Warehouse container automatic operation method and system based on three-dimensional mechanism

Abstract

The invention relates to the technical field of warehouse logistics management, and discloses a warehouse container automatic operation method and system based on a three-dimensional mechanism. The method comprises the following steps of: counting characteristic parameters of an operation scene, calculating a target weight when the characteristic parameters continuously exceed a threshold value, obtaining a weight on that day according to five-day linear interpolation smooth transition, scoring candidate storage locations according to the weight on that day, selecting an optimal storage location, and selecting an optimal crane by integrating optimal scheduling of a moving distance, operation time and a queue length. According to the invention, the adaptive capability of the storage location allocation decision to the scene change, the smoothness and strategy continuity of the weight adjustment process, and the weight configuration convergence and operation efficiency optimization effect based on feedback correction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of warehouse logistics management, and particularly relates to a warehouse container automated operation method and system based on a three-dimensional mechanism. BACKGROUND

[0002] The warehouse container automated operation management method based on a three-dimensional mechanism in the prior art generally includes vehicle arrival identification, storage location allocation, crane scheduling, and container hoisting, etc. The storage location allocation module selects an idle storage location from candidate storage locations according to container attributes and inventory states for allocation. The crane scheduling module issues operation instructions to available cranes according to a task queue. The system obtains basic information such as container types, weights, and sizes by querying a database and performs matching calculation in combination with parameters such as storage location bearing capacities and spatial positions. When a lower container needs to be accessed, the system automatically generates a destacking task to first transfer an upper container to a temporary storage location. After the operation is completed, the system updates container positions and storage location occupancy state records in the inventory database. This kind of method realizes information management of warehouse operation processes to a certain extent.

[0003] However, the prior art has significant deficiencies. The storage location allocation strategy adopts a fixed evaluation weight configuration, that is, the weight coefficients of evaluation factors such as out-of-warehouse frequency, weight distribution, and stacking stability remain constant values during system operation. This static weight configuration cannot adapt to dynamic changes in warehouse operation scenarios. When the warehouse area switches from a low-frequency out-of-warehouse scenario in the off-season to a high-frequency out-of-warehouse scenario in the peak season, the fixed weight still allocates storage locations according to the off-season mode, resulting in that high-frequency out-of-warehouse containers are stored in positions far from the out-of-warehouse passageway, increasing the subsequent carrying distance. When the weight distribution of containers changes from a balanced state to a discrete state, the fixed weight fails to timely improve the priority of the weight balance factor, causing excessive load in local areas and safety hazards. The destacking decision mechanism only considers the immediate transfer path of the upper container and does not comprehensively evaluate the future out-of-warehouse plan of the container and the availability of the temporary storage location. The crane scheduling algorithm performs serial allocation according to the order of tasks without considering the differences in current positions and operation loads of cranes. The system lacks a feedback evaluation and parameter optimization mechanism based on actual operation effects, resulting in that the storage location allocation and scheduling strategy cannot be continuously improved. SUMMARY

[0004] The present application provides a warehouse container automated operation method and system based on a three-dimensional mechanism, which solves the problems in the prior art that the weight configuration of storage location evaluation cannot adapt to dynamic changes in operation scenarios, the weight adjustment process lacks a smooth transition mechanism, resulting in unstable allocation strategies, and there is a lack of weight iterative optimization mechanism based on actual operation effect feedback, improves the self-adaptability of storage location allocation decisions to scenario changes, the smoothness of the weight adjustment process, the strategy continuity, and the convergence of the weight configuration and operation efficiency optimization effect based on feedback correction.

[0005] In a first aspect, this application provides a method for automated operation of warehouse containers based on a three-dimensional mechanism, the method comprising: Step S1: Receive the job task data packet and container information, and calculate the inbound / outbound ratio, weight dispersion and stacking pressure index of the past 7 days as scene feature parameters; Step S2: When any parameter in the scene feature parameters exceeds the preset threshold for 3 consecutive days, calculate the target weight vector. Decompose the difference between the current weight vector and the target weight vector into daily adjustment increments using 5-day linear interpolation. Accumulate the daily adjustment increments to the current weight vector to obtain the weight vector for the day. Step S3: Predict the outbound probability based on the container type, query the weight deviation, load-bearing capacity, and distance to the outbound channel of the candidate storage location, and obtain the suitability score by weighted summation of the outbound probability, weight deviation, load-bearing capacity, and distance with the weight vector of the day, and select the storage location with the highest suitability score as the target storage location. Step S4: Calculate the three-dimensional moving distance and operation time of the crane from the container source location and the target storage location. Based on the moving distance, operation time and queue length, optimize the scheduling and select the crane with the highest scheduling score to generate a scheduling instruction.

[0006] Secondly, this application provides an automated warehouse container operation system based on a three-dimensional mechanism, the automated warehouse container operation system based on a three-dimensional mechanism comprising: The receiving module is used to receive job task data packets and container information, and to collect data on the inbound and outbound ratios, weight dispersion, and stacking pressure index over the past 7 days as scene feature parameters. The decomposition module is used to calculate the target weight vector when any parameter in the scene feature parameters exceeds a preset threshold for 3 consecutive days. The difference between the current weight vector and the target weight vector is decomposed into daily adjustment increments by 5-day linear interpolation. The daily adjustment increments are accumulated to the current weight vector to obtain the weight vector for the day. The weighting module is used to predict the outbound probability based on the container type, query the weight deviation, load-bearing capacity, and distance to the outbound channel of the candidate storage location, and sum the outbound probability, weight deviation, load-bearing capacity, and distance with the weight vector of the day to obtain the suitability score, and select the storage location with the highest suitability score as the target storage location. The scheduling module is used to calculate the three-dimensional moving distance and operation time of the crane from the container source location and the target storage location, optimize the scheduling based on the moving distance, operation time and queue length, and select the crane with the highest scheduling score to generate a scheduling instruction.

