Manufacturing industry big data management system based on supply chain

Abstract

The invention relates to the technical field of data management, in particular to a supply chain-based manufacturing industry big data management system, which comprises an event attribution module, a data integration module, an inventory trend module, a batch judgment module and a parameter tracing module. According to the method, automatic association analysis of multi-dimensional parameters such as process, logistics, inventory and environment is realized through time sequence synchronous acquisition of multi-source data and dynamic linkage of business events, and accurate circulation of whole-process data is promoted by adopting an event-level feature recognition and parameter screening mechanism; furthermore, positioning and traceability of an abnormal fluctuation process are supported in a process traceability and trend discrimination mode, information interconnection and parameter relation chain establishment of each manufacturing link are enhanced, synchronous processing and hierarchical clustering of multi-parameter heterogeneous data are realized, and the real-time decision support capability and traceability precision of the data are improved; and the response speed of an abnormal process and inventory fluctuation in a manufacturing scene and the intelligent level of data application are effectively enhanced.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and in particular to a big data management system for manufacturing based on the supply chain. Background Technology

[0002] The field of data management involves the collection, storage, processing, analysis, and utilization of data, including structured data management, efficient retrieval, secure storage, and data sharing and circulation. This field encompasses everything from underlying data storage solutions to high-level data governance standards, aiming to improve the efficiency and reliability of data application across various industries. Traditional manufacturing big data management systems refer to the aggregation, management, and application of various data generated from raw material procurement, production, manufacturing, inventory management, and logistics transportation within the manufacturing supply chain. For the large-scale, heterogeneous, and multi-source data generated during the supply chain process, centralized databases are typically used for data collection and batch entry. Information is stored using fixed-format data tables, and information updates across multiple stages are achieved through regular manual or automatic data synchronization. Data retrieval and report generation are performed using pre-defined data query statements.

[0003] Existing technologies use centralized databases and fixed-structure data tables to manage supply chain data. Business node information relies on periodic batch entry and manual synchronization. Data flow and analysis are limited by rule-driven static queries. Production and inventory data are difficult to dynamically link, cross-link information has weak timeliness, and the integration of multi-source parameters is not high. As a result, process anomalies in the manufacturing process cannot be identified in real time and accurately traced. Inventory change trends and logistics status are also difficult to be fed back in a timely manner. Data management and traceability capabilities are significantly limited, affecting the overall management efficiency of the supply chain and the depth of data application. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a supply chain-based manufacturing big data management system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a manufacturing big data management system based on the supply chain, the system comprising:

[0006] The event attribution module analyzes changes in business processes based on warehouse transfer instructions, combines material inbound labels and logistics outbound task records to identify changes in action status fields, and attributes each status change to each business event by dynamically counting the task record sequence, thus obtaining the event trigger feature set.

[0007] The data integration module analyzes the distribution of temperature and humidity monitoring node parameters under task tags based on the event trigger feature set, records real-time inventory quantity changes, monitors vehicle location information collected by the locator, and obtains a multi-source joint dataset.

[0008] Based on the multi-source joint dataset, the inventory trend module compares the changing trends of inventory data for each category, calculates the synchronicity between the dynamic inventory level of the finished goods warehouse and the periodic changes, analyzes the changes in the inventory of the same category during the same period, determines the correlation between the changes in the turnover inventory of the transit warehouse and the overall trend, and obtains the key fluctuation characteristics of inventory.

[0009] Based on the key fluctuation characteristics of the inventory, the batch discrimination module analyzes the temperature fluctuations in the coating process parameter records, judges the degree of matching between the equipment operation time and the process standard, compares the parameter fluctuations with the process boundary differences, archives the associated parameters, and obtains the batch process anomaly sequence.

[0010] The present invention improves upon the following: the event triggering feature set includes event classification coding, event priority labels, and node change descriptions; the multi-source joint dataset includes an environmental perception data structure, inventory monitoring records, and logistics tracking entries; the key inventory fluctuation features include category fluctuation indicators, periodic change mapping, and associated status attributes; and the batch process anomaly sequence includes batch identification information, abnormal process features, and deviation source annotations.

[0011] The present invention is improved in that the event attribution module includes:

[0012] The instruction deconstruction submodule, based on warehouse transfer instruction information, analyzes material inbound labels and logistics outbound task content through operation type, target location and time records, judges the logical consistency and process integrity between related information, and calculates the task transfer path between nodes in combination with transfer business links to obtain the transfer status node set.

[0013] The state discrimination submodule calls the allocation state node set to judge the changes in the action state field, analyze the continuity of the state before and after the task execution, identify the state switching between each business action by comparing the task time and the allocation start node time, identify the key task execution stage, and obtain the state evolution trajectory index.

[0014] The event mapping submodule analyzes the action state content corresponding to each stage based on the state evolution trajectory index. By analyzing the action type, state switching, and task time information, it determines the business event attribution of each data item, filters priority and classification features, and obtains the event trigger feature set.

