Shelf item classification method and system based on commodity attributes

Abstract

The present application relates to the technical field of inventory management, in particular to a shelf article classification method and system based on commodity attributes, comprising the following steps: obtaining commodity attributes based on a laser range finder and an electronic scale, forming a commodity characteristic parameter set in combination with a code scanning gun and a warehouse-out frequency, analyzing the fit degree of the commodity characteristic parameter set with shelf carrying capacity, space and disturbance resistance performance, calculating AGV path and circulation activity, sorting to obtain a classification partition sequence, collecting carrying process energy consumption and progress, and outputting classification operation offset characteristics. The present application comprehensively collects and parameterizes describes multi-dimensional attributes of commodities, establishes an actual adaptation-based attribution relationship between commodities and shelves, automatically analyzes space adaptation, carrying capacity and disturbance factors, dynamically sorts by fusing path efficiency and circulation activity, generates partition allocation results and collects carrying energy consumption and progress information in real time, feeds back state offsets in the classification and carrying process, and realizes fine control of shelf classification and operation.

Description

Technical Field

[0001] This invention relates to the field of inventory management technology, and in particular to a method and system for classifying shelf items based on product attributes. Background Technology

[0002] Inventory management involves the systematic management of the entire process of goods entering and leaving the warehouse, storage, classification, replenishment, and shelf management. This field covers multiple aspects such as warehousing and logistics, retail terminals, and supply chain management. Among them, traditional shelf classification methods refer to the classification method of placing goods in specific shelf locations based on human experience or rules. Usually, the category, use, or sales frequency of goods are manually identified and divided into specific areas before being manually placed on the shelves. Some methods use static coding to establish a mapping relationship between fixed locations and product categories, or ABC classification based on the turnover rate of goods in historical sales, and set the corresponding shelf locations based on this.

[0003] Existing technologies rely on human judgment and static rules to classify shelf items, lacking detailed consideration of the physical attributes and circulation characteristics of goods. Classification standards are difficult to adjust in real time according to actual sales changes and product updates. Shelf space division and item placement are arbitrary, lacking analysis of multi-dimensional attributes such as the volume, weight, and packaging of different goods. The setting and allocation of storage locations ignore path efficiency and operational data feedback, which easily leads to problems such as wasted shelf space, repeated handling, item misplacement, and untimely replenishment. This results in a rigid inventory structure, insufficient ability to respond to market changes, and limitations on overall management accuracy and warehousing operation efficiency. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a method and system for classifying shelf items based on product attributes. The technical solution is as follows:

[0005] On the one hand, a method for classifying shelf items based on product attributes is provided, including the following steps:

[0006] S1: Based on a laser rangefinder, collect the three-dimensional dimensions of the product, combine the weight data of the electronic scale to identify abnormal fluctuations, enter the packaging material information through a barcode scanner, verify the consistency with the standard catalog, and obtain the product feature parameter set;

[0007] S2: Based on the set of product characteristic parameters, determine the compatibility between the product and the shelf load-bearing capacity, shelf height and shock resistance, analyze the load trend of the weighing sensor monitoring data and the product weight, compare the product volume with the shelf space capacity, screen the matching of shelf disturbance resistance performance and impact resistance performance, and obtain the classification adaptation factor.

[0008] S3: Based on the AGV navigation path, calculate the distance from the sorting point to the shelf, analyze the changes in path nodes, combine the outbound frequency, judge the trend of commodity circulation speed, arrange the priority of the storage location in order, and obtain the path priority index.

[0009] S4: Based on the classification adaptation factor and path priority index, select the shelf numbers that meet the conditions, compare the matching of each shelf with the product attributes, analyze the arrival path efficiency, determine the allocation sequence, and obtain the classification partition sequence.

[0010] S5: Based on the classification and partitioning sequence, adjust the preferred shelf arrangement, analyze the energy consumption changes collected during AGV handling, determine the difference between the operation progress and the planned progress, screen for classification data offset phenomena, and obtain classification operation offset characteristics.

[0011] On the other hand, the commodity feature parameter set includes size attributes, quality attributes, material attributes, and circulation attributes; the classification adaptation factor includes spatial adaptation parameters, carrying capacity adaptation parameters, and anti-disturbance adaptation parameters; the path priority index includes path distance parameters, circulation frequency parameters, and priority ranking parameters; the classification partition sequence includes partition number, allocation order, and scheduling identifier; and the classification operation offset features include energy consumption features, progress features, and path offset features.

[0012] On the other hand, the specific steps for obtaining the product feature parameter set are as follows:

[0013] S101: Based on a laser rangefinder, analyze the acquired spatial coordinate data of the product surface, fit the boundary of the product point cloud shape model, compare the three-dimensional range of the boundary with the standard three-dimensional size parameter set of the product, determine the index position of the spatial boundary that is abnormal, identify the point position that differs from the standard size, and obtain the three-dimensional size deviation set.

[0014] S102: Analyze the weighing data of goods collected by the electronic scale, compare the continuous weighing collection sequence with the standard quality parameters of the goods, determine the quality change of the same goods in a continuous weighing process, identify the time period of continuous fluctuation, and obtain the set of quality fluctuation intervals.

[0015] S103: Based on the quality fluctuation interval set and the three-dimensional size deviation set, compare the packaging material label collected by the barcode scanner with the standard packaging catalog template, determine the packaging material discrepancies, and at the same time analyze the periodic trend of the corresponding product sales system's outbound frequency, identify the codes of abnormal fluctuation amplitude, and obtain the product feature parameter set.

[0016] On the other hand, the specific steps for obtaining the classification adaptation factor are as follows:

[0017] S201: Based on the product feature parameter set, analyze the quality and size information, compare the product weight with the shelf load monitoring data, determine whether the load capacity matches, analyze the spatial compatibility between the shelf height and the three dimensions of the product, identify combinations with limited space or incompatible load, and obtain shelf structure adaptation data.

[0018] S202: Based on the shelf structure adaptation data, compare the volume data with the remaining shelf volume information, calculate the ratio difference of the product volume in the shelf space, determine the ratio change in each shelf unit, identify the combination of ratio changes, and obtain the space volume ratio index.

[0019] S203: Based on the space volume ratio index, call the continuous operation cycle shelf vibration spectrum data collected by the vibration recorder, compare the degree of overlap between the impact resistance structural parameters of the goods and the vibration frequency band distribution, determine the combination of shelves and goods with impact risk, and obtain the classification adaptation factor.

[0020] On the other hand, the steps for obtaining the path priority index are as follows:

[0021] S301: Based on the AGV navigation path, calculate the driving path relationship from the product sorting point to each shelf, determine the changes in the coordinates of the path nodes during continuous movement, filter out sections with dense turning or concentrated structural changes in the path, and obtain the path change feature group.

[0022] S302: Based on the path change feature group, call the outbound frequency record of the corresponding shelf goods in the sales system, compare the frequency change status in each period, determine the changing trend of the goods circulation speed, and group and classify the goods according to the trend consistency to obtain the circulation trend distribution group.

