A Visualization Method for the Goods Turnover Model Oriented to Digital Warehousing

Abstract

The present invention discloses a visualization method for the goods turnover mode facing digital warehousing. It obtains the goods-related data of the digital warehousing system, retains the valid information after data cleaning, preprocessing and frequent pattern mining; designs a goods attribute screening view and a goods outbound mode view, and performs visual mapping using visual channels such as position, shape, color, etc.; in the goods attribute screening view, the parallel coordinate rectangle is used to map the distribution of the turnover attribute values of the goods in the specified year and month; in the goods outbound mode view, the nested rectangle is used to map the proportion of the goods outbound times, and the pattern unit is used to map the outbound frequent pattern and picking time of the selected goods. Combining the rich interactions within the views and the linkage between the views, the present invention can assist warehouse operators in exploring the distribution of the turnover attributes of goods during the warehouse operation process, screening the goods of interest, and analyzing the frequent patterns of goods outbound, so as to better make decisions and optimize the warehouse operation efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of information visualization and visual analysis, and specifically provides a visualization method for the goods turnover mode in a digital warehouse. Background Art

[0002] With the booming development of the Internet e-commerce economy, the operation efficiency of traditional warehouses can no longer meet people's needs, and digital warehouses have emerged as the times require. Digital warehouses use Internet of Things and big data as core technologies to automatically collect data on various links such as the inbound and outbound of goods. These data contain rich information such as goods characteristics and warehouse operation performance. Exploring the hidden patterns therein can assist in supervising the warehouse operation status and guiding the allocation of inbound storage locations. However, due to the characteristics of large scale, high dimension, and time series of digital warehouse data, it is difficult for analysts to mine and understand the valuable information contained therein. Therefore, how to analyze and understand the system operation status and goods turnover rules based on digital warehouse data, explore the hidden patterns therefrom to optimize decisions, and thereby improve the system operation efficiency has become the focus issue of concern to digital warehouse operation and management personnel.

[0003] Most of the existing methods for improving the operation efficiency of digital warehouses are based on machine learning and data mining. However, such automated methods do not integrate the user's cognitive and reasoning abilities into the analysis process, and the output results are often obscure and difficult to understand, lacking good interpretability. Information visualization and visual analysis methods provide new solutions for warehouse operation and management personnel. They can use intuitive visual presentations and flexible interactions between views to explore the rules and patterns behind the data. Information visualization methods convert data into perceivable graphics, symbols, colors, etc. to enhance people's recognition efficiency of data and convey effective information. At the same time, visual analysis methods can combine machine intelligence with human wisdom, and introduce human knowledge and experience into the analysis process through an interactive interface that integrates multiple views, thereby assisting warehouse operation and management personnel to comprehensively and intuitively complete data analysis and reasoning decisions.

[0004] Specifically, in the operation of a digital warehouse system, in order to analyze the inbound and outbound turnover rules of goods, warehouse operation and management personnel need to analyze the turnover performance of goods and the distribution of their attribute values, screen the goods of interest, and also need to analyze the frequent patterns of goods outbound, so as to provide insights for the discovery of abnormal goods and the adjustment of storage locations. However, the existing research on warehouse data visualization mainly focuses on operation process analysis and warehouse scenario simulation, lacking the ability to interactively explore and analyze the goods turnover mode in the warehouse system. The existing event sequence visualization work also rarely focuses on warehouse turnover events, and it is difficult for general methods to reflect the impact of characteristics such as the combination method of goods on the event sequence, and it is difficult to accurately reflect the operation status of the warehouse system and intuitively discover abnormal outbound patterns of goods. Summary of the Invention

[0005] To address the above problems, the objective of the present invention is to provide a visualization method for the goods turnover mode in digital warehousing. Based on various goods-related data in digital warehousing, an interactive goods attribute screening view and a goods outbound mode view are designed and implemented. The present invention applies visual analytics to the research of warehousing turnover events, which can not only display the distribution of various attributes of goods during the warehousing operation process, but also screen the goods of interest and further analyze their frequent outbound modes. Warehouse operation managers can obtain effective insights and cognitions from this invention by combining their own domain knowledge and experience, such as discovering abnormal goods and unreasonable storage locations, and then better make decisions to optimize the warehousing operation efficiency. The specific technical solutions are as follows:

[0006] A visualization method for the goods turnover mode in digital warehousing, comprising the following steps:

[0007] S1: Data acquisition and processing

[0008] Obtain digital warehousing data. After data cleaning, data preprocessing, and frequent pattern mining on the original data, extract and retain the valid information, and store the goods turnover performance data and the goods outbound mode data in the database;

[0009] S2: Visual mapping

[0010] Perform visual mapping on the data obtained in step S1 through visual channels:

[0011] Design a goods attribute screening view, use a parallel coordinate rectangle to map the distribution of each turnover attribute value of the goods in a specified year and month, and map the goods and their attribute values with line segments inside the rectangle; in addition, display the screened goods with a goods screening list;

[0012] Design a goods outbound mode view, use a three-layer nested rectangle to map the proportion of the outbound times of the selected goods in their respective owners and their respective warehouses; use a pattern unit to map the frequent patterns of the selected goods co-outbound with other goods, the occurrence frequency of this frequent pattern, as well as the picking time of the corresponding goods and the average picking time of this frequent pattern;

[0013] S3: Visual layout and implementation

[0014] Perform visual layout and implementation on the visual modules that have completed mapping in S2:

[0015] In the goods attribute screening view, the parallel coordinate rectangles on the left are arranged in sequence from top to bottom, and the line segments are arranged from left to right inside the rectangles according to the size of the attribute values of the goods, and the vertical line segments representing the same goods in each rectangle are connected; the goods screening list on the right includes a general selection list above and a selected list below;