[0015] The present invention is improved in that the data integration module includes:

[0016] The data acquisition and distribution submodule determines the monitoring nodes and environmental parameters corresponding to the task tags based on the event trigger feature set, optimizes the correspondence between the monitoring node numbers and task tags, calculates the distribution of environmental parameters of each node under the different task stages, filters out nodes with fluctuation characteristics in the monitoring node data, and obtains the node parameter distribution density.

[0017] The inventory trajectory submodule calls the node parameter distribution density and compares it with the inventory change records under each task label to determine the increase or decrease trend of inventory quantity, calculates the inventory fluctuation amplitude in the continuous time series, and obtains the inventory trend structure.

[0018] The location information integration submodule analyzes vehicle location data within the time period corresponding to the task tag based on the inventory trend structure, filters key spatial trajectory points during inventory fluctuation phases, optimizes the matching relationship between task time and vehicle trajectory, judges the integrity of spatial movement trajectory, and obtains a multi-source joint dataset.

[0019] The present invention is improved in that the inventory trend module includes:

[0020] The category trend analysis submodule analyzes the inventory monitoring content and task tags based on the multi-source joint dataset, compares the changing trends of the inventory quantity of each category in the same period, judges the frequency and growth of fluctuations, and obtains the category inventory fluctuation frequency.

[0021] The finished goods synchronization calculation submodule compares the inventory fluctuation frequency of the product category with the finished goods inventory change process to obtain the finished goods inventory trend synchronization degree.

[0022] The transit correlation judgment submodule determines the correlation between transit warehouse inventory changes and finished goods warehouse trends within key periods based on the finished goods inventory trend synergy. It filters time periods with the same fluctuation direction and synchronous inventory changes, and identifies key category characteristics of the changes to obtain key inventory fluctuation characteristics.

[0023] The present invention is improved in that the batch discrimination module includes:

[0024] The temperature fluctuation judgment submodule analyzes the coating process parameter records based on the key fluctuation characteristics of the inventory, judges the temperature change trend of each batch at the temperature measurement node, compares it with the allowable range of the process standard, judges whether the actual temperature fluctuation meets the process requirements, and obtains the process temperature fluctuation status.

[0025] The time sequence comparison submodule analyzes the start and end times of each batch of equipment operation based on the process temperature fluctuation status, calculates the continuous running time of the equipment, compares the operation process with the operation range specified by the process standard, determines whether the equipment operation process meets the process specifications, and obtains the process operation time sequence status.

[0026] The process deviation archiving submodule filters batches that fail to meet process standard requirements during temperature fluctuations and equipment operation based on the process operation sequence status. It calculates the deviation performance of the operation process, temperature fluctuations and ambient temperature, obtains the process abnormality deviation degree, archives the parameter information that meets the judgment conditions, and obtains the batch process abnormality sequence.

[0027] The present invention has an improvement, wherein the system further includes:

[0028] Based on the batch process anomaly sequence, the parameter traceability module traces the changes in environmental control parameters before and after the process, analyzes equipment operating efficiency and operation duration, judges the continuous trend changes of environmental parameters, collects related process segments, and obtains parameter trend traceability segments.

[0029] The parameter trend tracing segment includes trend variation nodes, tracing path indexes, and process stage identifiers.

[0030] The present invention is improved in that the parameter traceability module includes:

[0031] The trend recognition submodule analyzes the environmental control parameters before and after the process based on the batch process anomaly sequence, compares the changing trends of temperature, humidity and air pressure in each time period, judges the continuity and consistency of the parameter change direction, and filters out data that show the same changing trend in a period of time to obtain the stage trend sequence.

[0032] The reversal calculation submodule analyzes the changes in the direction of parameter change within each continuous interval based on the stage trend sequence, identifies the nodes where the parameter direction reverses, calculates the number of times the trend reversal occurs, and obtains the set of direction change nodes.

[0033] The process aggregation submodule determines the process segment where the reverse node is located based on the set of directional change nodes, identifies the process segment identifier associated with the node, and aggregates the associated process segments to obtain the parameter trend traceability segment.

[0034] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0035] In this invention, by synchronously collecting multi-source data in a time sequence and dynamically linking business events, automatic correlation analysis of multi-dimensional parameters such as process, logistics, inventory, and environment is achieved. Event-level feature recognition and parameter screening mechanisms are used to promote the accurate flow of data throughout the entire process. Furthermore, process traceability and trend discrimination methods are used to support the location and traceability of abnormal process fluctuations. Information interconnection and parameter relationship chain construction in various manufacturing links are strengthened, and synchronous processing and hierarchical clustering of multi-parameter heterogeneous data are achieved. This improves the real-time decision support capability and traceability accuracy of data, and effectively enhances the response speed and intelligent level of data application in manufacturing scenarios for abnormal process and inventory fluctuations. Attached Figure Description