[0023] S303: Based on the aforementioned distribution trend group, analyze the relationship between the path distance and outbound frequency of the goods, compare the correlation between the distance change within each path segment and the degree of circulation activity, determine the matching order between the path and the circulation of goods, and arrange the shelf path nodes in order to obtain the path priority index.

[0024] On the other hand, the specific steps for obtaining the classification partition sequence are as follows:

[0025] S401: Based on the classification adaptation factor and path priority index, the shelf number is retrieved, the load monitoring status, spatial parameters and path sorting identifier of each shelf are compared, the correspondence between each set of shelf parameters and product attributes is determined, and the shelf code that fully meets the matching rules in terms of load capacity, spatial configuration and path order is identified to obtain a set of candidate shelf numbers.

[0026] S402: Based on the candidate shelf number set, compare the matching degree between the shelf load status, shelf height conditions and product parameters, determine the matching result of each shelf for the same product, group and classify each shelf code according to the degree of matching, and obtain the shelf adaptation order index.

[0027] S403: Based on the shelf adaptation order index, analyze the path sorting information, compare the changes in the order of the shelf arrival paths, determine the combination of the shelf in the path order and adaptation sorting, and arrange the shelf codes in sequence to obtain the classification and partitioning sequence.

[0028] On the other hand, the specific steps for obtaining the classification job offset features are as follows:

[0029] S501: Based on the classification and partitioning sequence, compare the energy consumption data during the AGV transport to the shelf, determine the energy change trend on each transport path, identify batches where energy consumption fluctuates, and obtain the transport energy consumption change trajectory.

[0030] S502: Based on the energy consumption change trajectory of the handling, analyze the operation node time of the corresponding batch, compare the difference between the operation progress and the planned progress, determine the progress deviation of each stage of the operation, filter the time segments in which the progress is out of sync, and obtain the operation cycle offset index.

[0031] S503: Based on the operation cycle offset index, analyze the handling process parameters, compare the changing trends of the parameter collection data and the process standard data, determine the fluctuations or offset phenomena in the classification and handling process, identify abnormal batches and summarize them to obtain the classification operation offset characteristics.

[0032] On the other hand, the abnormal fluctuation refers to the fluctuation of the collected commodity weight data that exceeds the reasonable range, and the standard catalog is a list of packaging materials and specifications preset in the warehousing system.

[0033] On the other hand, the load trend is recorded in real time by weighing sensors to change the load on the shelf and compared with the weight of the goods to be sorted and put on the shelf, so as to monitor the change trend of the shelf load over time or the shelving action. The shelf anti-disturbance performance refers to the stability and deformation resistance of the shelf structure under the influence of external forces and vibrations.

[0034] On the other hand, a shelf item classification system based on product attributes is provided. This system is applied to shelf item classification methods based on product attributes, including:

[0035] The attribute acquisition module is based on a laser rangefinder to collect the three-dimensional dimensions of the product, and combines the weight data of the electronic scale to identify abnormal fluctuations. It also uses a barcode scanner to input packaging material information, verifies consistency with the standard catalog, and obtains a set of product feature parameters.

[0036] Based on the product feature parameter set, the adaptation analysis module determines the compatibility between the product and the shelf load-bearing capacity, shelf height, and shock resistance. It analyzes the load trend of the weighing sensor monitoring data and the product weight, compares the product volume with the shelf space capacity, screens the matching of the shelf's anti-interference performance and impact resistance, and obtains the classification adaptation factor.

[0037] The path sorting module calculates the distance from the sorting point to the shelf based on the AGV navigation path, analyzes the changes in path nodes, combines the outbound frequency to determine the trend of commodity circulation speed, and arranges the priority of the storage locations in order to obtain the path priority index.

[0038] Based on the classification adaptation factor and path priority index, the partitioning decision module filters the shelf numbers that meet the conditions, compares the matching of each shelf with the product attributes, analyzes the arrival path efficiency, determines the allocation sequence, and obtains the classification partitioning sequence.

[0039] Based on the classification and partitioning sequence, the status monitoring module adjusts the preferred shelf arrangement, analyzes the energy consumption changes collected during AGV handling, determines the difference between the operation progress and the planned progress, screens for classification data offset phenomena, and obtains classification operation offset characteristics.

[0040] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0041] By comprehensively collecting and parametrically describing the multidimensional attributes of goods, an affiliation relationship between goods and shelves based on actual adaptation is established. The system automatically analyzes space adaptation, load-bearing capacity, and anti-interference factors, integrates path efficiency and circulation activity to complete dynamic sorting, generates zoning allocation results, and collects handling energy consumption and progress information in real time. It also provides feedback on state deviations during the classification and handling process, enabling refined management of shelf classification and operations, significantly improving space utilization and zoning accuracy, and enhancing the dynamic response capability to changes in demand and operational anomalies. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a flowchart of the main steps of the present invention;

[0044] Figure 2 This is a flowchart of steps S1 of the present invention;

[0045] Figure 3 This is a flowchart of steps S2 of the present invention;

[0046] Figure 4 This is a flowchart of steps S3 of the present invention;

[0047] Figure 5 This is a flowchart of step S4 of the present invention;

[0048] Figure 6 This is a flowchart of steps S5 of the present invention;

[0049] Figure 7 This is a system block diagram of the present invention. Detailed Implementation

[0050] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0051] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0052] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0053] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0054] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0055] This invention provides a method for classifying shelf items based on product attributes, such as... Figure 1 As shown, it includes the following steps:

[0056] S1: Based on a laser rangefinder, obtain the three-dimensional dimensions of the product, analyze the difference between the rangefinder data and the product shape, combine the weight data of the electronic scale to determine whether there are abnormal fluctuations, compare the packaging material recorded by the barcode scanner with the standard packaging catalog, check the product outbound frequency in the sales system, determine the seasonal variation pattern, and obtain the product feature parameter set.

[0057] S2: Based on the set of product feature parameters, determine the fit between product parameters and shelf load-bearing capacity, shelf height and seismic resistance, analyze the load change trend of weighing sensor monitoring data and product weight, compare the fit between product volume and remaining shelf space capacity, screen the matching between shelf disturbance resistance performance reflected by vibration recorder and product impact resistance performance, and obtain classification fit factor.

[0058] S3: Based on the AGV navigation path, calculate the distance from the product sorting point to each shelf, analyze the continuous changes in path nodes, combine the outbound frequency of the sales system to judge the trend of changes in product circulation speed, compare the correlation between path distance and product circulation activity, arrange the priority of storage locations according to the analysis order, and obtain the path priority index.

[0059] S4: Based on the classification adaptation factor and path priority index, select the shelf numbers that meet the conditions, compare the adaptability differences of each shelf to the product attributes, analyze the efficiency of the arrival path, and use progressive sorting to determine the allocation sequence to obtain the classification partition sequence.