[0016] In the product delivery mode view, the left side is the nested rectangle corresponding to the product, and the right side is the n most frequently appearing pattern units in the frequent delivery mode corresponding to the product. The n rows are arranged from top to bottom for comparison;

[0017] S4: Interaction Design

[0018] In the product attribute filtering view, after brushing a line segment within any range in any rectangle on the left, the product lines that meet all the brushing conditions will be highlighted, and the general selection list on the right will also display all the selected product information. After clicking on the product of interest from the general selection list and adding it to the selected list, the product delivery mode view will be updated in conjunction; after clicking the square of the mode unit in the product delivery mode view, the corresponding product will be added to the last row of the view, and the selected list in the product attribute filtering view will be updated in conjunction; in addition, when the mouse is hovered over the elements in the two views, there will be an information floating box to display detailed information.

[0019] Furthermore, in step S1, the specific process of data acquisition and processing is as follows:

[0020] S11: Obtain three types of data, including inbound orders, outbound orders, and daily inventory of goods from the digital warehouse management platform. The attributes of the inbound / outbound order data include: order ID, consignor ID, product ID, timestamps of various inbound / outbound operations, and quantity, volume, and quality of the goods; outbound order data also includes outbound batch ID; the attributes of the daily inventory data include: inventory record ID, date, product ID, consignor ID, shelf ID, and inventory quantity;

[0021] S12: Screen and delete the invalid parts of the original goods inbound order data and outbound order data; for the missing parts of the goods inventory data, calculate, extract and complete them based on other existing information; convert and integrate the data to extract goods turnover performance data;

[0022] S13: Construct a transaction data set to record the types of goods included in each outbound batch; mine the frequent outbound goods combinations of each product, construct corresponding new transaction data for each product, use the FP-Growth algorithm to perform frequent pattern mining on the goods combinations in the outbound batch, and generate outbound pattern data containing the goods;

[0023] S14: Design a database model based on data characteristics, and store the goods turnover performance data and goods outbound model data in the MySQL database.

[0024] Furthermore, the step of constructing the transaction data set in S13 specifically includes:

[0025] S131: Group the outbound order data by shipper ID based on the goods turnover performance data.

[0026] S132: For each order group belonging to a shipper, group it by outbound batch ID.

[0027] S133: For each order group belonging to an outbound batch, remove duplicates from all goods IDs according to the goods information included in the outbound order to obtain a combination of goods IDs.

[0028] S134: Construct a goods outbound transaction data set based on the shipper ID, outbound batch ID, and combination of goods IDs.

[0029] Furthermore, the specific process of mining the frequently outbound goods combination for each type of goods described in S13 is as follows:

[0030] 1) For each type of goods, filter out the corresponding transaction items from the original transaction data set according to the corresponding shipper ID.

[0031] 2) Filter out the transaction items containing this type of goods according to the combination of goods IDs corresponding to the outbound batch, and then form a new transaction data set for this type of goods.

[0032] 3) Apply the frequent pattern mining algorithm FP-Growth to the new transaction data set of this type of goods to obtain the frequent patterns containing this type of goods. In the subsequent description, this type of goods will be referred to as the "source goods", and the goods other than this type of goods in the frequent pattern will be referred to as "other goods".

[0033] Furthermore, in step S2, the specific process of visual mapping is as follows:

[0034] S21: Perform shape and position mapping on the parallel coordinate rectangles of the goods attribute filtering view: Map the turnover attributes of the goods with rectangles and the goods with connection lines; The six rectangles map the six turnover attributes of the total inbound quantity, average receiving time, average shelving time, total outbound quantity, average picking time, and average packaging time of the goods in the selected month from top to bottom in sequence; Map the size of the corresponding attribute values of the goods with the position of the vertical line segments, increasing proportionally from left to right, and connect the same goods with diagonal lines.

[0035] S22: Perform shape and position mapping on the goods filtering list of the goods attribute filtering view: Map the operation of adding goods with a "+" shaped icon and the operation of deleting goods with a "×" shaped icon; Map different goods in the vertical direction of the list, and map the three attributes of "goods ID, shipper ID, operation" from left to right in sequence in the horizontal direction.

[0036] S23: Map the shape, position, and color of the nested rectangles in the view of the goods outbound mode: Use three nested rectangles to map the outbound times of goods at three levels; the innermost dark gray rectangle maps the outbound times of the selected goods in the current month, the middle gray rectangle maps the outbound times of all goods under the owner to which the goods belong, and the outermost light gray rectangle maps the total outbound times of all goods in the warehouse;

[0037] S24: Map the shape, position, color, radius, and arc length of the mode units in the view of the goods outbound mode: For the upper circular part, use the arc length of the outer arc to map the frequency of this frequent mode, and use the radii of the multiple inner rings to map the outbound picking times of each good in this combination; use the radius of the background circle to map the average outbound picking time of this combination; for the lower rectangular part, use the leftmost small rectangle to map the source good, and use the other small rectangles to map the other goods that are outbound together with the source good under this frequent mode; use two colors to map the source good and other goods respectively, and the corresponding rings and rectangles are kept consistent.

[0038] Furthermore, in step S3, the specific process of the visual layout and implementation of the parallel coordinate rectangles in the goods attribute screening view is as follows:

[0039] S3a: The parallel coordinate rectangles use a parallel coordinate layout with the X-axis perpendicular to the Y-axis. Based on the predefined boundary distances and the height and width of the rectangles, calculate the coordinates of the six rectangles respectively, and mark the attribute names in the upper left corner;

[0040] S3b: The length of the intersection point between the goods connection line and the parallel rectangle from the left boundary of the rectangle maps the magnitude of the turnover attribute value corresponding to the goods. Use a logarithmic scale to map the turnover attribute value of the goods and the length of the intersection point from the left boundary of the rectangle. The left boundary of the rectangle maps the minimum attribute value, and the right boundary maps the maximum attribute value;

[0041] S3c: Determine the positions of the six vertical line segments of the goods in the six rectangles and connect them with oblique lines.