[0007] The technical solution provided in this application establishes a scene recognition mechanism based on historical operation data by statistically analyzing the inbound / outbound ratio, weight dispersion, and stacking pressure index over the past 7 days as scene feature parameters. This mechanism can quantitatively reflect the dynamic change trend of warehouse operation scenarios. The inbound / outbound ratio directly characterizes the intensity of outbound operations in the warehouse area through the ratio of the total number of outbound tasks to the total number of inbound tasks. The weight dispersion reveals the dispersion of container weight distribution through the ratio of the standard deviation of container weight to the average weight. The stacking pressure index quantifies the impact of warehouse stacking layout on operation efficiency through the ratio of the number of stacking operations to the total number of outbound tasks.

[0008] Three scenario feature parameters comprehensively characterize the operational scenario status from different dimensions. When any of the scenario feature parameters exceeds a preset threshold for three consecutive days, a weight adjustment process is triggered. This continuous judgment mechanism effectively avoids false triggers caused by daily data fluctuations. The difference between the current weight vector and the target weight vector is decomposed into daily adjustment increments using 5-day linear interpolation and accumulated to the current weight vector daily. This smooth transition mechanism distributes the weight adjustment evenly over a 5-day adjustment period, adjusting only one-fifth of the total difference each day. This allows the weight configuration of the storage location assessment model to gradually transition from the current state to the target state rather than abruptly changing. Containers of the same type that enter the storage at adjacent times are assigned to storage locations with similar spatial positions due to the continuity of weight configuration. The smooth weight adjustment process maintains the storage location allocation. The stability and predictability of the allocation strategy are assessed by predicting the outbound probability based on container type and querying the weight deviation, load-bearing capacity, and distance to the outbound channel of candidate storage locations. These evaluation indicators are then weighted and summed with the daily weight vector to obtain a fit score. This multi-dimensional evaluation mechanism achieves adaptive matching of the storage location evaluation strategy to the characteristics of the current operating scenario through dynamically adjusted weight vectors. When the inbound / outbound ratio is high, the weight of outbound frequency increases, allowing containers with high outbound probability to be preferentially allocated to convenient storage locations near the outbound channel. When the weight dispersion is large, the weight of weight balance increases, guiding heavy containers to be stored in a dispersed manner to avoid excessive load on local areas. When the stacking pressure index is high, the weight of stacking stability increases, prompting containers to be preferentially allocated to storage locations with sufficient load-bearing capacity and less overhead obstruction. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of one embodiment of the warehouse container automation operation method based on a three-dimensional mechanism in this application. Figure 2This is a schematic diagram illustrating the convergence process of the weight feedback correction algorithm in the embodiments of this application; Figure 3 This is a schematic diagram showing the distribution of each score item for the candidate storage location in an embodiment of this application. Detailed Implementation

[0011] This application provides an automated operation method and system for warehouse containers based on a three-dimensional mechanism. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0012] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the warehouse container automation operation method based on a three-dimensional mechanism in this application includes: Step S1: Receive the job task data packet and container information, and calculate the inbound / outbound ratio, weight dispersion and stacking pressure index of the past 7 days as scene feature parameters; Specifically, the scene feature parameters are collected based on statistical analysis of historical operation data. The inbound / outbound ratio is obtained by extracting the total number of task records marked as outbound in the database over the past 7 days and dividing it by the sum of the total number of outbound and inbound task records. The weight dispersion is obtained by calculating the arithmetic mean and standard deviation of the weight field of all inbound containers and then dividing the result to obtain a quantitative indicator reflecting the degree of weight distribution dispersion. The stacking pressure index reflects the frequency of stacking operations by dividing the number of operation records marked as stacking operations by the total number of outbound tasks. The scene recognition and judgment mechanism sets up three independent threshold detection channels to monitor whether the inbound / outbound ratio exceeds 0.6 for 3 consecutive days, whether the weight dispersion exceeds 0.3 for 3 consecutive days, and whether the stacking pressure index exceeds 0.2 for 3 consecutive days. Triggering any channel will set the scene change flag.

[0013] Step S2: When any parameter in the scene feature parameters exceeds the preset threshold for 3 consecutive days, calculate the target weight vector. Decompose the difference between the current weight vector and the target weight vector into daily adjustment increments using 5-day linear interpolation. Accumulate the daily adjustment increments to the current weight vector to obtain the weight vector for the day. Specifically, the adaptive weight calculation uses a gain superposition method. When the inbound / outbound ratio exceeds 0.6, the difference between this ratio and 0.6 is extracted, multiplied by 0.5 to obtain the outbound frequency gain coefficient, and added to the baseline weight of 0.35 to form the target outbound frequency weight. The weight dispersion and stacking pressure index are calculated using the same gain calculation logic to generate corresponding target weight components. The weight smooth transition mechanism subtracts the target weight vector from the current weight vector component by component to obtain four weight differences. Each difference is multiplied by a fixed coefficient of 0.2 to obtain the daily adjustment increment. Through a daily accumulation mechanism, that is, the current weight plus the adjustment increment multiplied by the number of days adjusted, the weight changes gradually rather than abruptly.