[0060] S5: Based on the classification and partition sequence, adjust the corresponding preferred shelf arrangement, analyze the changes in energy consumption collected during the AGV transport to the shelf, determine the difference between the operation progress and the planned progress, screen for data offset phenomena in the classification process, and summarize the current classification and handling status to obtain the classification operation offset characteristics.

[0061] The commodity feature parameter set includes size attributes, quality attributes, material attributes, and circulation attributes; the classification adaptation factors include spatial adaptation parameters, load-bearing adaptation parameters, and disturbance resistance adaptation parameters; the path priority index includes path distance parameters, circulation frequency parameters, and priority ranking parameters; the classification partition sequence includes partition number, allocation order, and scheduling identifier; and the classification operation offset features include energy consumption features, progress features, and path offset features.

[0062] In S1, abnormal fluctuations refer to fluctuations in the collected commodity weight data that exceed a reasonable range. For example, the weighing results of the same type of commodity deviate significantly from the expected standard, which may be due to equipment errors or abnormal commodity conditions (such as damage or inclusions). The standard packaging catalog is a list of pre-set packaging materials and specifications in the warehousing system. The packaging information identified by the barcode scanner must correspond to this catalog to verify the compliance of the commodity packaging. Outbound frequency refers to the actual number of times a commodity is taken out (picked up or sold) within a certain period of time (such as daily or weekly), which is a data measure of the activity level of commodity circulation. Seasonal variation pattern refers to the regular changes in commodity outbound frequency with the season or cycle. For example, sales of cold drinks increase significantly in summer, and down jackets are frequently outbound in winter. This change is used to guide dynamic zoning and replenishment strategies.

[0063] In S2, the load change trend is recorded in real time by weighing sensors, and compared with the weight of the goods to be sorted and shelved. This observation of the change trend of the shelf load over time or shelving action is used to assess the shelf load safety. The fit relationship refers to the actual accommodation relationship between the size (volume, shape) of the goods and the remaining space on the shelf, that is, to determine whether the shelf space can safely and compliantly store the goods. The vibration recorder is a device used to monitor the vibration response of the shelf structure in warehousing operations. It can collect the physical vibration signals of the shelf during handling, shelving, etc. Shelf disturbance resistance refers to the stability and deformation resistance of the shelf structure under the influence of external forces and vibrations, which is often reflected by vibration recorder data. Impact resistance refers to the ability of the packaging and materials of the goods themselves to resist the influence of external forces such as collisions and drops. It is used to determine whether fragile or high-value goods can be classified into a certain area of ​​the shelf.

[0064] In S3, the sorting point refers to the workstation or designated space in warehouse operations where AGVs or workers perform preliminary sorting, identification, and shelving preparation of goods, and it is the starting point for path planning. Path nodes refer to each turning point, station, or task node along the path during AGV navigation. Continuous path nodes are used to calculate the actual movement route of the AGV from the sorting point to the target shelf. The circulation speed change trend refers to the characteristics of the outbound speed of goods at different time periods, used to determine whether the demand for goods has increased or decreased recently, providing a decision reference for dynamic zoning. Path distance refers to the actual travel distance of the AGV from the sorting point to each candidate shelf, including the total length of straight and turning segments. Correlation refers to analyzing the relationship between path distance and the activity of goods circulation, and determining whether high-circulation goods should be prioritized for allocation to locations with shorter or easier access. Location priority refers to assigning a priority ranking to each shelf location based on multi-factor analysis, guiding automatic zoning and shelving actions.

[0065] In S4, the eligible shelf numbers are selected from all shelves after preliminary screening and adaptation judgment, meeting the product attribute requirements, having sufficient space, and being easy to handle; adaptability difference refers to the differences in parameters such as load-bearing capacity, size, and environmental adaptability of different shelves for the current product, and parameter comparison clarifies which set of shelves is more suitable for the product; arrival path refers to the path taken by the AGV or other automated handling device from the current position to the target shelf, used to evaluate the convenience and efficiency of handling; progressive sorting compares all candidate shelves in sequence according to priority rules, sorts them from best to second best, forming a clear allocation queue; allocation sequence refers to the specific sequence or list obtained after sorting, indicating which shelves the product should be allocated to and what the priority order is.

[0066] In S5, the preferred shelf arrangement refers to the shelf number and its arrangement selected after classification and zoning, which is the target location for actual handling and shelving operations; energy consumption change refers to the trend data of the electrical energy or power resources consumed by the AGV during actual handling as a function of path, weight, time, etc.; operation progress refers to the actual progress of the operation process such as handling, classification, and shelving of goods, including time nodes, process completion, etc.; planned progress refers to the planned goals such as the pre-established operation completion timetable and standard process nodes; data deviation refers to the difference or deviation between the data collected in the operation and the planned or standard data, which is used to reflect possible problems or anomalies in the execution process; classification and handling status refers to the overall performance result of actual classification and handling, which is derived by summarizing various parameters such as time, energy consumption, and path execution during the handling process.

[0067] like Figure 2 As shown, the specific steps for obtaining the product feature parameter set are as follows:

[0068] S101: Based on a laser rangefinder, analyze the acquired spatial coordinate data of the product surface, fit the boundary of the product point cloud shape model, compare the three-dimensional range of the boundary with the standard three-dimensional size parameter set of the product, determine the index position of the spatial boundary that is abnormal, identify the point position that differs from the standard size, and obtain the three-dimensional size deviation set.

[0069] To acquire spatial coordinate data of the product surface, a ranging device is fixed at the edge of the work area. Continuous rotational scanning is performed on the product to collect positional data at fixed angular intervals. These coordinate points are then aggregated into a 3D point cloud model. After the point cloud is formed, the outer boundary of the product model is extracted using boundary scanning. From this boundary, the maximum outer edge and minimum inner edge of the product in the length, width, and height directions are extracted to form a three-dimensional boundary range. This range is then compared direction-by-direction with the standard 3D dimensional parameters registered in the warehousing system. For example, in the length direction, if the currently measured range is 12 mm to 212 mm, while the standard is 200 mm, the error is 12 mm. The system then determines whether this exceeds a preset deviation threshold. If the error exceeds 5 mm, the direction is marked as abnormal. Then, all points at the boundary in that direction whose error exceeds the judgment value are extracted and aggregated into an offset set. At the same time, all offset points are compared with the total number of points to determine the proportion of offset points. For example, if the total number of points is 30,000 and the number of offset points is 3,500, the offset ratio is 11.7%, which exceeds the set abnormal ratio threshold of 10%. Then, it is determined that the overall size of the product is abnormal, and the offset position is recorded to form a three-dimensional size deviation set. If a piece of protruding soft plastic material on the left side of the product is found to extend by 15 mm during a certain collection, causing the width to increase from 180 mm to 195 mm, this change is judged as a boundary offset and included in the deviation data set.

[0070] S102: Analyze the weighing data of goods collected by the electronic scale, compare the continuous weighing collection sequence with the standard quality parameters of the goods, determine the quality change of the same goods in a continuous weighing process, identify the time period of continuous fluctuation, and obtain the set of quality fluctuation intervals.