[0036] Figure 1 This is a system flowchart of the present invention;

[0037] Figure 2 This is a flowchart of the event attribution module in this invention;

[0038] Figure 3This is a flowchart of the data integration module in this invention;

[0039] Figure 4 This is a flowchart of the inventory trend module in this invention;

[0040] Figure 5 This is a flowchart of the batch discrimination module in this invention;

[0041] Figure 6 This is a flowchart of the parameter traceability module in this invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0043] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0044] Example

[0045] Please see Figure 1 This invention provides a technical solution: a supply chain-based manufacturing big data management system comprising:

[0046] The event attribution module analyzes changes in supply chain business processes based on warehouse allocation instructions, combines material inbound labels and logistics outbound task records to identify changes in action status fields, and performs attribution judgment on each status change process by dynamically checking the time sequence of task records. It then classifies and maps business events according to action status characteristics to obtain event trigger feature sets.

[0047] The data integration module analyzes the distribution of the collected parameters of the temperature and humidity monitoring nodes under the task tags based on the event trigger feature set, records the real-time inventory quantity change trajectory, monitors the vehicle location information obtained by the locator, and then synchronizes and organizes it with the task tags and collection time to obtain a multi-source joint dataset.

[0048] The inventory trend module is based on a multi-source joint dataset. It compares the changing trends of inventory data for each category, calculates the synchronicity between the dynamic inventory of finished goods warehouse and the inventory changes within the cycle, analyzes the concurrent changes of other categories of inventory, judges the correlation between inventory changes in the transit warehouse and the overall trend, identifies categories with key change characteristics, and obtains key inventory fluctuation characteristics.

[0049] The batch discrimination module analyzes the temperature fluctuation range in the coating process parameter records based on key fluctuation characteristics of inventory, judges the degree of conformity between the equipment operation time sequence and the process standard, compares the difference between parameter fluctuation and process boundary, archives parameter information that meets the judgment conditions, and obtains the batch process abnormal sequence.

[0050] The parameter traceability module traces changes in environmental control parameters before and after a process based on a batch process anomaly sequence, analyzes equipment operating efficiency and operation duration, determines the trend changes of environmental control parameters in continuous stages, calculates the number of trend reversals, and aggregates related process segments to obtain parameter trend traceability segments.

[0051] The event trigger feature set includes event classification coding, event priority labels, and node change descriptions. The multi-source joint dataset includes environmental perception data structures, inventory monitoring records, and logistics tracking entries. The key inventory fluctuation features include category fluctuation indicators, periodic change mapping, and associated status attributes. The batch process anomaly sequence includes batch identification information, abnormal process features, and deviation source annotations. The parameter trend traceability fragment includes trend variation nodes, traceability path index, and process stage identifiers.

[0052] In Module 1, business process changes refer to actual operational changes occurring in the supply chain, such as the progress or changes in key business actions like material allocation, warehousing, outbound, and inventory counting; changes in action status fields refer to changes in the fields representing business status in the system's business records (such as database tables or message logs) (e.g., "incomplete" becomes "complete"); the time sequence of task records refers to the sequential relationship of business operations in the time dimension, i.e., the order in which various tasks occur; attribution judgment refers to analyzing the causes of a certain status change to determine which type of business event it belongs to (e.g., inventory change caused by inventory counting or inventory increase caused by warehousing); action status features refer to status parameters or markers that reflect the essence of business actions, such as system identification fields like "sorted," "in transit," and "warehousing completed"; mapping business events refers to establishing a one-to-one correspondence between the above features and business event types to facilitate subsequent automatic identification and data collection.

[0053] In Module 2, temperature and humidity monitoring nodes refer to sensor devices such as temperature and humidity devices deployed in scenarios such as warehouses and logistics, used to collect environmental parameter data; task tags refer to unique identifiers or classification tags assigned to each type of business event or collection task (such as "20240616_Inbound_A1"), which facilitate data archiving and retrieval; change trajectory refers to the process of inventory quantity changing over time, that is, a data sequence generated through continuous collection, reflecting the dynamic change status of inventory.

[0054] In Module 3, "Change Trend" refers to the overall direction or pattern of change of a certain type of inventory data over a period of time, such as continuous increase, decrease, or cyclical fluctuation; "Synchronization Calculation" refers to aligning two or more sets of data (such as finished goods warehouse and cyclical changes) on the time axis and calculating the synchronous fluctuation relationship over time; "Other Category Inventory" refers to the inventory data of all material categories involved in the analysis in the system, excluding the target category; "Turnover Link" refers to the link in the supply chain from one node to another, such as the circulation process from a transit warehouse to the production line or customer warehouse; "Overall Trend" refers to the overall fluctuation characteristics or common changes of all inventory categories involved in the analysis within the same time window.