[0014] Step S3: Predict the outbound probability based on the container type, query the weight deviation, load-bearing capacity, and distance to the outbound channel of the candidate storage location, and obtain the suitability score by weighted summation of the outbound probability, weight deviation, load-bearing capacity, and distance with the weight vector of the day, and select the storage location with the highest suitability score as the target storage location. Specifically, the storage location suitability assessment converts four different evaluation indicators into weighted scoring items with a unified dimension, and then aggregates them. The outbound probability is derived from the normalized ratio of the number of outbounds of this container type in the most recent 30 days to the total number of outbounds in the most recent 90 days in the historical database. The weight deviation is quantified by subtracting the average weight of all areas of the entire storage facility from the total weight of the containers already stored in the candidate storage location area, taking the absolute value, and then dividing by the average weight. The load-bearing margin is directly read from the candidate storage location attribute table, which is the difference between the designed load-bearing limit field and the currently carried weight field. The distance to the outbound channel is calculated as the straight-line distance in space using the Euclidean norm of the difference between the three-dimensional coordinates of the candidate storage location and the coordinates of the nearest outbound channel. The four original indicators are multiplied by the corresponding weighted components in the daily weight vector and then summed to form a comprehensive score value. The larger the value, the more suitable the storage location is for the current container storage.

[0015] Step S4: Calculate the three-dimensional moving distance and operation time of the crane from the container source location and the target storage location. Based on the moving distance, operation time and queue length, optimize the scheduling and select the crane with the highest scheduling score to generate a scheduling instruction.

[0016] Specifically, crane scheduling optimization comprehensively considers three dimensions: spatial cost, time cost, and load balancing. Spatial cost is calculated by adding the Euclidean distance from the crane's current 3D coordinates to the container source's 3D coordinates to the Euclidean distance from the container source to the target storage location, thus obtaining the total travel distance. Time cost is calculated by dividing the coordinate differences between the crane and the target storage location in the horizontal, vertical, and longitudinal directions by the corresponding equipment movement speed parameters, taking the maximum value, and adding it to the fixed container grabbing and releasing time cost. The load balancing index reflects the equipment's workload by dividing the number of tasks awaiting execution in the crane's current work queue by the system's set queue capacity limit. After normalization, these three indicators are weighted and summed using fixed weighting coefficients of 0.4, 0.35, and 0.25 to obtain a scheduling score. The crane number with the highest score is selected, encapsulated in a scheduling instruction data packet, and sent for execution.

[0017] In one specific embodiment, step S1 includes: Receive job task data packets and parse the task type field, task area field, job task field and task priority field to obtain the task type, task area number, number of container targets and task priority value; Acquire the 3D point cloud data of the container and perform contour extraction processing to obtain the 3D dimension coordinates and orientation tilt angle of the container; Extract the total number of outbound tasks and the total number of inbound tasks for the past 7 days from the historical task record database, calculate the total number of outbound tasks divided by the sum of the total number of outbound tasks and the total number of inbound tasks, and obtain the outbound-inbound ratio. Extract the weight data of all containers that have been put into storage in the past 7 days, calculate the average weight and weight standard deviation, divide the weight standard deviation by the average weight to obtain the weight dispersion, and calculate the total number of stacking operations in the past 7 days and divide it by the total number of outbound tasks to obtain the stacking pressure index.

[0018] Specifically, after the task data packet is transmitted to the warehouse management system via a standard interface protocol, the parsing module locates the positions of each field according to a predefined JSON or XML format structure. The task type field uses an enumerated value encoding method to store three operation types: inbound, outbound, or palletizing. The task area field uses a two-digit code to map to the warehouse physical partition identification table. The task field contains an integer value for the target number of containers and a timestamp range for the completion time window. The task priority field stores an integer value from 1 to 5, with the value directly reflecting the urgency of the task. After outliers are removed from the container's 3D point cloud data using a point cloud filtering algorithm, the contour extraction process obtains the container height by identifying the difference between the maximum and minimum Z-axis coordinates in the point cloud dataset. The extreme differences in the X and Y axes correspond to the container's length and width, respectively. The attitude tilt angle is calculated by fitting the angle between the normal vector of the container's top surface point cloud and the normal vector of the horizontal plane, obtaining the angle parameter reflecting the container's placement attitude.

[0019] The calculation of the inbound / outbound ratio involves filtering records marked as "outbound" from the task type field of the historical operation record database. The total number of these records is used as the numerator, and the denominator is the arithmetic sum of the outbound and inbound record numbers. The ratio reflects the intensity of outbound operations within the statistical period. Weight dispersion calculation involves two layers of data processing. The first layer extracts the set of weight field values ​​from the container attribute table and calculates the arithmetic mean as a weight central trend indicator. The second layer calculates the square root of the sum of the squares of the differences between each weight value and the mean, divided by the sample size, to obtain the weight standard deviation, reflecting the degree of dispersion. The ratio of the standard deviation to the mean eliminates the influence of dimensions, forming a dimensionless dispersion indicator. The stacking pressure index is calculated by querying the operation log table and using the number of records marked as "stacking" as the numerator. The denominator reuses the total number of outbound tasks. The ratio quantifies the average number of stacking operations required for each outbound operation, reflecting the impact of warehouse stacking layout on outbound efficiency.

[0020] In one specific embodiment, step S2, calculating the target weight vector, includes: Read the baseline weight vector; When the inbound / outbound ratio is greater than 0.6, the outbound frequency gain coefficient is calculated, and the outbound frequency base weight is added to the outbound frequency gain coefficient to obtain the target outbound frequency weight. The weight balance gain coefficient is calculated when the weight dispersion is greater than 0.3, and the stacking stability gain coefficient is calculated when the stacking pressure index is greater than 0.2. The target weight vector is obtained by normalizing the target outbound frequency weight, target weight balance weight, target stacking stability weight, and distance convenience benchmark weight.