[0042] Furthermore, in step S3, the specific process of the visual layout and implementation of the goods outbound mode view is as follows:

[0043] S31: In the nested rectangles on the left, predefined the heights of the three rectangles from the outside to the inside, decreasing in turn; predefined the width of the outermost rectangle, and then calculate the widths of the middle and inner rectangles according to the linear proportional relationship between the monthly warehouse outbound times, the owner's outbound times, and the outbound times of this good; use gradually changing shades of color to fill the rectangles from the outside to the inside;

[0044] S32: In the circular part of the right mode unit, the arc of the outer arc is calculated by the frequency of the mode and the number of times the source goods corresponding to the mode are shipped out of the warehouse, and then multiplied by the predefined maximum radius to obtain the arc length of the outer arc; the radius of the ring is calculated by the linear ratio of the goods picking time and the maximum picking time of the month, and the radius of the background circle is calculated by the linear ratio of the average picking time of the mode and the maximum picking time of the month;

[0045] S33: In the rectangular part of the pattern unit on the right, predefine the height and overall width of the rectangle, and divide the overall width by the number of goods included in the pattern to calculate the width of each rectangle; then use two colors to fill the rectangle to map the source goods and other goods respectively.

[0046] Furthermore, in step S4, the specific interaction design is as follows:

[0047] S41: The product attribute filtering view supports four mouse interactive operations: swiping, hovering, clicking and scrolling: each line segment in the rectangle can be swiped, and the selected line represents the products with the corresponding attribute in the range; when the mouse hovers over a product line or a row in the filtering list, the corresponding information floating box is displayed; click the "+" icon and "×" icon in the "Operation" column on the far right of the product filtering list to update the view in a linked manner; when there are many selected products, the list supports mouse scrolling;

[0048] S42: The goods outbound mode view supports two mouse interactive operations: floating and clicking. When the mouse is hovered over a nested rectangle, the corresponding information floating box is displayed. When the mouse is hovered over the outer arc, background circle or small rectangle of the pattern unit, the corresponding information floating box is displayed and the corresponding element is highlighted. Clicking the small rectangle representing the goods at the bottom of the pattern unit with the mouse can increase the frequent patterns to be explored and update the view in a linked manner.

[0049] The beneficial effects of the present invention are as follows: the present invention obtains the goods-related data of the digital warehousing system, performs data cleaning, preprocessing and frequent pattern mining, and retains effective information; designs a goods attribute screening view and a goods outbound pattern view, and uses visual channels such as position, shape, and color for visual mapping; in the goods attribute screening view, parallel coordinate rectangles are used to map the distribution of the turnover attribute values of goods in the specified year and month; in the goods outbound pattern view, nested rectangles are used to map the proportion of the number of goods outbound times, and pattern units are used to map the selected goods' outbound frequent patterns and picking time. Combined with the rich interactions within the view and the linkage between views, the present invention can assist warehouse operators in exploring the distribution of goods' turnover attributes during warehouse operations, screening goods of interest, and analyzing the frequent patterns of goods outbound, thereby making better decisions and optimizing warehouse operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1It is a schematic diagram of the overall process framework of the visualization method for the goods turnover mode in the digital warehouse of the present invention.

[0051] Figure 2 It is a schematic diagram of the overall layout and visual mapping of the goods attribute screening view in the present invention.

[0052] Figure 3 It is a schematic diagram of the visual layout calculation of the parallel coordinate rectangle in the present invention.

[0053] Figure 4 It is a schematic diagram of the overall layout of the goods outbound mode view in the present invention.

[0054] Figure 5 It is a schematic diagram of the visual mapping of nested rectangles and pattern units in the present invention; (a) Visual mapping of nested rectangles; (b) Visual mapping of pattern units.

[0055] Figure 6 It is a schematic diagram of the visual layout calculation of nested rectangles and pattern units in the present invention; (a) Layout calculation of nested rectangles; (b) and (c) Layout calculation of the circular part of the pattern unit; (d) Layout calculation of the rectangular part of the pattern unit.

[0056] Figure 7 It is a schematic diagram of the interaction design of the goods attribute screening view in the present invention.

[0057] Figure 8 It is a schematic diagram of the interaction design of the goods outbound mode view in the present invention; (a) Interaction design of nested rectangles; (b) Interaction design of pattern units. Detailed implementation manners

[0058] The following further elaborates the present invention in detail in combination with the attached drawings and specific data.

[0059] Through information visualization and visual analysis methods, combined with the multi-view linkage strategy and flexible interaction means, the present invention realizes multi-angle analysis of digital warehouse data, helps warehouse operation and management personnel intuitively explore the distribution of attribute values of goods turnover performance, find interested goods, and analyze the frequent patterns of goods outbound. The technical solutions include: data acquisition and processing, visual mapping, visual layout and implementation, and interaction design. The schematic diagram of the overall process framework of the visualization method for the goods turnover mode in the digital warehouse of the present invention is as Figure 1 shown, and the specific steps are as follows:

[0060] Step 1: Data acquisition and processing

[0061] Obtain goods-related data from the digital warehouse management platform, perform data cleaning, data preprocessing, and frequent pattern mining, and then store them in the database.