[0055] In Module 4, coating process parameters refer to process parameters such as temperature, time, and ambient humidity in the "coating" process (such as automotive painting and electronic product casing coating) in manufacturing; equipment operation time refers to the continuous running time of production equipment within a production batch or process segment; process standard: refers to the range of operating parameters specified in the production process (such as temperature range, operation duration range), which is usually derived from process specifications or technical manuals; meeting the judgment criteria means that the data characteristics of a certain parameter or batch meet the discrimination requirements set by the system and can be identified as abnormal, normal, or other states.

[0056] In Module 5, changes in environmental control parameters refer to the changes in key parameters (such as temperature, humidity, and air pressure) in the production or storage environment at different stages of the production process; trends in continuous stages refer to the analysis of the changing trends of the same parameter in continuous stages of the process when the system is traced, such as gradual increase, slowing down of fluctuations, and sudden reversal.

[0057] Please see Figure 2 The event attribution module includes:

[0058] The instruction deconstruction submodule, based on warehouse transfer instruction information, analyzes material inbound labels and logistics outbound task content through operation type, target location and time records, judges the logical consistency and process integrity between related information, and calculates the task transfer path between nodes in combination with transfer business links to obtain the transfer status node set.

[0059] The operation type is extracted from the transfer records. For example, if the record contains "inbound transfer," it is marked as a first-type operation and converted into a fixed number for subsequent unified processing. Next, the target location, such as warehouse A1, is extracted. By consulting the warehouse structure coding table, A1 is mapped to a preset coordinate position, and its physical node level is recorded. Then, the operation time is extracted and standardized into a unified time format for subsequent sorting and comparison. For example, "2025-05-03 14:25:00" is converted into a standard timestamp for calculation. Next, the label number is extracted from the material inbound label and compared with the current transfer instruction number to confirm whether they belong to the same transfer batch. For example, if the transfer number is "DT20250503-17" and the material label is "L20250503-17," the intermediate date range and the number field are extracted. The system compares each character individually. If they match perfectly, they are recorded as valid tags. Then, it analyzes whether there is a corresponding logistics outbound task for the transfer task. It searches for logistics task records with the same batch number and extracts fields such as vehicle number, starting location, and loading time. Each record is compared with the fields of the transfer task to confirm whether the location matches and whether the loading time is later than the transfer time. If the conditions are met, it is considered a successfully matched upstream and downstream task chain. Then, it is determined whether there is a spatial displacement between the transfer target location and the logistics task target location. By consulting the task path mapping table, the length of the task transfer path between the two points is identified. For example, there are 3 task jump points from L001 to L004. The path nodes are recorded and the time and location correspondence is marked to form a set of transfer status nodes, which represents the spatial path and temporal evolution of the transfer task from the source location to the target location.

[0060] The status discrimination submodule calls the allocation status node set to judge the changes in action status fields, analyzes the continuity of status before and after task execution, identifies the status transition between various business actions by comparing the task time and the allocation start node time, identifies key task execution stages, and obtains status evolution trajectory indicators.

[0061] First, all allocation tasks are categorized and summarized by task number. Each task contains several allocation records. Then, the records are sorted in ascending order by operation time. The status field corresponding to each record is extracted, and the status between adjacent records is analyzed. For example, if the first record's status is "Not Executed" and the second is "Executing," a status switch is determined. The location of the status switch is marked and a timestamp is recorded. Next, the start time of the allocation task is compared with the inventory task time recorded in the system. If the allocation task's start time is earlier than the inventory task's time, it is included in the status evolution judgment process; otherwise, the task record is considered not to belong to the same business chain and is therefore excluded. After eliminating the previous stage, we continue to analyze whether there are any continuous and unchanging stages in the status fields. For example, if three consecutive status fields are "Executing", then we determine that the stage is a stable stage and record the start and end times of the stage. We further identify whether the next stage changes. If the status changes to "Completed", then we mark the transition from "Executing" to "Completed" as a state transition. We repeat this process to identify all continuous status stages and record the start and end times of each stage. We calculate the duration of each stage and integrate the information of all continuous status stages to form a clear sequence of state evolution trajectories that reflects the entire process of the task from not executing to executing and then to completing.

[0062] The event mapping submodule analyzes the action state content corresponding to each stage based on the state evolution trajectory index. By using action type, state switching and task time information, it determines the business event attribution of each data item, filters priority and classification features, and obtains the event trigger feature set.