[0021] Specifically, the baseline weight vector is stored in the weight configuration table as a system initialization parameter. The read operation retrieves four preset values ​​by querying this table: outbound frequency baseline weight (0.35), weight balance baseline weight (0.25), stacking stability baseline weight (0.25), and distance convenience baseline weight (0.15). The calculation trigger condition for the outbound frequency gain coefficient is that the outbound / inbound ratio is greater than the threshold of 0.6. The calculation logic is to extract the difference between the outbound / inbound ratio and the threshold of 0.6, multiply it by the gain amplification factor of 0.5, and add this gain factor to the outbound frequency baseline weight of 0.35 to form the target outbound frequency weight. The gain mechanism is designed so that when outbound operations in the warehouse area are frequent, the weight of the outbound frequency factor in the warehouse location evaluation is automatically increased. The weight balance gain coefficient and the stacking stability gain coefficient adopt the same condition triggering and calculation mode. When the weight dispersion is greater than 0.3, the weight dispersion is calculated, minus 0.3, and then multiplied by 0.4 to obtain the weight balance gain coefficient, which is added to the weight balance reference weight of 0.25. When the stacking pressure index is greater than 0.2, the stacking pressure index is calculated, minus 0.2, and then multiplied by 0.5 to obtain the stacking stability gain coefficient, which is added to the stacking stability reference weight of 0.25.

[0022] The normalization process corrects for cases where the sum of the four weight components might exceed 1. First, the arithmetic sum of the four values—target outbound frequency weight, target weight balance weight, target stacking stability weight, and distance accessibility baseline weight—is calculated as a normalization factor. Then, each of the four weight components is divided by this normalization factor to obtain the normalized weight values. The normalization operation ensures that the sum of the four weight components is strictly equal to 1, satisfying the normalization constraint of the probability distribution. The target weight vector is formed by combining the four normalized weight components in a fixed order of outbound frequency, weight balance, stacking stability, and distance accessibility to create a one-dimensional array structure. This vector serves as the target state parameter for subsequent weight smoothing calculations and is passed to the weight adjustment module.

[0023] In one specific embodiment, the difference between the current weight vector and the target weight vector is decomposed into daily adjustment increments using 5-line linear interpolation, including: Read the current weight vector, and subtract the corresponding weight component in the current weight vector from each weight component in the target weight vector to obtain the weight difference; Multiply the weight difference by 0.2 to obtain the daily adjustment increment; Based on the adjustment days n, each weight component in the current weight vector is added to the corresponding daily adjustment increment multiplied by n to obtain the weight components for that day. The weight vector for that day is then obtained by combining the weight components for that day.

[0024] Specifically, the current weight vector is read from the weight storage module as an array structure containing four components: current outbound frequency weight, current weight balance weight, current stacking stability weight, and current distance convenience weight. The weight difference is calculated by subtracting each component sequentially: the first component of the target weight vector is subtracted from the first component of the current weight vector to obtain the outbound frequency weight difference; the second component is subtracted to obtain the weight balance weight difference; the third component is subtracted to obtain the stacking stability weight difference; and the fourth component is subtracted to obtain the distance convenience weight difference. These four differences reflect the direction and magnitude of the weight adjustment. The daily adjustment increment is calculated by multiplying each of the four weight differences by a fixed coefficient of 0.2. This coefficient corresponds to the reciprocal of the 5-day adjustment cycle, i.e., one-fifth of the total adjustment amount completed each day. The daily adjustment increments for outbound frequency, weight balance, stacking stability, and distance convenience constitute the increment vector and are stored in the weight adjustment cache.

[0025] The adjustment period 'n' is read from the weight adjustment counter, indicating the current day of the 5-day adjustment cycle. The value is an integer from 1 to 5. The daily weight components are calculated using an accumulation formula: the first component of the current weight vector is added to the daily adjustment increment of the outbound frequency, multiplied by 'n' to obtain the outbound frequency weight for that day. The other three components are calculated using the same accumulation logic to obtain the daily weight balance weight, daily stacking stability weight, and daily distance convenience weight, respectively. The daily weight vector is formed by combining the four daily weight components in a fixed order into a one-dimensional array structure. This vector is updated daily within the weight adjustment cycle until 'n' reaches 5, at which point the daily weight vector converges to the target weight vector. The weight storage module then overwrites the current weight vector with the daily weight vector, completing one weight status update.

[0026] In one specific embodiment, it further includes: The total number of stacking operations and the total crane travel distance were calculated 7 days after the implementation of the weight adjustment, and the baseline number of stacking operations and baseline travel distance before the weight adjustment were read. Calculate the rate of change of stacking frequency and the rate of change of running distance; When the rate of change of the number of stacking cycles is less than the expected stacking reduction threshold, the stacking stability compensation gain is calculated and the target stacking stability weight is corrected. When the rate of change of the running distance is less than the expected distance reduction threshold and the ratio of the outbound frequency weight to the distance convenience weight is greater than 4, the distance convenience weight is corrected and stored in the weight correction parameter table.

[0027] Specifically, the weight adjustment effect evaluation module initiates a data statistics process on the 7th day after the weight adjustment is completed. It queries the operation log database for records whose timestamp field is within 7 days of the weight adjustment implementation and whose operation type field is marked as "stack reversing," summing these records to obtain the total number of stack reversing operations. It then queries the operating trajectory table of all crane equipment to extract the values ​​of the lateral movement distance, longitudinal movement distance, and vertical movement distance fields. These values ​​are summed for each crane and then summed for all cranes to obtain the total crane operating distance. The baseline stack reversing count and baseline operating distance are read from the historical baseline data table, which records the stack reversing operations and crane operating distances within the same time span (7 days) before the weight adjustment as a reference value. The stack reversing count change rate is calculated by subtracting the total stack reversing operations from the baseline stack reversing count and dividing by the baseline stack reversing count to obtain a ratio reflecting the reduction in stack reversing count. The operating distance change rate uses the same calculation logic: subtracting the total crane operating distance from the baseline operating distance and dividing by the baseline operating distance to obtain a ratio reflecting the degree of shortening of operating distance.