[0071] When collecting continuous weighing data of goods on an electronic scale, the data acquisition frequency is set to twice per second, and recording continues for 30 seconds to obtain a set of weighing values. This set of data is then compared with the standard weight of the goods frame by frame for difference analysis. During the analysis, five consecutive data points are selected as a group for sliding processing. It is determined whether most data points in each group differ from the standard weight exceeding a set error threshold. For example, if the standard weight is 2000 grams, and more than three values ​​in each group fluctuate outside ±50 grams, then that time period is judged as an abnormal fluctuation, and the time interval is recorded. The start and end points are defined as abnormal fluctuation intervals. If the data detected between the 12th and 18th seconds are 1980g, 2060g, 2070g, 1955g, 2090g, and 2010g respectively, all exceeding the judgment value range multiple times, then this time period is defined as an abnormal fluctuation. If a total of seven similar fluctuation segments are detected during the entire weighing process, with each segment lasting 3 to 8 seconds, then the segments are clustered to form a set of quality fluctuation intervals for the product. At the same time, the maximum and minimum weight difference ranges corresponding to each interval are marked and numbered for subsequent tracking.

[0072] S103: Based on the quality fluctuation interval set and the three-dimensional size deviation set, compare the packaging material labels collected by the barcode scanner with the standard packaging catalog template, determine the packaging material discrepancies, and at the same time analyze the periodic trend of the corresponding product sales system's outbound frequency, identify the codes of abnormal fluctuation amplitude, and obtain the product feature parameter set.

[0073] After obtaining the three-dimensional dimension deviation set and quality fluctuation range set of the product, the product label data collected by the barcode scanner is retrieved. The packaging material field is identified and compared with the standard packaging material template. If the product is actually identified as "plastic soft film" while the standard template is "corrugated cardboard box", the product packaging material is considered inconsistent, and this inconsistency is marked as abnormal packaging material. Then, the outbound frequency data of the product recorded in the sales system is searched, and the daily outbound records of the past two months are extracted. These are divided into seven-day periods, and the difference between the maximum and minimum values ​​of each period is calculated to determine the outbound fluctuation. For example, if the maximum outbound record in the second period is 30 times and the minimum is 12 times, the fluctuation range is 18 times, which exceeds the pre-set fluctuation range judgment standard value of 10 times. Therefore, this period is marked as an abnormal period. At the same time, the corresponding product number is bound to the abnormal period identifier. The four types of information, namely three-dimensional dimension deviation, quality fluctuation range, abnormal packaging material, and abnormal outbound frequency fluctuation, are integrated to form a product feature parameter set for subsequent shelf adaptation and classification analysis tasks.

[0074] like Figure 3 As shown, the specific steps for obtaining the classification adaptation factor are as follows:

[0075] S201: Based on the product feature parameter set, analyze quality and size information, compare product weight with shelf load monitoring data, determine whether the load capacity matches, analyze the spatial compatibility between shelf height and product three-dimensional dimensions, identify combinations with limited space or incompatible load, and obtain shelf structure adaptation data.

[0076] First, the standard weight and actual measured weight of each item are extracted, and the real-time load-bearing data of the corresponding shelf unit in the current shelf monitoring system is read. The process involves comparing the actual weight of the item with the maximum load-bearing capacity of the shelf unit. When the item weight exceeds 80% of the shelf's maximum load-bearing capacity, it is marked as a warning zone; when it exceeds 90%, it is marked as a mismatch. For example, if an item weighs 18 kg, the shelf's maximum load-bearing capacity is 20 kg, and the currently loaded item weighs 3 kg, then the remaining load-bearing capacity is only 17 kg, which is considered a mismatch. This combination is recorded. Then, the effective height of the shelf unit is analyzed and compared with the height, width, and length of the item's three-dimensional dimensions. The maximum height data of the item is extracted and compared with the vertical clearance between shelves. When the difference is less than 10 mm, it is marked as a space restriction. If the clearance is only 95% of the item's height... The following are considered unsuitable for placement: For example, if the shelf height is 280 mm and the product height is 275 mm, the remaining space is only 5 mm. This situation is considered a high-density placement risk. If the product width is 380 mm and the shelf width is 400 mm, then horizontal placement meets the requirements. Such combinations are compiled into structural adaptation assessment records. Further, by tracing back the current analysis object's partition and corresponding shelf identifier through the record number, it is confirmed whether there are multiple structural conflicts. For example, if a product has dimensions of 400×300×280 mm and a weight of 18 kg, and it is allocated to a shelf unit with a net space of only 380×290×270 mm and a current remaining load capacity of 15 kg, then this combination has both three-dimensional space insufficiency and weight exceeding limits. All mismatched combination records are identified, and shelf structure adaptation data is generated.

[0077] S202: Based on shelf structure adaptation data, compare volume data with shelf remaining volume information, calculate the ratio difference of product volume in shelf space, determine the ratio change in each shelf unit, identify the combination of ratio changes, and obtain the space volume ratio index.

[0078] Product volume data is extracted by directly calculating the product's length, width, and height. Then, the remaining effective volume information of the corresponding shelf unit is retrieved from the shelf database. The two volume values ​​are compared to calculate the proportion of product volume to shelf space. Furthermore, the remaining volume status of each shelf at the current moment is recorded. For example, if a product has a volume of 0.025 cubic meters and is to be placed in a shelf unit with a remaining volume of 0.03 cubic meters, the ratio is 83.3%. This ratio is then assessed to determine if it falls within the high occupancy range: over 90% is a high-risk area for overcrowding, 80%-90% is moderate, 60%-80% is ideal, and below 60% is a waste of space. Recorded units with a ratio exceeding 90% are marked as having a high ratio. If a shelf already contains two products with proportions of 92% and 89% respectively, and a new product is added with a proportion of 88%, then the shelf unit is marked as a combination with a high proportion. By summarizing the current proportion changes of each shelf unit and comparing the change in proportion between two consecutive times, such as from 60% to 95%, the change is 35%. If it exceeds the set change benchmark value of 25%, it is recorded as a combination with drastic proportion changes. This benchmark value is set based on the average daily volume change rate in historical warehouse operation records. In a specific warehouse environment, the average value is 22%, so 25% is set as the judgment threshold. In this way, combination data with prominent changes in multiple time slices are identified and organized into a space volume proportion index.

[0079] S203: Based on the spatial volume ratio index, call the continuous operation cycle shelf vibration spectrum data collected by the vibration recorder, compare the degree of overlap between the impact resistance structural parameters of the goods and the vibration frequency band distribution, determine the combination of shelves and goods with impact risk, and obtain the classification adaptation factor.