[0063] For each state segment, a corresponding action state identifier is established. "Not Executed" is marked as the starting state, "Executing" as the intermediate execution state, and "Completed" as the completed state. Each state segment is assigned a task time tag based on the time information recorded in the state segment. Then, the operation type information from the transfer task record is combined. For example, if a record shows the operation type as "transfer transfer," the state "Executing" is bound to this type, marking this stage as "transfer execution." Then, "Completed" is combined with the operation type and marked as "transfer completed." A complete mapping table of action types and state content is established accordingly. Next, state transition time nodes are extracted from the trajectory, and each transition is checked to see if it meets the time interval conditions set in the task flow. For example, it is checked whether the time for the state to change from "Not Executed" to "Executing" is less than the set maximum. Switching intervals are allowed; if so, the state switch is considered valid. The state segment is then jointly judged with the operation type of the transfer task. For example, if the continuous state segment is "Executing", the type is "Inbound Transfer", and the duration exceeds the specified threshold, it is judged as a valid "Inbound Execution" event. Subsequently, multiple action type events that occur within the same time period are selected from all events, and the corresponding event weight values ​​are compared. For example, the weight of the "Transfer Completed" event is 0.75, and the weight of "Inbound Completed" is 0.65. The event with the higher weight, "Transfer Completed", is selected as the main event type for this stage. Finally, the corresponding classification code and priority label are assigned according to the event type. For example, "Transfer Completed" is classified as category code C03, and the priority is marked as medium or high. The set of all marked event triggering features is output.

[0064] Please see Figure 3 The data integration module includes:

[0065] The data acquisition and distribution submodule determines the monitoring nodes and environmental parameters corresponding to the task tags based on the event trigger feature set, optimizes the correspondence between the monitoring node numbers and task tags, calculates the distribution of environmental parameters of each node under the different task stages, filters out nodes with fluctuation characteristics in the monitoring node data, and obtains the node parameter distribution density.

[0066] First, extract the task tag information from each event. The task tag format includes time, task type, and location fields, such as "20240616_Inbound_A1". Based on this tag, retrieve the associated temperature and humidity node number from the monitoring configuration table. For example, location A1 corresponds to nodes N021 and N022. Next, extract the temperature and humidity data of the node within the corresponding task time period. The raw data of each node is collected at 10-minute intervals, and the actual values ​​are recorded to form a time series. Then, analyze the uniqueness based on the frequency of task tag and node binding. When the same node number appears in multiple task tags, prioritize retaining the binding relationship with the highest frequency, readjust the mapping table, and ensure that subsequent tasks and nodes correspond one-to-one, completing the binding optimization. Subsequently, divide the task time period into multiple stages, such as start, processing, and completion, and summarize the temperature data of each node in each stage. For humidity parameters, the maximum, minimum, and average values ​​are statistically analyzed. The average value and standard deviation of the same node between adjacent stages are compared. If the difference in temperature standard deviation exceeds 1℃ or the difference in humidity standard deviation exceeds 5%, the node is marked as a fluctuating node. For example, the temperature standard deviation of node N021 is 1.8℃ in the processing stage and 0.6℃ in the completion stage, with a difference of 1.2℃, which is greater than the judgment threshold of 1℃. Therefore, it is recorded as fluctuating. Nodes with similar fluctuation characteristics under the same task are further screened, and the fluctuation frequency of the node in the task stage is counted. Finally, the density of temperature and humidity values ​​of all fluctuating nodes is calculated, that is, the frequency of occurrence in different intervals is counted. For example, the frequency in the temperature interval of 21℃ to 22℃ is 34 times, and the frequency in the temperature interval of 23℃ to 24℃ is 51 times. The parameter distribution density structure of the node is output based on the frequency.

[0067] The inventory trajectory submodule calls the node parameter distribution density and compares it with the inventory change records under each task tag to determine the increase or decrease trend of inventory quantity, using the following formula:

[0068]

[0069] Calculate the inventory fluctuation range in a continuous time series to obtain the inventory trend structure, where VE i SE represents the change in inventory during task period i. i SE represents the inventory quantity record at the i-th time point. i+1 TE represents the inventory quantity record at time point i+1. i TE represents the task execution timestamp at the i-th time point. i+1 DE represents the task execution timestamp at the (i+1)th time point. i ME represents the task complexity factor associated with task label i. i PE represents the number of inventory action types involved in task i. j n represents the average temperature and humidity collected at monitoring node j during the same period as task i.VE This indicates the total number of monitoring nodes;

[0070] Compare the inventory change records under each task tag to determine the increase or decrease trend of inventory quantity, and calculate the inventory fluctuation range in the continuous time series. First, extract the inventory record data within the time period corresponding to task tag T1001. The inventory quantity at time point t1 is 320, and the inventory quantity at time point t2 is 305, with a difference of -15. The inventory quantity is in units, and the execution timestamps of the records are TE. i =1609459200 seconds and TE i+1 = 1609459260 seconds, with a time interval of 60 seconds, and the task complexity factor is set to DE. i =5, the number of inventory movement types is ME i =3. The task involves three action types: warehousing, outbound, and inventory. It calls the environmental data of monitoring nodes N01 and N02 during this time period. The temperatures are 22.5℃ and 23.1℃, and the humidity is 45.3%RH and 46.7%RH, respectively. After normalization, the normalized temperature is 0.312 and 0.340, and the normalized humidity is 0.621 and 0.640, which corresponds to the node environmental parameter PE. j The sum is:

[0071]

[0072] Substitute the above parameters into the calculation formula:

[0073]

[0074] This result indicates that the inventory volatility parameter VE i A value of approximately -0.4589 is negative, indicating that the inventory quantity within the time period corresponding to task label T1001 shows a decreasing trend. Furthermore, the value is close to zero, suggesting that the relative strength of inventory fluctuations is weak considering the combined effects of time consumption, task complexity, and environmental disturbances. This result is directly related to the inventory trend structure of the steps in the inventory trajectory submodule. In other words, this value reflects the directionality and sensitivity of inventory fluctuations through the changing trend of the numerator and the stability of the denominator. Further simplification is as follows: using VE from the continuous task phase... i The data series is used to construct time series curves, and structural features such as trend slope changes, inflection points, peaks and troughs are extracted to form a complete inventory trend structure that can be used for inventory scheduling and prediction.

[0075] The location information integration submodule analyzes vehicle location data within the time period corresponding to the task tag based on the inventory trend structure, filters key spatial trajectory points during inventory fluctuation phases, optimizes the matching relationship between task time and vehicle trajectory, judges the integrity of spatial movement trajectory, and obtains a multi-source joint dataset.

[0076] Inventory volatility refers to the intensity, speed, and complexity of inventory changes between two adjacent points in time during a specific task period in manufacturing data management. It is a numerical value that combines factors such as changes in inventory quantity, time span, task complexity, and environmental conditions to reflect the activity level and volatility characteristics of inventory changes within a certain task phase.

[0077] First, extract the associated task tags and task execution time range from each inventory trend record. For example, the task "20240616_Transit_B3" executes from 9:00 AM to 11:00 AM. Based on this, filter out all vehicle trajectory data appearing within this time range from the vehicle location records. The trajectory data includes vehicle number, location coordinates, and timestamp, and is recorded in 5-second increments. The continuous trajectories of each vehicle are integrated into a coordinate time series. Then, clean up consecutive points with a position offset of less than 1 meter in the trajectory to remove duplicate records when the vehicle is stationary. Next, determine the center time of the fluctuation based on the inventory fluctuation time period indicated in the inventory trend. Then, extract the coordinate points within 10 minutes before and after the center time from the vehicle trajectory as key trajectory points. For example, if the inventory fluctuation time is 10:15, the trajectory records between 9:55 and 10:35 will be extracted. Finally, extract the target points based on the location field parsed from the task tags. For example, if the coordinates of position B3 are (14.5, 28.0), the vehicle trajectory points are compared one by one to see if any records fall within a 5-meter radius of that position. If they do, the vehicle trajectory is considered to match the task label; otherwise, it is recorded as an unmatched trajectory. Then, the vehicle's movement path is analyzed to see if there are any missing trajectories in consecutive time slices. For example, if a vehicle has no location points between 10:10 and 10:12, that segment is marked as a trajectory breakpoint. If the breakpoint time exceeds 30 seconds, the vehicle trajectory is marked as an incomplete path and will not participate in subsequent integration. All vehicles with a trajectory integrity rate of 95% or higher are selected. Then, the vehicle trajectory points are horizontally concatenated with their time, inventory change value, and associated task label to construct a unified joint data record. Each record contains fields such as vehicle coordinates, time, corresponding task, node number, environmental data, and inventory status. These are combined to form a unified format multi-source joint dataset output.

[0078] Please see Figure 4 The inventory trend module includes:

[0079] The category trend analysis submodule is based on a multi-source joint dataset. It analyzes inventory monitoring content and task tags, compares the changing trends of inventory quantity of each category in the same period, judges the frequency and growth of fluctuations, and obtains the category inventory fluctuation frequency.

[0080] Extract task labels, category numbers, inventory quantities, and record time fields from the dataset. Group each record by category number, and then sort each group of data by period according to the time field to form a time series. For example, extract the daily inventory quantity records from June 1st to June 10th, 2025 for category "P001", constructing a sequence such as: 210, 220, 198, 240, 235, 210, 205, 230, 225, 220. Next, calculate the difference between adjacent data in the sequence to obtain the daily fluctuation amount and record the direction of fluctuation. The fluctuation frequency can be obtained by counting the number of times the fluctuation direction changes in the sequence. For example, if the fluctuation direction changes 5 times in the above sequence, the fluctuation frequency is recorded as 5 times / 10 days. Then, count the number of differences greater than a threshold to determine the growth situation. The threshold is set to 1 / 10 of the average daily outflow of the category. 0.5 times, for example, if the average daily outbound volume of P001 is 8 pieces, and the threshold is set to 12 pieces, then if the inventory increase is 15 pieces on a certain day, it is recorded as a significant increase. If the cumulative number of significant increases is 3, then the growth situation is marked as high. Then compare the fluctuation frequency and growth situation of multiple categories, and define the frequency classification standard as: 02 is low frequency, 35 is medium frequency, and more than 6 times is high frequency. The growth situation classification standard is that the proportion of significant increase times in the period exceeds 40% as "high growth", 20% to 40% is "medium growth", and less than 20% is "low growth". For each category, determine its classification level in turn, and record the fluctuation frequency value and growth type label. For example, if the fluctuation frequency of P001 is 5 times and the significant growth rate is 30%, then its classification is medium frequency and medium growth. Combine the frequency values ​​and growth situation labels of all categories to form the category inventory fluctuation frequency output.