[0028] The calculation of the stacking stability compensation gain is triggered when the rate of change of stacking frequency is less than the expected stacking reduction threshold of 0.15. The calculation logic is to extract the difference between the expected stacking reduction threshold of 0.15 and the actual rate of change of stacking frequency, multiply it by the compensation coefficient of 0.1 to obtain the stacking stability compensation gain value. This compensation gain is added to the target stacking stability weight to form the corrected target stacking stability weight and written to the corresponding field of the weight correction parameter table. The triggering condition for the distance convenience weight correction includes two judgment logics: the first level judges whether the rate of change of running distance is less than the expected distance reduction threshold of 0.1, and the second level judges whether the ratio of the outbound frequency weight of the day to the current distance convenience weight is greater than the threshold of 4. When both conditions are met, the correction operation is performed, that is, the current distance convenience weight is added to the fixed compensation gain of 0.05 to obtain the corrected distance convenience weight. The weight correction parameter table stores two fields: the corrected target stacking stability weight and the corrected distance convenience weight. This table is read in the next round of weight adaptive calculation. The correction parameters are added to the newly calculated target weight vector to form a closed loop of weight iterative optimization based on the feedback of actual operation effect.

[0029] Figure 2This is a schematic diagram illustrating the convergence process of the weight feedback correction algorithm in this application embodiment. The graph, a line graph, shows the trends of the reduction rate of stacking frequency and the reduction rate of running distance over 10 adjustment cycles. The solid line represents the reduction rate of stacking frequency, which gradually increases from the initial 0% and reaches 41% in the 7th adjustment cycle before stabilizing and entering a convergence state. The dashed line represents the reduction rate of running distance, which increases from 0% to 26% in the 7th cycle, also exhibiting convergence characteristics. The expected stacking reduction threshold of 15% and the expected distance reduction threshold of 10% marked in the graph are two horizontal reference lines. When the actual reduction rate exceeds these two threshold lines in the 2nd and 3rd cycles respectively, the system determines that the weight adjustment effect has reached the expected target. The convergence point marked in the 7th cycle indicates that the weight feedback correction mechanism has completed iterative optimization at this moment. Afterward, the growth rate of the reduction rate of stacking frequency and the reduction rate of running distance slows significantly, indicating that the weight configuration has converged to the optimal solution for the current operating scenario. The convergence curve in the graph verifies the effectiveness of the dynamic weight adaptive algorithm in achieving continuous performance improvement and eventual stability through feedback correction.

[0030] In one specific embodiment, step S3 includes: Retrieve historical outbound records that match the container type, calculate the outbound frequency in the last 30 days, divide by the total outbound frequency in the last 90 days, and multiply by 3 to obtain the outbound probability; The weight deviation is obtained by querying the total weight of containers in the candidate storage location area and the average weight of the entire warehouse. The load-bearing capacity and spatial distance to the outbound passage of the candidate storage location are also queried. The outbound frequency score is obtained by multiplying the outbound probability by the outbound frequency weight in the daily weight vector; the weight balance score is obtained by subtracting the weight deviation from 1 and multiplying it by the weight balance weight in the daily weight vector; the stacking stability score is obtained by multiplying the load-bearing margin by the stacking stability weight in the daily weight vector; and the distance convenience score is obtained by subtracting the spatial distance from the maximum distance threshold, dividing it by the maximum distance threshold, and then multiplying it by the current distance convenience weight in the daily weight vector. The suitability score is obtained by summing the scores for outbound frequency, weight balance, stacking stability, and distance convenience.

[0031] Specifically, the outbound probability prediction module performs a matching query on the historical outbound record database using the container type field. It filters out all historical records whose container type field value is the same as the type of container to be received. In the filtering results, it further filters by the timestamp field to obtain the number of outbound record rows in the most recent 30 days as the short-term outbound frequency. It expands the time window to count the total number of outbound record rows in the most recent 90 days as the long-term outbound base. The number of short-term outbounds is divided by the total number of long-term outbounds to obtain the outbound frequency ratio of 30 days to 90 days. This ratio is multiplied by a magnification factor of 3 to realize the time scale conversion from the 90-day base to the 30-day prediction window to obtain the estimated outbound probability value for the next 30 days.

[0032] The weight deviation query module first reads the region number field from the storage location attribute table based on the candidate storage location number. Then, it queries all container records in the inventory data table that match the region number field using this region number. The total weight of containers in the candidate storage location region is obtained by summing the weight field values ​​of these container records. The average weight of the entire warehouse is obtained by calculating the arithmetic mean of the weight fields of all container records in the inventory data table. The weight deviation index, which reflects the degree of deviation in the weight distribution of the region, is obtained by subtracting the average weight of the entire warehouse from the total weight of containers in the candidate storage location region, taking the absolute value, and then dividing by the average weight of the entire warehouse.

[0033] The load-bearing capacity query obtains the remaining load-bearing capacity of the candidate storage location by subtracting the design load-bearing limit field and the current load-bearing weight field from the candidate storage location attribute table. The spatial distance query obtains the three-dimensional coordinates of the storage location center point, including the horizontal, vertical, and longitudinal coordinates, from the candidate storage location attribute table, and the three-dimensional coordinates of the nearest outbound channel from the outbound channel configuration table. The spatial distance is obtained by summing the square of the horizontal coordinate of the candidate storage location minus the square of the horizontal coordinate of the outbound channel, the square of the difference between the vertical and longitudinal coordinates, and the square of the difference between the longitudinal and longitudinal coordinates, and then taking the square root.