[0062] 1. Data Acquisition

[0063] Three types of data, namely inbound orders, outbound orders, and daily inventory of goods, are obtained from the digital warehousing management platform. The important attributes of inbound / outbound order data of goods are: order ID, shipper ID, goods ID, timestamps of various inbound / outbound operations, as well as the quantity, volume, and quality of the goods; in addition, the outbound order data also includes the outbound batch ID. The important attributes of daily inventory data of goods are: inventory record ID, date, goods ID, shipper ID, shelf ID, and inventory quantity.

[0064] 2. Data Cleaning and Preprocessing

[0065] In a real warehousing environment, missing or invalid data may occur due to omissions or incorrect operations by workers. This method will screen and complete the original data.

[0066] (1) Delete invalid data

[0067] In the inbound / outbound order data of goods, the operation sequence in some order records may not conform to the actual scenario. For example, in some inbound orders, the goods shelving time is earlier than the goods receiving time. To address this issue, this method takes the inbound process of "order creation - goods arrival - goods receipt - goods shelving" and the outbound process of "order creation - batch allocation - goods picking (order picking) - goods packaging" as standards to screen and delete incorrect and invalid data.

[0068] (2) Complete missing data

[0069] Regarding the problem of missing records for some goods in the goods inventory data, this method first calculates the daily inbound and outbound quantity of goods based on the inbound / outbound order data of goods, and then finds a valid record in the daily inventory data of goods. Taking the inventory quantity of this record as the benchmark and combining the daily inbound and outbound quantity of goods, the missing records are completed forward and backward.

[0070] (3) Extract goods turnover performance data

[0071] Based on the inbound / outbound order data of goods, count the goods with inbound and outbound activities each month, and calculate the total inbound quantity, average receiving time, average shelving time, total outbound quantity, average picking time, and average packaging time of each good in the current month to obtain the goods turnover performance data for analyzing and discovering the performance differences of each good in the warehousing system during the operation process. Since different shippers may have the same goods, the goods described in this method use "shipper ID - goods ID" as the primary key, that is, the same goods of different shippers are calculated and analyzed separately.

[0072] 3. Frequent Pattern Mining

[0073] Explore the frequently occurring combinations of goods in the outbound batches, calculate the average picking time of these combinations, and identify combinations of goods with abnormal performance in the picking operation, providing a decision-making basis for warehouse operators to adjust the shelf positions of goods.

[0074] This method uses a frequent pattern mining algorithm to obtain the high-frequency combinations of goods in the outbound batches. Before applying the algorithm, a transaction dataset needs to be constructed. Each outbound order in the warehouse data represents a shopping behavior of a user, and an outbound batch consists of multiple outbound orders. The records in the transaction dataset are the types of goods included in each outbound batch, and the construction method is as follows:

[0075] (1) Group the outbound order data by the owner ID based on the goods turnover performance data;

[0076] (2) For each order group belonging to an owner, group it by the outbound batch ID;

[0077] (3) For each order group belonging to an outbound batch, remove duplicates from all the goods IDs according to the goods information included in the outbound order to obtain a combination of goods IDs;

[0078] (4) Construct a goods outbound transaction dataset based on the owner ID, outbound batch ID, and combination of goods IDs.

[0079] Furthermore, to reduce the user's selection cost and improve the analysis effect, this method will mine the frequently occurring combinations of goods for each item, focusing the user's attention from all the frequent patterns in the warehouse system on the frequent patterns of a single item, and only mining the high-frequency outbound combinations containing that item. The specific method is as follows:

[0080] (1) For each item, filter out the corresponding transaction items from the original transaction dataset according to the corresponding owner ID;

[0081] (2) Filter out the transaction items containing that item according to the combination of goods IDs corresponding to the outbound batch, and then form a new transaction dataset for that item;

[0082] (3) Apply the frequent pattern mining algorithm FP-Growth in the new transaction dataset of that item to obtain the frequent patterns (combinations of goods) containing that item. Call that item the "source item", and the items other than that item in the frequent pattern are called "other items".

[0083] After executing the frequent pattern mining algorithm, the obtained data includes the following attributes: time (year and month), goods ID, owner ID, all the frequent patterns corresponding to the source item and their frequencies, that is, the number of times the combination of goods in the pattern is shipped out together in the selected month. In addition, calculate the total number of outbound times of the source item to analyze the proportion of different frequent patterns.

[0084] To further analyze the outbound time consumption of a certain frequent pattern, calculate the average picking time of the pattern based on the average picking time of each item in the frequent pattern, because the picking step in the outbound process of items has the greatest uncertainty and is the most time-consuming.

[0085] In addition, to analyze the proportion of the outbound times of source items in a month in their respective owners and warehouses, calculate the outbound times of the item in that month, the outbound times of all items under the corresponding owner, and the total outbound times of all items in the warehouse according to the item ID, owner ID, and outbound batch ID.

[0086] Furthermore, obtain the item outbound pattern data containing the above attributes.

[0087] 4. Data storage

[0088] After completing the above operations, store the item turnover performance data and the item outbound pattern data in a MySQL database for subsequent visualization.

[0089] Step 2: Visualization mapping

[0090] After data acquisition and processing, design a visualization mapping scheme for the item attribute screening view (as shown in Figure 2 ) and the item outbound pattern view (as shown in Figure 5 (a) and (b)) in the present invention.

[0091] 1. Item attribute screening view

[0092] (1) Parallel coordinate rectangle

[0093] Shape: Map the turnover attributes of items with rectangles and map items with connecting lines.

[0094] Position: Map different item attributes with parallel rectangles at different positions, and map the six turnover attributes of the total inbound quantity, average receiving time, average shelving time, total outbound quantity, average picking time, and average packaging time of the item in the selected month from top to bottom in sequence; map the magnitude of the item attribute values with the position of vertical line segments, increasing proportionally from left to right, and connect the same item with oblique lines.