[0081] The finished goods synchronization calculation submodule compares the frequency of category inventory fluctuations with the changes in finished goods inventory using the following formula:

[0082]

[0083] Obtain the finished goods inventory trend correlation KG, where ΔTG cv ΔTG represents the change in the inventory quantity of the target category at the v-th time point, i.e., the change in the inventory quantity of the target category between two adjacent time points. fv This represents the change in the quantity of finished goods inventory at time point v, i.e., the change in the quantity of finished goods inventory between two adjacent time points, n. KG This represents the total number of time points, i.e., the total number of all time points that are statistically analyzed within the analysis period.

[0084] Obtain the inventory time series of the finished goods warehouse within the same period, and perform one-to-one correspondence processing on the two series according to the time nodes to construct the change series of the target category and the finished goods warehouse at each time point, denoted as ΔTG. cv With ΔTG fvThen, the difference square operation was performed on the two sets of changes, and the degree of cooperability was calculated using the formula:

[0085] During the calculation process, all participating parameters must be uniformly converted to "pieces / hour". After normalization, the changes at each time point are as follows:

[0086] ΔTG cv : 2.40, -1.60, 1.20, -2.00;

[0087] ΔTG fv : 2.00, -1.00, 0.80, -1.80;

[0088] First, calculate the numerator, i.e., the sum of squares of differences:

[0089] (2.40-2.00) 2 +(-1.60+1.00) 2 +(1.20-0.80) 2 +(-2.00+1.80) 2 ;

[0090] =0.16 + 0.36 + 0.16 + 0.04 = 0.72;

[0091] Then calculate the denominator, which is the sum of the squares of the two changing sequences:

[0092] 2.40 2 +(-1.60) 2 +1.20 2 +(-2.00) 2 +2.00 2 +(-1.00) 2 +0.80 2 +(-1.80) 2 ;

[0093] =5.76+2.56+1.44+4.00+4.00+1.00+0.64+3.24=22.64;

[0094] The degree of synergy is:

[0095]

[0096] The results indicate that the overall difference between the finished goods inventory change trend and the target category inventory fluctuation is small within the analyzed period, and the trend changes show a high degree of consistency. This suggests that the magnitude and direction of their changes over time tend to be synchronized. Specifically, the numerical result KG = 0.0318 is close to zero, reflecting that the cumulative squared deviation between the two inventory change sequences is relatively low, and the relative error accounts for a small proportion of the overall change energy. Therefore, this numerical result can be directly used as a quantitative expression of trend consistency and can be determined as a highly synchronized state. It can be further organized and archived as the finished goods inventory trend synergy degree KG defined in this step, and used as the input basis for subsequent judgment of transit warehouse inventory changes to identify whether there is a stage linkage or trend transmission effect in key categories.

[0097] The transit correlation judgment submodule determines the correlation between transit warehouse inventory changes and finished goods warehouse trends based on the finished goods inventory trend synergy, filters time periods with the same fluctuation direction and synchronous inventory changes, and identifies key category characteristics of the changes to obtain key inventory fluctuation characteristics.

[0098] Finished goods inventory trend synergy refers to the degree of synchronization and consistency between the trend of finished goods inventory changes and the trend of target category inventory changes within the same period. It reflects the degree of mutual coordination, consistent fluctuations, or deviations between the two sets of inventory data in terms of the direction and magnitude of change. It is an indicator for measuring the close correlation between the dynamic changes of finished goods inventory and specific category inventory.

[0099] Records within the same time window are extracted from finished goods warehouse inventory data and transit warehouse inventory data, and synchronized at a daily or hourly granularity. For example, from June 3rd to June 7th, 2024, inventory records at 9:00, 12:00, 15:00, and 18:00 each day are retained and grouped into two trend sequences based on record time: one for finished goods warehouse and the other for transit warehouse. Then, using the same time point as a benchmark, the direction of numerical changes in the two sequences is compared to determine whether there is a co-directional fluctuation at the same time point, i.e., the difference between the two sequences at adjacent time points has the same sign. For example, from 9:00 to 12:00 on June 4th, if finished goods warehouse inventory decreases by 20 units and transit warehouse inventory also decreases by 15 units, it is marked as a co-directional fluctuation. The number of co-directional fluctuation periods within the period is then counted and the percentage is calculated. When the percentage exceeds 60%, it is considered a co-directional fluctuation. During this period, the finished goods warehouse and the transit warehouse showed strong trend synergy. Subsequently, during the period of unidirectional fluctuation, the category numbers and inventory change magnitudes involved at each time point were analyzed. The top 3 category numbers with the largest inventory changes were selected and their task tags were extracted. Then, the operation type and inventory change value corresponding to the task tag were traced back. For example, if the inventory of category "P010" decreased by 60 units during a certain period and the operation type was "transit dispatch", then this category was marked as the key fluctuation category for this period. The total inventory change value of this category was then compared with the category's total volume. If the change value percentage exceeded 30%, it was included in the record as a key feature category. All categories that met the conditions were numbered sequentially and labeled with the "key fluctuation" tag to form the output structure of key inventory fluctuation characteristics. Each record in this structure includes the key category number, fluctuation direction, consistent time period, and inventory synergy ratio.