[0034] The outbound frequency score is calculated by multiplying the outbound probability value by the first component of the daily weight vector, i.e., the outbound frequency weight, to obtain the weighted score for this dimension. The weight balance score is calculated by multiplying the difference between 1 and the weight deviation, which represents the degree of weight distribution balance, by the second component of the daily weight vector, i.e., the daily weight balance weight. The stacking stability score is calculated by directly multiplying the load-bearing margin value by the third component of the daily weight vector, i.e., the daily stacking stability weight. The distance convenience score is calculated by subtracting the spatial distance from the maximum distance threshold and then dividing by the maximum distance threshold to achieve distance normalization and inverse mapping, i.e., the smaller the distance, the higher the score. The normalized distance convenience index is multiplied by the fourth component of the daily weight vector, i.e., the current distance convenience weight, to obtain the weighted score for the distance dimension.

[0035] The suitability score is obtained by directly summing four scores: outbound frequency score, weight balance score, stacking stability score, and distance convenience score. The higher the score, the more suitable the candidate storage location is for container storage under the current weight configuration. The storage location evaluation module repeats the above calculation process for each storage location in all candidate storage location sets to obtain its own suitability score. The scoring and sorting unit sorts all candidate storage locations from high to low according to their suitability scores, and selects the storage location number that is first in the sorting, i.e., the storage location with the highest suitability score, as the target storage location and outputs it to the scheduling instruction generation module.

[0036] Figure 3 This diagram illustrates the distribution of scores for each candidate storage location in this embodiment. It shows the score distribution of the five candidate storage locations across four dimensions: outbound frequency, weight balance, stacking stability, and proximity convenience. Storage location D has the highest outbound frequency score of 0.25, indicating its proximity to the outbound channel and historical outbound records showing high-frequency outbound characteristics for the corresponding container type. Storage location C has a weight balance score of 0.11, indicating a small deviation between the total weight of containers in the area and the average weight of the entire storage facility, which is beneficial for balanced weight distribution. Storage location E has a stacking stability score of 0.14, indicating sufficient load-bearing capacity and few obstructing containers. By weighted summing the scores across the four dimensions, storage location D achieves the highest overall suitability score and is selected as the target storage location. The diagram uses bar charts with different filling patterns to visually compare the advantages and disadvantages of each candidate storage location across different evaluation dimensions, providing a visual basis for storage location allocation decisions.

[0037] In one specific embodiment, step S4 includes: Obtain the current position coordinates, operating status, and queue length of each crane, and eliminate faulty cranes to obtain a set of available cranes; For each crane in the set of available cranes, calculate the spatial distance from the current position to the container source position and the spatial distance from the container source position to the target storage location, sum them to obtain the total moving distance, and calculate the expected operation time based on the position coordinate difference and the equipment moving speed parameters; After normalizing the total travel distance, expected operation time, and queue length, the data are weighted and summed using weighting coefficients of 0.4, 0.35, and 0.25. The crane with the highest weighted sum is selected to generate the scheduling instruction.

[0038] Specifically, the crane status query module reads all crane records from the crane status table in the equipment monitoring database. Each record contains a crane number field, a current position coordinate field including three subfields: horizontal position coordinate, vertical position coordinate, and vertical position coordinate, an operation status field using enumerated values ​​to identify three states: idle, running, or fault, and a queue length field recording the number of tasks currently pending execution for the crane.

[0039] The available crane filtering logic iterates through all crane records and checks if the operation status field equals the fault flag value. Crane records with an idle or running operation status are retained to form an available crane set, while faulty cranes are excluded from the scheduling candidate range. Spatial distance calculation performs two Euclidean distance calculations for each crane in the available crane set. The first calculation involves summing the square root of the sum of the crane's current lateral coordinates minus the square of the container source's lateral coordinates, the square of the difference between the longitudinal and vertical coordinates, and the current lateral coordinates to obtain the spatial distance from the crane's current position to the container source position. The second calculation involves summing the square root of the sum ...

[0040] The calculation of the expected operation time involves independent time estimation in three directions. The time in the lateral direction is obtained by subtracting the absolute value of the crane's current lateral position coordinate from the lateral coordinate of the target storage location and then dividing it by the lateral movement speed parameter read from the equipment parameter table. The time in the longitudinal direction is obtained by dividing the absolute value of the difference in the longitudinal coordinates by the longitudinal movement speed parameter. The time in the vertical direction is obtained by dividing the absolute value of the difference in the vertical coordinates by the vertical movement speed parameter. Since the crane's three-dimensional mechanism, including the trolley, hoisting mechanism, has some parallel movement capabilities, the expected operation time is taken as the maximum value of the time in the three directions, which represents the bottleneck direction that limits the overall operation speed. The maximum value is added to the fixed time overhead read from the operation parameter table, including the container grabbing time and release time, to obtain the complete expected operation time value.

[0041] The normalization processing module performs numerical transformations on three indicators: total travel distance, expected job time, and queue length. Total travel distance normalization is obtained by subtracting the total travel distance from the maximum distance threshold read from the configuration table and then dividing by the maximum distance threshold to obtain the normalization score for the distance dimension. Expected job time normalization is obtained by subtracting the expected job time from the maximum time threshold and then dividing by the maximum time threshold to obtain the normalization score for the time dimension. Queue length normalization is obtained by subtracting the queue length from 1 and then dividing by the maximum queue capacity to obtain the normalization score for the load dimension. The larger the values ​​of the three normalization scores, the more obvious the scheduling advantage in the corresponding dimension.