[0095] (2) Item screening list

[0096] Shape: In the upper general selection list, map the operation of adding items with a "+" shaped icon; in the lower selected list, map the operation of deleting items with a "×" shaped icon.

[0097] Position: Map different items in the vertical direction of the list, and map the three attributes of "item ID, owner ID, operation" from left to right in the horizontal direction in sequence.

[0098] 2. Goods Outbound Mode View

[0099] (1) Nested Rectangles

[0100] Shape: Use three nested rectangles to map the outbound times of goods at three levels.

[0101] Position and Color: The innermost dark gray rectangle maps the outbound times of the selected goods in the current month. The middle gray rectangle maps the outbound times of all goods owned by the owner of this good. The outermost light gray rectangle maps the total outbound times of all goods in the warehouse.

[0102] (2) Mode Unit

[0103] Shape and Position: The upper circular part consists of an outer arc, an inner ring, and a background circle with the same center. The outer arc maps the frequency of this frequent mode. The multiple inner rings map the outbound picking times of each good in this combination. The background circle maps the average outbound picking time of this combination. The lower rectangular part consists of multiple small rectangles arranged horizontally. The leftmost small rectangle maps the source good of this frequent mode, and the other small rectangles map the other goods in this mode.

[0104] Color: Use colors to map the types of goods corresponding to the rings. The orange ring corresponds to the outbound picking time of the source good of this frequent mode, and the green ring corresponds to the outbound picking times of other goods in this frequent mode. Use a blue background circle to map the average outbound picking time of this combination of goods. In the rectangular part, use orange to map the source good and green to map other goods. For example, a frequent outbound mode of good A is "A, B, C". Good A is the source good of this mode and is mapped with orange; goods B and C are other goods in this mode and are mapped with green.

[0105] Radius: Use the radii of the rings and the circle to map the corresponding picking times. The longer the time, the larger the radius;

[0106] Arc Length: Use the arc length of the outer arc to map the frequency of the frequent mode. The larger the frequency, the longer the arc length. The arc length is calculated based on the ratio of the frequency of this mode to the total outbound times of the source good.

[0107] By comparing the picking time of a single item in the pattern unit with the average picking time of the frequent pattern, abnormal item combinations can be discovered. When the average picking time of the pattern is close to the picking time of each item when it is separately shipped out, the radius of the background circle will be close to the radii of the orange and green rings, indicating that the simultaneous shipment of the items in this frequent pattern does not significantly affect their picking time. When the average picking time of the pattern is significantly higher than the picking time of each item when it is separately shipped out, the radius of the background circle will be significantly larger than the radii of the orange and green rings, indicating that the simultaneous shipment of the items in this frequent pattern will prolong the picking time. Discovering such abnormal patterns can provide a reference for warehouse operators to adjust the storage locations.

[0108] Step Three: Visual Layout and Implementation

[0109] Read the warehouse item data for the specified year and month, and complete the layout and implementation of the left parallel coordinate rectangle and the right item screening list in the item attribute screening view; complete the layout and implementation of the left nested rectangle, the right pattern unit, and the upper legend in the item shipment pattern view, as follows:

[0110] 1. Visual Layout and Implementation of the Item Attribute Screening View

[0111] The overall layout of the item attribute screening view is as Figure 2 shown. The screening list on the right can be implemented using conventional methods, and the following focuses on the visual layout and implementation of the left parallel coordinate rectangle (as Figure 3 shown):

[0112] The parallel coordinate rectangle uses a parallel coordinate layout based on the X-axis and Y-axis. Taking the rectangle PQLM in Figure 3 as an example, the method for calculating the rectangle coordinates is defined as follows:

[0113] P = <leftMargin, topMargin>

[0114] Q = <leftMargin, topMargin + height rect >

[0115] L = <leftMargin + width rect , topMargin + height rect >

[0116] M = <leftMargin + width rect , topMargin>

[0117] In the formula: leftMargin is the distance of the parallel rectangle from the left boundary; topMargin is the distance of the parallel rectangle from the upper boundary; height rectis the height of the parallel rectangle; width rect is the width of the parallel rectangle.

[0118] Taking Figure 3 the midpoint A and point B of the connection line ABCDEFGHIJKL of the goods in it as an example, the method for calculating the coordinates of the points on the connection line is defined as follows:

[0119] A = <leftMargin + width inN , topMargin>

[0120] B = <leftMargin + width inN , topMargin + height rect >

[0121] In the formula: width inN is the length of the distance from the intersection point of the connection line and the parallel rectangle to the left boundary of the rectangle.

[0122] The length width of the distance from the intersection point between the connection line and the parallel rectangle to the left boundary of the rectangle inN maps the magnitude of the turnover attribute value corresponding to the goods. Since the turnover attribute values of some goods deviate too much from the mean value, using a linear scale will result in most of the connection lines being concentrated on the left side of the rectangle, and there are only sporadic connection lines distributed on the right side of the rectangle. This situation is not conducive to users' analysis. Therefore, this method uses a logarithmic scale to map the turnover attribute value and the length of the distance from the intersection point to the left boundary of the rectangle. Taking Figure 3 the parallel rectangle PQLM representing the inbound quantity of goods in the current month and the connection line ABCDEFGHIJKL of the goods in it as an example, the method for calculating the width width inN is defined as follows:

[0123]

[0124] In the formula: InNum unit is the inbound quantity of the goods corresponding to the connection line in the current month; InNum max is the maximum inbound quantity of goods in the current month; adding 1 to the true number of the logarithm is to avoid negative values.