[0080] Based on the shelf units marked as having a high or drastically changing ratio in the space volume ratio index, continuous data collected by the vibration recorders attached to these shelves during the corresponding work cycle is retrieved. The vibration amplitude and frequency distribution curves for each event, such as goods handling, AGV passage, and manual vibration, are extracted from the recorders. The frequency range of each vibration event is read and divided into intervals. Then, the impact resistance structural parameters marked for the product are extracted from the product characteristic parameter set. For example, there are three categories: high impact resistance products, ordinary impact resistance products, and low impact resistance products. High impact resistance corresponds to a vibration frequency range of 20 to 80 Hz, ordinary is 20 to 50 Hz, and low is 20 to 30 Hz. When the frequency in the vibration recording data... If most of the vibration frequency range is concentrated in the high-frequency range, such as 60 to 90 Hz, and the product is of low shock resistance type, it is judged as a high-risk combination. When two or more combinations are in the risk zone in the same shelf unit, the shelf unit is marked as a risk structure as a whole. Then, the numbers of each risk combination are filled back into the space volume ratio index to form a vibration impact cross-record table. For example, if product A is of low shock resistance type and is assigned to shelf Y, and Y has a vibration spectrum with a main frequency range of 60 to 85 Hz within three days, and there are 5 events with vibration duration exceeding 3 minutes, this combination is confirmed to have shock risk. By traversing all combinations with such overlapping ranges, the shelf and product combinations with shock risk are screened out and the output is the classification adaptation factor.

[0081] like Figure 4 As shown, the specific steps for obtaining the path priority metric are as follows:

[0082] S301: Based on the AGV navigation path, calculate the driving path relationship from the product sorting point to each shelf, determine the changes in the coordinates of the path nodes during continuous movement, filter out sections with dense turning or concentrated structural changes in the path, and obtain the path change feature group.

[0083] First, all AGV planning path data within the warehouse is acquired. The starting point of each path is defined as the goods sorting station, and the ending point is the location of the corresponding shelf number. Path information is displayed as a continuous sequence of coordinate points. The change in coordinates of each pair of adjacent nodes is calculated to determine the change in movement direction and displacement distance. Then, the coordinate points in the path are analyzed sequentially to identify node sequences with frequent direction changes. A continuous node segment with an angle difference greater than 30 degrees between adjacent segments is defined as a turn. If the number of turns exceeds 3 within a 20-meter path, the segment is marked as a high-turning area. Further, the warehouse traversed by each path is extracted. Structural area type: Based on the structural attributes of different areas in the warehouse, such as aisle width, intersection density, number of obstacles, etc., the structural complexity is scored. If the structural score of the area traversed by a path segment exceeds the set benchmark value of 5 points, it is recorded as a path segment with concentrated structural changes. When dense turning and overlapping structural changes occur, the path segment is preferentially marked as a high-variable path segment. For example, an AGV path from the sorting point to shelf B requires 4 right-angle turns and crosses two intersections, with a total path length of 28 meters. It is judged to have obvious path changes and is included in the path change feature group to form a dataset of change feature groups for all shelf target paths.

[0084] S302: Based on the path change feature group, call the outbound frequency record of the corresponding shelf goods in the sales system, compare the frequency change status in each period, judge the changing trend of the goods circulation speed, and group and classify the goods according to the consistency of the trend to obtain the circulation trend distribution group.

[0085] The system retrieves product outbound frequency information recorded in the sales system, dividing the timeframe into daily periods. A comparison period of 7 days is set. Outbound records for each product within consecutive periods are compiled into a sequence, and the periodic increase or decrease is calculated. The frequency difference between periods is compared to see if it exceeds a set threshold. For example, an increase exceeding 15% of the previous period is considered an active upward trend, while a decrease exceeding 20% ​​is considered a declining trend. If the trend direction remains consistent across three consecutive periods, the product is classified as a unidirectional trend. If the trend shows an initial increase followed by a decrease or fluctuation across different periods, the product is classified as a volatile product. The classification results are then associated with the shelf numbers in the path feature groups to determine whether the products within the same path segment have a consistent trend. For example, if a path segment is associated with three shelves and the outbound frequency of the corresponding products has shown a stable upward trend in the past four periods, with an increase of 12%, 14%, 17%, and 15%, then the products are classified into the same circulation growth group. If the products in another path segment show an increase of 13%, a decrease of 9%, an increase of 15%, and a decrease of 12%, respectively, then they are classified into the circulation trend fluctuation group. This rule is followed to process the product combinations corresponding to all path segments in turn, and the product circulation trend distribution group is integrated.

[0086] S303: Based on the distribution trend group, analyze the relationship between the path distance and the frequency of outbound shipments of goods, compare the correlation between the distance changes within each path segment and the level of circulation activity, determine the matching order between the path and the circulation of goods, and arrange all shelf path nodes in order to obtain the path priority index.

[0087] To compare the correlation between distance changes within each path segment and the level of circulation activity, the following formula is used:

[0088] ;

[0089] Calculate the path flow matching index value, determine the matching order between the path and the commodity flow, and rank all shelf path nodes sequentially to obtain the path priority index; among which... Representing the The path flow matching index value of each path. Representative path Upper The frequency of outbound shipments for a single product within the statistical period. The representative will be the path Upper The path distance of each node after mapping. Representative path The path distance mapping value of the previous adjacent node is used to construct the distance change of consecutive path segments. Representative path Upper The circulation weight coefficient corresponding to each commodity Representative path The number of path nodes participating in the calculation of the path flow matching index.

[0090] The path flow matching index value refers to the value of a specific path. In this context, the "deviation" between the outbound frequency of each node (and its associated goods) on the path and the path distance to that node (after unified mapping and smoothing) is weighted and calculated. The weighted deviation values ​​of all nodes are then summed and normalized to obtain a single numerical result. This indicator reflects the overall numerical matching degree between the actual circulation frequency of goods on the path and their spatial location within the path. If, on a certain path, high-circulation goods are mostly distributed at nodes with shorter distances, and low-circulation goods are distributed at nodes with longer distances, then the path's... A lower value indicates a high degree of matching between circulation and distance; a large deviation (such as highly circulating goods being located at long-distance nodes) indicates a lower value. An increasing value indicates a lower matching degree; the smaller the value, the stronger the "fit" between the frequency of goods leaving the warehouse at each node in the path and the spatial distribution of the path, and the smaller the deviation; the larger the value, the more mismatches there are between the circulation frequency and spatial distance on the path, which is used for subsequent ranking and comparison of different path priorities.

[0091] Extract the dataset of the node with path number 1, and set the number of nodes. For each of the three consecutive path nodes, numbered Node 1, Node 2, and Node 3, the outbound frequency value of the corresponding product within the statistical period is obtained from the warehouse scheduling system. , , Then call the original distance data of the path segment recorded in the AGV path management module. , , Since path distance and outbound frequency are on different dimensions, linear minimum-maximum normalization is used to map them to intervals respectively. The normalized frequencies are as follows:

[0092] , , ;

[0093] The normalized path distance is:

[0094] , , ;

[0095] Then, the mean distance of continuous path segments for nodes 2 and 3 is calculated separately, where the mean smoothed distance of node 2 is... Node 3 is Further square root construction of path smoothing index:

[0096] The value corresponding to node 2 is ;

[0097] Node 3 is ;

[0098] Perform a difference operation on the frequency values ​​corresponding to each node and take the absolute value:

[0099] The deviation at node 2 is ;

[0100] Node 3 deviation is ;

[0101] Weighting coefficients are set based on the percentage of outbound goods corresponding to each node. , , The deviation value of each node is weighted and calculated together with its weight.