[0042] The scheduling score calculation involves multiplying the normalized score of the total travel distance by a weighting factor of 0.4 to obtain the distance-weighted score, multiplying the normalized score of the expected operation time by a weighting factor of 0.35 to obtain the time-weighted score, and multiplying the normalized score of the queue length by a weighting factor of 0.25 to obtain the load-weighted score. The three weighted scores are then summed to obtain the comprehensive scheduling score of the crane.

[0043] The scheduling decision module repeats the above calculation process for each crane in the available crane set to obtain its own scheduling score. The scoring comparison unit traverses all scheduling score values ​​to identify the maximum value and extracts the crane number corresponding to the maximum scheduling score as the executing crane. The scheduling instruction generation module encapsulates the executing crane number, the three-dimensional coordinates of the container source location, the three-dimensional coordinates of the target storage location, the task priority value, and the expected completion timestamp to form a scheduling instruction data packet. This data packet is sent to the control system of the selected crane through the communication interface to trigger the automated hoisting operation execution process.

[0044] The above describes the automated warehouse container operation method based on a three-dimensional mechanism in the embodiments of this application. The following describes the automated warehouse container operation system based on a three-dimensional mechanism in the embodiments of this application. One embodiment of the automated warehouse container operation system based on a three-dimensional mechanism in the embodiments of this application includes: The receiving module is used to receive job task data packets and container information, and to collect data on the inbound and outbound ratios, weight dispersion, and stacking pressure index over the past 7 days as scene feature parameters. The decomposition module is used to calculate the target weight vector when any parameter in the scene feature parameters exceeds a preset threshold for 3 consecutive days. The difference between the current weight vector and the target weight vector is decomposed into daily adjustment increments by 5-day linear interpolation. The daily adjustment increments are accumulated to the current weight vector to obtain the weight vector for the day. The weighting module is used to predict the outbound probability based on the container type, query the weight deviation, load-bearing capacity, and distance to the outbound channel of the candidate storage location, and sum the outbound probability, weight deviation, load-bearing capacity, and distance with the weight vector of the day to obtain the suitability score, and select the storage location with the highest suitability score as the target storage location. The scheduling module is used to calculate the three-dimensional moving distance and operation time of the crane from the container source location and the target storage location, optimize the scheduling based on the moving distance, operation time and queue length, and select the crane with the highest scheduling score to generate a scheduling instruction.

[0045] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automating warehouse container operations based on a three-dimensional mechanism, characterized in that, The method includes: Step S1: Receive the job task data packet and container information, and calculate the inbound / outbound ratio, weight dispersion and stacking pressure index of the past 7 days as scene feature parameters; Step S2: When any parameter in the scene feature parameters exceeds the preset threshold for 3 consecutive days, calculate the target weight vector. Decompose the difference between the current weight vector and the target weight vector into daily adjustment increments using 5-day linear interpolation. Accumulate the daily adjustment increments to the current weight vector to obtain the weight vector for the day. Step S3: Predict the outbound probability based on the container type, query the weight deviation, load-bearing capacity, and distance to the outbound channel of the candidate storage location, and obtain the suitability score by weighted summation of the outbound probability, weight deviation, load-bearing capacity, and distance with the weight vector of the day, and select the storage location with the highest suitability score as the target storage location. Step S4: Calculate the three-dimensional moving distance and operation time of the crane from the container source location and the target storage location. Based on the moving distance, operation time and queue length, optimize the scheduling and select the crane with the highest scheduling score to generate a scheduling instruction.

2. The automated warehouse container operation method based on a three-dimensional mechanism according to claim 1, characterized in that, Step S1 includes: Receive job task data packets and parse the task type field, task area field, job task field and task priority field to obtain the task type, task area number, number of container targets and task priority value; Acquire the 3D point cloud data of the container and perform contour extraction processing to obtain the 3D dimension coordinates and orientation tilt angle of the container; Extract the total number of outbound tasks and the total number of inbound tasks for the past 7 days from the historical task record database, calculate the total number of outbound tasks divided by the sum of the total number of outbound tasks and the total number of inbound tasks, and obtain the outbound-inbound ratio. Extract the weight data of all containers that have been put into storage in the past 7 days, calculate the average weight and weight standard deviation, divide the weight standard deviation by the average weight to obtain the weight dispersion, and calculate the total number of stacking operations in the past 7 days and divide it by the total number of outbound tasks to obtain the stacking pressure index.

3. The automated warehouse container operation method based on a three-dimensional mechanism according to claim 1, characterized in that, In step S2, calculating the target weight vector includes: Read the baseline weight vector; When the inbound / outbound ratio is greater than 0.6, the outbound frequency gain coefficient is calculated, and the outbound frequency base weight is added to the outbound frequency gain coefficient to obtain the target outbound frequency weight. The weight balance gain coefficient is calculated when the weight dispersion is greater than 0.3, and the stacking stability gain coefficient is calculated when the stacking pressure index is greater than 0.

2. The target weight vector is obtained by normalizing the target outbound frequency weight, target weight balance weight, target stacking stability weight, and distance convenience benchmark weight.

4. The automated warehouse container operation method based on a three-dimensional mechanism according to claim 3, characterized in that, The step of decomposing the difference between the current weight vector and the target weight vector into daily adjustment increments using 5-line linear interpolation includes: Read the current weight vector, and subtract the corresponding weight component in the current weight vector from each weight component in the target weight vector to obtain the weight difference; Multiply the weight difference by 0.2 to obtain the daily adjustment increment; Based on the adjustment days n, each weight component in the current weight vector is added to the corresponding daily adjustment increment multiplied by n to obtain the weight components for that day. The weight vector for that day is then obtained by combining the weight components for that day.