[0125] 2. Visualization layout and implementation of the goods outbound mode view:

[0126] The goods outbound mode view mainly includes a nested rectangle on the left and a mode unit on the right. As Figure 4 shown, the overall view consists of three rows. Each row represents the outbound mode of a kind of goods. The leftmost is a nested rectangle, and then three mode units are arranged from left to right in descending order of the frequency of the corresponding frequent mode of the goods. The specific layout and implementation process are as follows:

[0127] (1) Visualization layout and implementation of the nested rectangle

[0128] Data definition: The nested rectangles encode the goods, the owner to whom the goods belong, and the monthly outbound frequency of the warehouse. For each nested rectangle, define the data rectData as follows:

[0129] rectData = <fre unit ,fre owner ,fre mon >

[0130] Where: fre unit is the number of outbound times of a certain good in the current month; fre owner is the number of outbound times of all goods under the owner to whom the good belongs in the current month; fre mon is the total number of outbound times of all goods in the warehouse in the current month.

[0131] Layout calculation: As shown in (a) of Figure 6 , the heights of the three rectangles nested from the outside to the inside of the nested rectangle are respectively defined as fixed decreasing values Height mon , Height owner and Height unit . This method linearly encodes the outbound frequency of goods using the width of the rectangle. The monthly outbound frequency fre mon of the warehouse is mapped to the width Width mon of the outer rectangle, the monthly outbound frequency fre owner of the owner is mapped to the width Width owner of the middle rectangle, and the monthly outbound frequency fre unit of the goods is mapped to the width Width unit of the inner rectangle. Fix the value of the outer Width mon , and according to the proportional relationship of fre unit , fre owner and fre mon , the corresponding rectangle widths Width owner and Width unit can be calculated. Draw each rectangle based on the fixed height and the calculated width, and fill the rectangles from the outside to the inside with dark gray, gray, and light gray respectively.

[0132] (2) Visual layout and implementation of the pattern unit

[0133] The upper part of the pattern unit is a circular part encoding the picking time of the frequent pattern, and the lower part is a rectangular part encoding the composition of the goods in the frequent pattern.

[0134] Data definition:

[0135] The circular part encodes the frequency of the outbound mode, the average picking time of the mode, and the picking time of the goods in the mode. For each circular part, define the data circleData as follows:

[0136] circleData = <fre, PTimeList>

[0137]

[0138] Where: fre is the occurrence frequency of the mode; PTimeList is the picking time list of the goods in the mode; PTime mode is the average picking time of the mode; is the picking time of the goods i in the mode.

[0139] The rectangular part encodes the composition of the goods in the outbound mode. For each rectangular part, define the data rectData as follows:

[0140] rectData = <Unit0, … Unit i …, Unit n >

[0141] Unit i = <SKUNo i , PTime i >

[0142] Where: Unit i is the data corresponding to the goods i; SKUNo i is the code corresponding to the goods i; PTime i is the picking time of the goods i.

[0143] Layout calculation:

[0144] As Figure 6 shown in (b) and (c) below, the circular part consists of the outer gray arc, the inner ring, and the background circle.

[0145] The arc length l of the gray arc encodes the frequency fre of the mode mode , and through fre mode and the outbound times fre of the corresponding source goods of the mode unit the radian θ of the arc can be calculated frequency , and then multiplied by the radius R of the outermost circle max to obtain the arc length of the gray arc. Among them, the calculation methods of the radian θ frequency and the arc length l are defined as follows:

[0146]

[0147] l = θ frequency * Rmax

[0148] The radius of the inner ring encodes the picking time PTime. The radius of the background circle R mode Encode the average picking time PTime of this mode mode 、The radius of the orange ring R unit0 The picking time PTime of the source product corresponding to the coding mode unit0 , the radius of the green ring R unit1 , R unit2 Encode the picking time PTime of other goods unit1 、PTime unit2 . Through the picking time PTime and the maximum picking time PTime of the goods in the month max The radius R of the ring can be calculated, and then the drawing of the ring and the background circle can be completed. mode For example, the method for calculating the radius length is defined as follows:

[0149]

[0150] like Figure 6 As shown in (d), the rectangular part is composed of multiple rectangles. The orange rectangle on the far left corresponds to the source product Unit0 corresponding to the pattern, and the other green rectangles correspond to other products Unit1 and Unit2 in the pattern. The width len of each rectangle can be calculated by dividing the overall width len by the number of products n included in the pattern. rect , and then complete the drawing of the RectPart view.

[0151] Step 4: Interaction Design

[0152] In the product attribute filtering view, after swiping a line segment within any range in any rectangle on the left, the product lines that meet all the swiping conditions will be highlighted, and the general selection list on the right will also display all selected product information. After clicking on the product of interest from the general selection list and adding it to the selected list, the product delivery mode view will be updated in conjunction; after clicking the square of the mode unit in the product delivery mode view, the corresponding product will be added to the last row of the view, and the selected list in the product attribute filtering view will be updated in conjunction; in addition, when the mouse is hovered over the elements in the two views, there will be a floating prompt box to display detailed information.

[0153] 1. Interactive design of product attribute screening view:

[0154] like Figure 7 As shown in the figure, the product attribute filtering view supports four mouse interactive operations: swiping, hovering, clicking and scrolling:

[0155] Brushing: Without brushing, the parallel coordinate rectangle displays all the connections (such asFigure 2 As shown in the figure, it shows the attribute distribution of all commodities in the selected month. Each line segment in the rectangle can be selected by brushing. The selected line represents the commodities with the corresponding attribute in the range. The selected commodity line is highlighted and other lines are hidden.