[0102] Node 2 is ;

[0103] Node 3 is ;

[0104] Node 1, as the starting node of the path, does not meet the conditions for path segment calculation, so its deviation is counted as 0. Finally, the sum of all weighted deviation values ​​is divided by the sum of the weights. The path flow matching index value for path 1 is calculated as follows:

[0105] ;

[0106] This result indicates the path flow matching index value for path 1. Falling within the assessment range Within this range, there is a corresponding "medium deviation matching segment". This segment is used to indicate that there is a certain difference between the frequency of goods leaving the warehouse and the spatial distribution of the path, but it is still within the allowable adaptation range.

[0107] The preset criteria for dividing the complete interval are:

[0108] when When the time is "highly matched segment", it means that high-frequency goods on the path are concentrated in the shorter path position and low-frequency goods are distributed in the more distant nodes, with minimal deviation;

[0109] when At this point, it is considered a "moderate deviation matching segment," indicating that there is a certain degree of flow-space mismatch in the path, but the overall situation is within an acceptable range.

[0110] when When the path distribution is in a "low matching segment", it indicates a significant deviation between the path distribution and the circulation activity, and the path allocation scheme needs to be adjusted first.

[0111] The results of this calculation are directly related to the path matching judgment task and can serve as a quantitative basis for "judging the matching order between paths and commodity circulation". Subsequently, all path numbers will be ranked accordingly. Arrange them in ascending order to form a path priority index sequence.

[0112] like Figure 5 As shown, the specific steps for obtaining the categorical partition sequence are as follows:

[0113] S401: Based on the classification adaptation factor and path priority index, the shelf number is retrieved, the load monitoring status, spatial parameters and path sorting identifier of each shelf are compared, the correspondence between each set of shelf parameters and product attributes is determined, and the shelf code that meets all the matching rules in terms of load capacity, spatial configuration and path order is identified to obtain the candidate shelf number set.

[0114] Extract the spatial adaptation parameters, load-bearing adaptation parameters, and anti-interference adaptation parameters for each product, and compare them item by item with the load-bearing monitoring data recorded under all shelf numbers. Extract the maximum load-bearing value and remaining load capacity of each shelf layer. Then, directly compare the product weight with the remaining load capacity. When the product weight is less than the remaining load capacity and the proportion does not exceed 90%, it is determined to be a load-bearing match. Then, read the shelf layer height and width, and perform a three-way matching with the product's length, width, and height. If the dimension in any three-dimensional space exceeds the shelf's capacity in the corresponding direction, it is recorded as a mismatch group. For example, if the product dimensions are 400 mm long, 300 mm wide, and 250 mm high, the shelf layer dimension to be matched is 420 mm long. If the dimensions are 350 mm wide and 240 mm high, insufficient height is considered a spatial mismatch. Simultaneously, the path priority index records the path sorting identifier to obtain the sorting number of the shelf in the AGV arrival order. If the shelf sorting value is within the top 30%, it is marked as a path priority match. Then, the three judgment items, namely load capacity, space, and path, are assigned values ​​of 1 (match) or 0 (mismatch). Only when all three items are 1, the shelf number is judged as a complete match group. For example, if shelf A has a remaining load capacity of 20 kg and the product weight is 15 kg, it is judged as a load capacity match. The shelf height is 260 mm, which is a height match. The path sorting is 12th, accounting for the top 20%, which is a path match. Shelf A is included in the candidate shelf number set.

[0115] S402: Based on the candidate shelf number set, compare the adaptability of shelf load status, shelf height conditions and product parameters, determine the matching result of each shelf for the same product, group and classify each shelf code according to the degree of adaptability, and obtain the shelf adaptation order index;

[0116] The system retrieves the current real-time load-bearing status of each shelf and compares the difference between the product weight and the shelf's remaining load capacity. It then categorizes the load-bearing capacity into three levels: Excellent (remaining weight ≥ product weight × 2), Medium (remaining weight ≥ product weight × 1.2), and Low (remaining weight ≥ product weight but less than × 1.2). Next, it reads the shelf height information and compares the difference between the product height and the shelf's net height. If the shelf height is more than 50 mm greater than the product height, it's marked as Excellent; between 30 and 50 mm, Medium; and less than 30 mm, Low. Further values ​​are assigned based on a combination of load-bearing capacity and space level. For example, the load-bearing capacity... The overall level is A (Superior + Superior Space), B (Medium + Superior Space), C (Medium + Medium Space), and so on, dividing the system into six levels. These levels are then sorted from A to F. If multiple shelf numbers fall into the A category, the priority is further subdivided according to path order. For example, if three A-level shelf numbers are H01, H02, and H03, and their corresponding path order positions are 3rd, 6th, and 9th, then the priority order is H01 > H02 > H03. In this way, all candidate shelves are grouped and categorized to form a matching sequence corresponding to the appropriate level, thus creating a shelf adaptation order index.

[0117] S403: Based on the shelf adaptation order index, analyze the path sorting information, compare the changes in the order of the shelf arrival path, determine the combination of the shelf in the path order and adaptation sorting, and arrange all shelf codes in sequence to obtain the classification partition sequence.

[0118] The system retrieves the arrival path number of each shelf recorded in the path priority index and compares it with the ranking value of the path number in the AGV task scheduling table to form a path order list. Then, it combines the adaptation level and the path ranking value to score them. The adaptation level is assigned 6 to 1 points from A to F, the path ranking in the top 10% is assigned 3 points, 10%-30% is assigned 2 points, 30%-60% is assigned 1 point, and the rest are assigned 0 points. The two scores are added together to obtain the comprehensive adaptation score of each shelf. Then, all shelves are sorted from high to low according to the total score. If there are shelf numbers with the same score, the number with the higher path ranking is given priority. For example, shelf number H12 has an adaptation level of B with a score of 5, and is the 4th position in the path ranking with a score of 3, for a total score of 8. Shelf number H15 has an adaptation level of A with a score of 6, and is the 18th position in the path ranking with a score of 2, for a total score of 8. In this case, H12 is selected as the first priority, followed by H15. All shelf numbers are processed in turn to obtain the sorted shelf code arrangement, forming a classification and partitioning sequence.

[0119] like Figure 6 As shown, the specific steps for obtaining the classification job offset features are as follows:

[0120] S501: Based on the classification and partitioning sequence, compare the energy consumption data of AGV in the process of transporting to the shelf, determine the energy change trend on each transport path, identify the batches in which the energy consumption status fluctuates, and obtain the transport energy consumption change trajectory.