5. The automated warehouse container operation method based on a three-dimensional mechanism according to claim 4, characterized in that, Also includes: The total number of stacking operations and the total crane travel distance were calculated 7 days after the implementation of the weight adjustment, and the baseline number of stacking operations and baseline travel distance before the weight adjustment were read. Calculate the rate of change of stacking frequency and the rate of change of running distance; When the rate of change of the number of stacking cycles is less than the expected stacking reduction threshold, the stacking stability compensation gain is calculated and the target stacking stability weight is corrected. When the rate of change of the running distance is less than the expected distance reduction threshold and the ratio of the outbound frequency weight to the distance convenience weight is greater than 4, the distance convenience weight is corrected and stored in the weight correction parameter table.

6. The automated warehouse container operation method based on a three-dimensional mechanism according to claim 1, characterized in that, Step S3 includes: Retrieve historical outbound records that match the container type, calculate the outbound frequency in the last 30 days, divide by the total outbound frequency in the last 90 days, and multiply by 3 to obtain the outbound probability; The weight deviation is obtained by querying the total weight of containers in the candidate storage location area and the average weight of the entire warehouse. The load-bearing capacity and spatial distance to the outbound passage of the candidate storage location are also queried. The outbound frequency score is obtained by multiplying the outbound probability by the outbound frequency weight in the daily weight vector; the weight balance score is obtained by subtracting the weight deviation from 1 and multiplying it by the weight balance weight in the daily weight vector; the stacking stability score is obtained by multiplying the load-bearing margin by the stacking stability weight in the daily weight vector; and the distance convenience score is obtained by subtracting the spatial distance from the maximum distance threshold, dividing it by the maximum distance threshold, and then multiplying it by the current distance convenience weight in the daily weight vector. The suitability score is obtained by summing the scores for outbound frequency, weight balance, stacking stability, and distance convenience.

7. The automated warehouse container operation method based on a three-dimensional mechanism according to claim 1, characterized in that, Step S4 includes: Obtain the current position coordinates, operating status, and queue length of each crane, and eliminate faulty cranes to obtain a set of available cranes; For each crane in the set of available cranes, calculate the spatial distance from the current position to the container source position and the spatial distance from the container source position to the target storage location, sum them to obtain the total moving distance, and calculate the expected operation time based on the position coordinate difference and the equipment moving speed parameters; After normalizing the total travel distance, expected operation time, and queue length, the data are weighted and summed using weighting coefficients of 0.4, 0.35, and 0.

25. The crane with the highest weighted sum is selected to generate the scheduling instruction.

8. An automated warehouse container operation system based on a three-dimensional mechanism, characterized in that, For implementing the warehouse container automation operation method based on a three-dimensional mechanism as described in any one of claims 1-7, the warehouse container automation operation system based on a three-dimensional mechanism comprises: The receiving module is used to receive job task data packets and container information, and to collect data on the inbound and outbound ratios, weight dispersion, and stacking pressure index over the past 7 days as scene feature parameters. The decomposition module is used to calculate the target weight vector when any parameter in the scene feature parameters exceeds a preset threshold for 3 consecutive days. The difference between the current weight vector and the target weight vector is decomposed into daily adjustment increments by 5-day linear interpolation. The daily adjustment increments are accumulated to the current weight vector to obtain the weight vector for the day. The weighting module is used to predict the outbound probability based on the container type, query the weight deviation, load-bearing capacity, and distance to the outbound channel of the candidate storage location, and sum the outbound probability, weight deviation, load-bearing capacity, and distance with the weight vector of the day to obtain the suitability score, and select the storage location with the highest suitability score as the target storage location. The scheduling module is used to calculate the three-dimensional moving distance and operation time of the crane from the container source location and the target storage location, optimize the scheduling based on the moving distance, operation time and queue length, and select the crane with the highest scheduling score to generate a scheduling instruction.

9. The system according to claim 8, characterized in that, Receive job task data packets and container information, and statistically analyze the inbound / outbound ratio, weight dispersion, and stacking pressure index over the past 7 days as scene characteristic parameters, including: Receive job task data packets and parse the task type field, task area field, job task field and task priority field to obtain the task type, task area number, number of container targets and task priority value; Acquire the 3D point cloud data of the container and perform contour extraction processing to obtain the 3D dimension coordinates and orientation tilt angle of the container; Extract the total number of outbound tasks and the total number of inbound tasks for the past 7 days from the historical task record database, calculate the total number of outbound tasks divided by the sum of the total number of outbound tasks and the total number of inbound tasks, and obtain the outbound-inbound ratio. Extract the weight data of all containers that have been put into storage in the past 7 days, calculate the average weight and weight standard deviation, divide the weight standard deviation by the average weight to obtain the weight dispersion, and calculate the total number of stacking operations in the past 7 days and divide it by the total number of outbound tasks to obtain the stacking pressure index.

10. The system according to claim 8, characterized in that, Calculate the three-dimensional travel distance and operation time of the crane from the container source location and the target storage location. Based on the travel distance, operation time, and queue length, optimize the scheduling and select the crane with the highest scheduling score to generate scheduling instructions, including: Obtain the current position coordinates, operating status, and queue length of each crane, and eliminate faulty cranes to obtain a set of available cranes; For each crane in the set of available cranes, calculate the spatial distance from the current position to the container source position and the spatial distance from the container source position to the target storage location, sum them to obtain the total moving distance, and calculate the expected operation time based on the position coordinate difference and the equipment moving speed parameters; After normalizing the total travel distance, expected operation time, and queue length, the data are weighted and summed using weighting coefficients of 0.4, 0.35, and 0.

25. The crane with the highest weighted sum is selected to generate the scheduling instruction.

Images