window set for each storage location in S3 includes: mapping each storage location to its corresponding shelf column, layer, and depth; calculating the detour time required for the equipment to move from the trajectory point to the storage location's retrieval point based on the spatial distance between the trajectory point in the equipment's pre-planned trajectory and the retrieval point at the front of the storage location; marking the timestamp intervals corresponding to trajectory points whose detour time is less than the equipment's remaining task buffer time as the allocable time windows for that storage location; and merging the allocable time windows corresponding to each equipment in chronological order when multiple equipment pre-planned trajectories can reach the same storage location, and using the merged time windows as the allocable time window set for that storage location.

4. The method for optimizing the allocation of storage locations in an automated warehouse according to claim 1, characterized in that, The process of selecting candidate storage locations in S4 further includes: extracting three dimensions of features from the cargo attribute vector: cargo volume, cargo weight, and cargo turnover frequency; using the volume capacity, load-bearing limit, and aisle location of each available storage location in the current storage location state tensor as the constraint boundaries of the corresponding dimensions; and removing storage locations that do not meet any of the dimension constraints to obtain a preliminary storage location set. For each storage location in the preliminary storage location set, it is checked whether its allocable time window set contains an available time window that is longer than the cargo entry deadline. If it does, the storage location is retained in the candidate storage location set.

5. The method for optimizing the allocation of storage locations in an automated warehouse according to claim 1, characterized in that, The determination of target storage location and target equipment in step S4 further includes: for each storage location in the candidate storage location set, calculating the shortest spatial deviation from the storage location to the current pre-planned trajectory of each piece of equipment. The spatial deviation is calculated by: traversing each trajectory point on the pre-planned trajectory of the equipment, calculating the Euclidean distance between the coordinates of the trajectory point and the coordinates of the storage location's pickup point, and taking the minimum Euclidean distance among all trajectory points as the spatial deviation of the equipment from the storage location; forming a deviation matrix from the spatial deviations of each equipment to each storage location, and using the Hungarian algorithm to perform bipartite graph matching on the deviation matrix to obtain the storage location-equipment pair that minimizes the total deviation as the target storage location and target equipment.

6. The method for optimizing the allocation of storage locations in an automated warehouse according to claim 1, characterized in that, The insertion of a dwell trajectory point in S5 includes: obtaining the trajectory point with the smallest deviation from the target storage location space in the current pre-planned trajectory of the target equipment as the insertion anchor point; shifting the timestamps of all trajectory points after the insertion anchor point backward by the operation duration; inserting a dwell trajectory point at the position immediately adjacent to the insertion anchor point; setting the spatial coordinates of the dwell trajectory point to the picking point coordinates of the target storage location; and setting the dwell duration attribute of the dwell trajectory point to the operation duration; after the insertion is completed, restoring the spatial coordinates of the next trajectory point after the dwell trajectory point to the coordinates of the first trajectory point after the original insertion anchor point, thus forming the updated equipment execution trajectory.

7. The method for optimizing the allocation of storage locations in an automated warehouse according to claim 1, characterized in that, The real-time acquisition of trajectory deviation in S6 further includes: within each control cycle of the device, acquiring the actual position coordinates of the device and the expected position coordinates corresponding to the current timestamp in the pre-planned trajectory of the device, calculating the vector difference between the actual position coordinates and the expected position coordinates as the single-step deviation; taking the vector and magnitude of all single-step deviations within a preset time window as the cumulative trajectory deviation; when the cumulative trajectory deviation exceeds a preset deviation threshold, triggering a replanning flag, and replacing the position coordinates in the current actual device status parameter set with the actual position coordinates, and replacing the remaining execution time of the current task with the remaining time re-estimated based on the actual trajectory.

8. A system for optimizing the allocation of storage locations in an automated warehouse, characterized in that, Includes the following modules: The data acquisition module acquires the current location status tensor of the automated warehouse, the cargo attribute vector of the goods to be received, and the current equipment status parameter set of the inbound and outbound equipment. The trajectory pre-planning module takes the current equipment status parameter set as input and pre-plans the spatiotemporal motion trajectory of each inbound / outbound equipment in the preset prediction time domain, generating the equipment pre-planned trajectory for each equipment. The time window reverse mapping module calculates the passable time window set for each spatiotemporal grid unit based on the pre-planned trajectory of all equipment, and reverse maps the passable time window set to each cargo location to generate an allocable time window set for each cargo location. The cargo location screening and matching module matches the cargo attribute vector with the current cargo location state tensor, filters the candidate cargo location set under the constraint of the allocable time window set, and determines the target cargo location and the corresponding target equipment with the goal of minimizing spatial deviation. The trajectory update and execution module inserts a stop trajectory point on the pre-planned trajectory of the target device to obtain the updated device execution trajectory, and controls the device to perform the warehousing operation according to the device execution trajectory; The deviation monitoring and replanning module collects the trajectory deviation between the actual trajectory points of the device and the pre-planned trajectory of the device in real time. When the cumulative trajectory deviation exceeds the preset deviation threshold, the trajectory preplanning unit is triggered to restart execution with the current actual device state parameter set.

9. An electronic device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for optimizing the allocation of storage locations in an automated warehouse according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for optimizing the allocation of storage locations in an automated warehouse according to any one of claims 1 to 7.

Images