[0156] Hover: When the mouse hovers over a product line or a row in the filter list, a semi-transparent information pop-up box will be displayed synchronously near the cursor. Through this interaction, you can see the detailed information of the product in the prompt box: product ID, owner ID, and specific values of six turnover attributes.

[0157] Click: The "Action" column on the far right of the product filter list supports clicks. Click the "+" icon in the general selection list to add the corresponding product to the selected list and update the product delivery mode view simultaneously; click the "×" icon in the selected list to delete the product in both the selected list and the product delivery mode view simultaneously.

[0158] Scrolling: When there are many selected items, the selected list supports mouse scrolling.

[0159] 2. Interactive design of goods outbound mode view:

[0160] like Figure 8 As shown in the figure, the product outbound mode view supports two interactive operations: floating and clicking:

[0161] Hover: Mouse hover interaction is designed to help users understand the specific meaning of visual elements.

[0162] like Figure 8 As shown in (a), when the mouse hovers over the nested rectangle, the information prompt box corresponding to the rectangle is displayed.

[0163] like Figure 8 As shown in (b), when the mouse hovers over the background circle on the upper side of the pattern unit and the arc on the outer side, the corresponding element will be highlighted and the corresponding information floating box will be displayed. When the mouse hovers over the rectangle on the lower side of the pattern unit, the rectangle will be highlighted and the corresponding information floating box will be displayed, and the circle corresponding to the product represented by the rectangle will also be bolded and highlighted.

[0164] Click: Click the small rectangle representing the product at the bottom of the pattern unit to add frequent patterns to be explored. In the product attribute filtering view, the product items in the selected list are updated in a linked manner, and the product outbound pattern row in this view is updated at the same time, using a first-in-first-out strategy from top to bottom.

Claims

1. A visualization method for the turnover mode of goods in digital warehousing, characterized in that The following steps are involved: S1: Data acquisition and processing Obtain digital warehouse data, perform data cleaning, data preprocessing, and frequent pattern mining on the original data, extract and retain effective information, and store the goods turnover performance data and goods outbound pattern data in the database; S2: Visual Mapping Visually map the data obtained in step S1 through the visual channel: Design a product attribute filtering view, use parallel coordinate rectangles to map the distribution of each product's turnover attribute value in a specified year and month, and the line segments in the rectangles map the product and its attribute values; In addition, the filtered goods are displayed using the product filter list; Design a product outbound pattern view, using three layers of nested rectangles to map the proportion of the selected product's outbound times in its owner and its warehouse; use pattern units to map the frequent patterns of the selected product being outbound with other products, the frequency of occurrence of the frequent patterns, as well as the picking time of the corresponding product and the average picking time of the frequent patterns; S3: Visualization layout and implementation Visualize and layout the visual modules mapped in S2: In the product attribute filtering view, the parallel coordinate rectangle on the left arranges the parallel rectangles from top to bottom, and the line segments are arranged from left to right inside the rectangle according to the attribute values of the products, and connect the vertical line segments representing the same product in each rectangle; the product filtering list on the right includes the general selection list above and the selected list below; In the product delivery mode view, the left side is the nested rectangle corresponding to the product, and the right side is the n most frequently appearing pattern units in the frequent delivery mode corresponding to the product. The n rows are arranged from top to bottom for comparison; S4: Interaction Design In the product attribute filtering view, after swiping a line segment within any range in any rectangle on the left, the product lines that meet all the swiping conditions will be highlighted, and the general selection list on the right will also display all the selected product information. After clicking on the product of interest from the general selection list and adding it to the selected list, the product outbound mode view will be updated in conjunction; After clicking the square of the mode unit in the product outbound mode view, the corresponding product will be added to the last row of the view, and the selected list in the product attribute filtering view will be updated in conjunction; In addition, when you hover the mouse over the elements in both views, an information pop-up box will display detailed information.

2. The visualization method for the goods turnover mode oriented to digital warehousing according to claim 1, wherein In step S1, data acquisition and processing are specifically as follows: S11: Obtain three types of data, including inbound orders, outbound orders, and daily inventory of goods from the digital warehouse management platform. The attributes of the inbound / outbound order data include: order ID, consignor ID, product ID, timestamps of various inbound / outbound operations, and quantity, volume, and quality of the goods; outbound order data also includes outbound batch ID; the attributes of the daily inventory data include: inventory record ID, date, product ID, consignor ID, shelf ID, and inventory quantity; S12: Screen and delete the invalid parts of the original goods inbound order data and outbound order data; for the missing parts of the goods inventory data, calculate, extract and complete them based on other existing information; convert and integrate the data to extract goods turnover performance data; S13: Construct a transaction dataset to record the types of goods included in each outbound batch; mine the frequent outbound goods combinations for each type of goods, construct corresponding new transaction data for each type of goods, and use the FP-Growth algorithm to perform frequent pattern mining on the goods combinations in the outbound batches to generate outbound pattern data containing the goods. S14: Design the database schema according to the data characteristics, and store the goods turnover performance data and the goods outbound pattern data into the MySQL database.

3. The visualization method for the goods turnover mode oriented to digital warehousing according to claim 2, characterized in that, The construction of the transaction dataset described in S13 specifically includes: S131: Based on the goods turnover performance data, group the outbound order data by the shipper ID. S132: For each order group belonging to a shipper, group it by the outbound batch ID. S133: For each order group belonging to an outbound batch, remove duplicates from all the goods IDs according to the goods information included in the outbound order to obtain a combination of goods IDs. S134: Construct a goods outbound transaction dataset based on the shipper ID, the outbound batch ID, and the combination of goods IDs.