[0121] The system retrieves energy consumption logs recorded on each transport path, extracts the total energy consumption of each batch from its starting point to the target shelf, and records the corresponding task number and timestamp. These are then sorted by transport batch number to form a sequence. The energy consumption values ​​are then sequentially plotted as a time curve to assess continuous changes in energy consumption. If the energy consumption of any batch exceeds a set threshold compared to adjacent batches within three consecutive batches, it is considered an energy fluctuation. This threshold is set at 15% based on the standard fluctuation range of the vehicle's average energy consumption. For example, if the average energy consumption per batch on a normal path is 480 joules, and a batch reaches 560 joules, the fluctuation exceeds 80 joules, exceeding the 15% threshold, and the batch is considered a fluctuating batch. Its task number and time point are further recorded. The total number of fluctuating batches for each path is aggregated by path number. If a path exhibits 4 fluctuations out of 10 transport batches, it is marked as a high-fluctuation path. All fluctuating batches are categorized with their corresponding path numbers to generate a transport energy consumption change trajectory.

[0122] S502: Based on the energy consumption change trajectory of the handling process, analyze the operation node time of the corresponding batch, compare the difference between the operation progress and the planned progress, determine the progress deviation of each stage of the operation, filter the time segments where the progress is out of sync, and obtain the operation cycle offset index.

[0123] Extract the full-process operation time node data of the corresponding batch during task execution, including AGV start time, arrival time at the sorting point, loading completion time, start of handling time, arrival time at the shelf, and completion of shelving time. Compare the actual operation cycle sequence formed by these time points with the set standard operation cycle plan data, calculate the actual time taken for each stage, and perform a difference calculation with the planned standard time. If the actual time of any stage deviates from the planned time by more than 10%, it is marked as a progress deviation stage. The deviation threshold is set based on the average tolerance, generally between 5% and 15%. The median value of 10% is selected as the judgment threshold. For example, if the planned loading time is 60 seconds and a batch is recorded as 70 seconds, the offset is 16.7%, which exceeds the threshold and is recorded as an offset node. Then, the time period of continuous offset is identified by sorting by time. If two consecutive phases of offset occur continuously and the total time exceeds 15% of the original plan, it is marked as an asynchronous operation segment. For example, if a batch of handling takes 380 seconds and the planned time is 330 seconds, the offset exceeds 50 seconds, which is 15.2%. The batch number, total offset seconds, and affected links are recorded and included in the operation cycle offset index.

[0124] S503: Based on the work cycle offset index, analyze the handling process parameters, compare the changing trends of parameter collection data and process standard data, determine the fluctuations or offset phenomena in the classification and handling links, identify abnormal batches for summarization and sorting, and obtain the classification operation offset characteristics.

[0125] Real-time parameter collection records during the handling process are retrieved, including vehicle speed, acceleration, cargo status monitoring data (such as center of gravity shift and load stability), and surrounding environmental interference indicators (such as the number of obstacle avoidance triggers). Each data point is compared with the standard values ​​of each parameter in the process standard database for difference judgment. According to the set fluctuation judgment rules, the data is filtered. When a parameter value deviates from the standard value by more than the set upper and lower limits three times in a row, it is marked as a fluctuation. For example, if the average speed of a batch of AGVs during the handling process is 1.2 m / s, while the standard is 1.6 m / s, and there are 4 obstacle avoidance behaviors during the process, while the standard is no more than 1, it is judged as a serious interference fluctuation. Then, combined with the speed anomalies and load change data in the path segment, it is analyzed whether there is a mechanical performance imbalance. If multiple parameters deviate at the same time and are concentrated in the same handling task, the batch number is recorded as an abnormal batch. Then, the offset type of all abnormal batches is classified, such as "path interference type", "load instability type", "node delay type", etc. The batch number, offset parameter and type identifier of all offset batches are summarized to form a classified operation offset feature.

[0126] like Figure 7 As shown, a shelf item classification system based on product attributes includes:

[0127] The attribute acquisition module is based on a laser rangefinder to collect the three-dimensional dimensions of the product, and combines the weight data of the electronic scale to identify abnormal fluctuations. It also uses a barcode scanner to input packaging material information, verifies consistency with the standard catalog, and obtains a set of product feature parameters.

[0128] The adaptation analysis module, based on the product feature parameter set, determines the compatibility between the product and the shelf load-bearing capacity, shelf height, and shock resistance. It analyzes the load trend of the weighing sensor monitoring data and the product weight, compares the product volume with the shelf space capacity, screens the matching of the shelf's disturbance resistance performance and impact resistance performance, and obtains the classification adaptation factor.

[0129] The path sorting module calculates the distance from the sorting point to the shelf based on the AGV navigation path, analyzes the changes in path nodes, combines the outbound frequency to determine the trend of commodity circulation speed, and arranges the priority of the storage locations in order to obtain the path priority index.

[0130] The partitioning decision module uses classification adaptation factors and path priority indicators to filter eligible shelf numbers, compare the adaptability of each shelf to product attributes, analyze arrival path efficiency, determine the allocation sequence, and obtain the classification partitioning sequence.

[0131] The status monitoring module adjusts the preferred shelf arrangement based on the classification and partition sequence, analyzes the energy consumption changes collected during AGV handling, judges the difference between the operation progress and the planned progress, screens for classification data offset phenomena, and obtains the classification operation offset characteristics.

[0132] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A shelf item classification method based on product attributes, characterized in that, The method includes: S1: Based on a laser rangefinder, collect the three-dimensional dimensions of the product, combine the weight data of the electronic scale to identify abnormal fluctuations, enter the packaging material information through a barcode scanner, verify the consistency with the standard catalog, and obtain the product feature parameter set; S2: Based on the set of product feature parameters, determine the compatibility between the product and the shelf load-bearing capacity, shelf height and shock resistance, analyze the load trend of the weighing sensor monitoring data and the product weight, compare the product volume with the shelf space capacity, screen the matching of shelf disturbance resistance performance and impact resistance performance, and obtain the classification adaptation factor. S3: Based on the AGV navigation path, calculate the distance from the sorting point to the shelf, analyze changes in path nodes, combine with outbound frequency to determine the trend of commodity circulation speed, and prioritize storage locations in order to obtain path priority indicators; including: S301: Based on the AGV navigation path, calculate the driving path relationship from the product sorting point to each shelf, determine the changes in the coordinates of the path nodes during continuous movement, and filter out sections with dense turning or concentrated structural changes in the path to obtain path change feature groups; the path nodes are the turning points, stations or task nodes traversed by the path. S302: Based on the path change feature group, call the outbound frequency record of the corresponding shelf goods in the sales system, compare the frequency change status in each period, determine the changing trend of the goods circulation speed, and group and classify the goods according to the trend consistency to obtain the circulation trend distribution group. S303: Based on the aforementioned distribution trend group, analyze the relationship between the path distance and outbound frequency of the goods, compare the correlation between the distance change and the level of circulation activity within each path segment, determine the matching order between the path and the circulation of goods, and arrange the paths in order to obtain the path priority index; this index reflects the overall numerical matching degree between the actual circulation frequency of goods on the path and its spatial location on the path. If, on a certain path, high-circulation goods are mostly distributed at shorter distance nodes and low-circulation goods are distributed at longer distance nodes, it indicates that circulation and distance are highly matched; otherwise, it indicates a low matching degree. S4: Based on the classification adaptation factor and path priority index, select the shelf numbers that meet the conditions, compare the matching of each shelf with the product attributes, analyze the arrival path efficiency, determine the allocation sequence, and obtain the classification partition sequence. S5: Based on the classification and partitioning sequence, adjust the preferred shelf arrangement, analyze the energy consumption changes collected during AGV handling, determine the difference between the operation progress and the planned progress, screen for classification data offset phenomena, and obtain classification operation offset characteristics.

2. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The commodity feature parameter set includes size attributes, quality attributes, material attributes, and circulation attributes; the classification adaptation factor includes spatial adaptation parameters, carrying capacity adaptation parameters, and anti-disturbance adaptation parameters; the path priority index includes path distance parameters, circulation frequency parameters, and priority ranking parameters; the classification partition sequence includes partition number, allocation order, and scheduling identifier; and the classification operation offset features include energy consumption features, progress features, and path offset features.

3. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The specific steps for obtaining the product feature parameter set are as follows: S101: Based on a laser rangefinder, analyze the acquired spatial coordinate data of the product surface, fit the boundary of the product point cloud shape model, compare the three-dimensional range of the boundary with the standard three-dimensional size parameter set of the product, determine the abnormal index position of the spatial boundary, identify the points that differ from the standard size, and obtain the three-dimensional size deviation set. S102: Analyze the weighing data of goods collected by the electronic scale, compare the continuous weighing collection sequence with the standard quality parameters of the goods, determine the quality change of the same goods in a continuous weighing process, identify the time period of continuous fluctuation, and obtain the set of quality fluctuation intervals. S103: Based on the quality fluctuation interval set and the three-dimensional size deviation set, compare the packaging material label collected by the barcode scanner with the standard packaging catalog template, determine the packaging material discrepancies, and at the same time analyze the periodic trend of the corresponding product sales system's outbound frequency, identify the codes of abnormal fluctuation amplitude, and obtain the product feature parameter set.

4. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The specific steps for obtaining the classification adaptation factor are as follows: S201: Based on the product feature parameter set, analyze the quality and size information, compare the product weight with the shelf load monitoring data, determine whether the load capacity matches, analyze the spatial compatibility between the shelf height and the three dimensions of the product, identify combinations with limited space or incompatible load, and obtain shelf structure adaptation data. S202: Based on the shelf structure adaptation data, compare the product volume data with the remaining shelf volume information, calculate the ratio of product volume in the shelf space, determine the ratio change in each shelf unit, identify the combination of ratio changes, and obtain the space volume ratio index. S203: Based on the space volume ratio index, call the continuous operation cycle shelf vibration spectrum data collected by the vibration recorder, compare the degree of overlap between the impact resistance structural parameters of the goods and the vibration frequency band distribution, determine the combination of shelves and goods with impact risk, and obtain the classification adaptation factor.

5. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The specific steps for obtaining the classification partition sequence are as follows: S401: Based on the classification adaptation factor and path priority index, the shelf number is retrieved, the load monitoring status, spatial parameters and path sorting identifier of each shelf are compared, the correspondence between each set of shelf parameters and product attributes is determined, and the shelf code that fully meets the matching rules in terms of load capacity, spatial configuration and path order is identified to obtain a set of candidate shelf numbers. S402: Based on the candidate shelf number set, compare the matching degree between the shelf load status, shelf height conditions and product parameters, determine the matching result of each shelf for the same product, group and classify each shelf code according to the degree of matching, and obtain the shelf adaptation order index. S403: Based on the shelf adaptation order index, analyze the path sorting information, compare the changes in the order of the shelf arrival path, determine the combination of the shelf in the path order and shelf adaptation order, and arrange the shelf codes in sequence to obtain the classification and partitioning sequence.

6. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The specific steps for obtaining the classification job offset features are as follows: S501: Based on the classification and partitioning sequence, compare the energy consumption data during the AGV transport to the shelf, determine the energy change trend on each transport path, identify batches where energy consumption fluctuates, and obtain the transport energy consumption change trajectory. S502: Based on the energy consumption change trajectory of the handling, analyze the operation node time of the corresponding batch, compare the difference between the operation progress and the planned progress, determine the progress deviation of each stage of the operation, filter the time segments in which the progress is out of sync, and obtain the operation cycle offset index. S503: Based on the operation cycle offset index, analyze the handling process parameters, compare the changing trends of the parameter collection data and the process standard data, determine the fluctuations or offset phenomena in the classification and handling process, identify abnormal batches and summarize them to obtain the classification operation offset characteristics.

7. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The abnormal fluctuation refers to the fluctuation of the collected commodity weight data that exceeds the reasonable range. The standard catalog is a list of packaging materials and specifications preset in the warehousing system.

8. The shelf item classification method based on commodity attributes according to claim 1, characterized in that, The load trend is recorded in real time by weighing sensors, and compared with the weight of the goods to be sorted and put on the shelves. The load on the shelves is monitored over time or during the shelving process. The shelf disturbance resistance refers to the stability and deformation resistance of the shelf structure under the influence of external forces and vibrations.

9. A shelf item classification system based on product attributes, the system being used to implement the shelf item classification method based on product attributes as described in any one of claims 1-8, characterized in that, The system includes: The attribute acquisition module is based on a laser rangefinder to collect the three-dimensional dimensions of the product, and combines the weight data of the electronic scale to identify abnormal fluctuations. It also uses a barcode scanner to input packaging material information, verifies consistency with the standard catalog, and obtains a set of product feature parameters. Based on the product feature parameter set, the adaptation analysis module determines the compatibility between the product and the shelf load-bearing capacity, shelf height, and shock resistance. It analyzes the load trend of the weighing sensor monitoring data and the product weight, compares the product volume with the shelf space capacity, screens the matching of the shelf's anti-interference performance and impact resistance, and obtains the classification adaptation factor. The path sorting module calculates the distance from the sorting point to the shelf based on the AGV navigation path, analyzes the changes in path nodes, combines the outbound frequency to determine the trend of commodity circulation speed, and arranges the priority of the storage locations in order to obtain the path priority index. The partitioning decision module, based on the classification adaptation factor and path priority index, filters the shelf numbers that meet the conditions, compares the matching of each shelf with the product attributes, analyzes the arrival path efficiency, determines the allocation sequence, and obtains the classification partitioning sequence. Based on the classification and partitioning sequence, the status monitoring module adjusts the preferred shelf arrangement, analyzes the energy consumption changes collected during AGV handling, determines the difference between the operation progress and the planned progress, screens for classification data offset phenomena, and obtains classification operation offset characteristics.

Images