4. The visualization method for the goods turnover mode oriented to digital warehousing according to claim 2, wherein The mining of the frequent outbound goods combinations for each type of goods described in S13 specifically includes: 1) For each type of goods, filter out the corresponding transaction items from the original transaction dataset according to the corresponding shipper ID. 2) Filter out the transaction items containing the goods according to the combination of goods IDs corresponding to the outbound batch, and then form a new transaction dataset for the goods. 3) Apply the frequent pattern mining algorithm FP-Growth to the new transaction dataset of the goods to obtain the frequent patterns containing the goods. In the subsequent description, the goods will be referred to as the "source goods", and the goods other than the source goods in the frequent patterns will be referred to as the "other goods".

5. The visualization method for the goods turnover mode for digital warehousing according to claim 4, characterized in that, In step S2, the visual mapping is specifically as follows: S21: Perform shape and position mapping on the parallel coordinate rectangles of the goods attribute filtering view: use rectangles to map the turnover attributes of the goods, and use lines to map the goods; the six rectangles map the six turnover attributes of the total inbound quantity, average receiving time, average shelving time, total outbound quantity, average picking time, and average packaging time of the goods in the selected month from top to bottom in sequence. Use the position of the vertical line segment to map the magnitude of the corresponding attribute value of the goods, which increases proportionally from left to right, and connect the same goods with a slanted line. S22: Perform shape and position mapping on the goods filtering list of the goods attribute filtering view: use a "+" shaped icon to map the operation of adding goods, and use an "×" shaped icon to map the operation of deleting goods; the vertical direction of the list maps different goods, and the horizontal direction maps the three attributes of "goods ID, shipper ID, operation" from left to right in sequence. S23: Perform shape, position, and color mapping on the nested rectangles of the goods outbound pattern view: use three nested rectangles to map the outbound times of the goods at three levels; the innermost dark gray rectangle maps the outbound times of the selected goods in the current month, the middle gray rectangle maps the outbound times of all the goods under the name of the shipper to which the goods belong, and the outermost light gray rectangle maps the total outbound times of all the goods in the warehouse. S24: Mapping the shape, position, color, radius and arc length of the pattern unit of the product outbound pattern view: the upper circular part uses the arc length of the outer arc to map the frequency of the frequent pattern, the radius of the inner multiple circular rings maps the outbound picking time of each product in the product combination, and the radius of the background circle maps the average outbound picking time of the product combination; The rectangular part at the bottom uses the leftmost small rectangle to map the source goods, and the other small rectangles map other goods that are shipped out together with the source goods in this frequent mode; two colors are used to map the source goods and other goods respectively, and the corresponding circles are consistent with the rectangles.

6. The visualization method for the goods turnover mode for digital warehousing according to claim 1, characterized in that In step S3, the visualization layout and implementation of the parallel coordinate rectangle in the product attribute screening view are as follows: S3a: Parallel coordinate rectangles use parallel coordinate layout with the X-axis perpendicular to the Y-axis. Based on the predefined boundary distance and the height and width of the rectangle, the coordinates of the six rectangles are calculated respectively, and the attribute name is marked in the upper left corner; S3b: The distance between the intersection of the product line and the parallel rectangle and the left border of the rectangle maps the size of the corresponding turnover attribute value of the product. A logarithmic scale is used to map the turnover attribute value of the product and the distance between the intersection and the left border of the rectangle. The left border of the rectangle maps the minimum attribute value, and the right border maps the maximum attribute value. S3c: Determine the position of the product on the six vertical line segments in the six rectangles and connect them with diagonal lines.

7. The visualization method for the goods turnover mode oriented to digital warehousing according to claim 4, characterized in that In step S3, the visual layout and implementation of the goods outbound mode view are specifically as follows: S31: In the nested rectangle on the left, the heights of the three rectangles from the outside to the inside are predefined, decreasing in sequence; the width of the outermost rectangle is predefined, and then the widths of the middle and inner rectangles are calculated according to the linear proportional relationship between the number of warehouse outbound shipments, the number of cargo owners outbound shipments, and the number of cargo outbound shipments in the current month; and the rectangles from the outside to the inside are filled with light and dark gradient colors; S32: In the circular part of the right mode unit, the arc of the outer arc is calculated by the frequency of the mode and the number of times the source goods corresponding to the mode are shipped out of the warehouse, and then multiplied by the predefined maximum radius to obtain the arc length of the outer arc; the radius of the ring is calculated by the linear ratio of the goods picking time and the maximum picking time of the month, and the radius of the background circle is calculated by the linear ratio of the average picking time of the mode and the maximum picking time of the month; S33: In the rectangular part of the pattern unit on the right, predefine the height and overall width of the rectangle, and divide the overall width by the number of goods included in the pattern to calculate the width of each rectangle; then use two colors to fill the rectangle to map the source goods and other goods respectively.

8. The visualization method for the goods turnover mode oriented to digital warehousing according to claim 1, characterized in that In step S4, the interaction design is specifically as follows: S41: The product attribute filtering view supports four mouse interactive operations: brushing, hovering, clicking and scrolling: each line segment in the rectangle can be brushed, and the selected line represents the products with the corresponding attribute in the range; when the mouse hovers over a product line or a row in the filtering list, the corresponding information floating box is displayed; Click the "+" icon and "×" icon in the "Operation" column on the far right of the product filter list to update the view in a linked manner; When there are many selected items, the list supports mouse scrolling; S42: The goods outbound mode view supports two mouse interactive operations: floating and clicking. When the mouse is hovered over a nested rectangle, the corresponding information floating box is displayed. When the mouse is hovered over the outer arc, background circle or small rectangle of the pattern unit, the corresponding information floating box is displayed and the corresponding element is highlighted. Clicking the small rectangle representing the goods at the bottom of the pattern unit with the mouse can increase the frequent patterns to be explored and update the view in a linked manner.

Images