[0100] Please see Figure 5 The batch discrimination module includes:

[0101] The temperature fluctuation judgment submodule analyzes the coating process parameter records based on the key fluctuation characteristics of the inventory, judges the temperature change trend of each batch at the temperature measurement node, compares it with the allowable range of the process standard, judges whether the actual temperature fluctuation meets the process requirements, and obtains the process temperature fluctuation status.

[0102] First, the task tag corresponding to each fluctuating product category is extracted and associated with the temperature acquisition data of the same batch in the coating process parameter record database. This data is recorded by sensors deployed at the temperature measurement nodes in the coating section, with a recording frequency of once per minute. The resulting sequence data consists of timestamps and temperature values. For example, the data recorded by temperature measurement node T001 of a certain batch B20240612 from 10:00 to 10:30 are 165.2, 167.0, 166.5, 168.3, 167.9, 165.8, and 169.2, respectively. By traversing the temperature sequence of this node, the maximum and minimum values ​​are counted and the difference is calculated, which is recorded as the temperature fluctuation range of this batch. This difference is compared with the allowable fluctuation threshold set by the process standard. For example, if the spraying process specifies that the temperature fluctuation should not exceed 4℃ within the range of 165℃ to 170℃, an actual fluctuation of 3.4℃ is considered compliant, while a fluctuation of 5.1℃ is considered non-compliant. The fluctuation values ​​recorded at different temperature measurement nodes are then compared one by one. If a value exceeds the process range, the batch is marked as having an over-temperature record. It is then determined whether this is a process-wide deviation or a momentary anomaly. Specifically, the duration of the fluctuation is calculated. If the temperature exceeds the upper limit for more than 5 consecutive minutes, it is considered a process-wide deviation; otherwise, it is considered a momentary fluctuation. The batch number, temperature measurement node number, maximum fluctuation value, and judgment status are written into the intermediate result table. All batches with a record status of "not meeting process requirements" are summarized and output according to the type of temperature anomaly as the process temperature fluctuation status.

[0103] The time sequence comparison submodule analyzes the start and end times of each batch of equipment operation based on the process temperature fluctuation status, calculates the continuous running time of the equipment, compares the operation process with the operation range specified by the process standard, determines whether the equipment operation process meets the process specifications, and obtains the process operation time sequence status.

[0104] First, extract the batch numbers of all batches with temperature fluctuation records. Then, retrieve the start and end times of the batch's operation in the coating section from the equipment operation log table and convert them into a standard time series structure. For example, if equipment E01 for batch B20240612 starts at 10:00 and stops at 10:38, the operation time is 38 minutes. Next, consult the coating process specification standard table to obtain the required operation time range for this product category in the current process section. If the standard is 35 to 40 minutes, compare whether the batch's operation time falls within the standard range. If it is within the range, mark it as "compliant"; if it is less than 35 minutes or more than 40 minutes, mark it as "non-compliant". Finally, check for any mid-operation shutdowns by checking the equipment status log. If there are records of "idle" or "stopped" status during the operation period, and the cumulative interruption exceeds 3 minutes within the time period, it is recorded as "discontinuous operation". In this case, even if the total time meets the requirements, it is still marked as "non-standard operation". Then, all equipment numbers, batch numbers, start and end times, total running time, and operation status are summarized and output. For batches marked as "non-standard", their temperature fluctuation status is further compared. If the batch has both temperature deviation and abnormal operation time, a "high risk" label is added. For example, B20240612 has a temperature fluctuation of 5.1℃ and an operation time of only 33 minutes, so it is recorded as a process deviation risk batch, forming a structured process operation time sequence status dataset for subsequent batch traceability processing.

[0105] The process deviation archiving submodule filters batches that fail to meet process standards during both temperature fluctuations and equipment operation based on the process operation sequence status. It calculates the deviations from ambient temperature during the operation process, temperature fluctuations, and the overall temperature using the following formula:

[0106]

[0107] Obtain the process anomaly offset, archive parameter information that meets the judgment criteria, and obtain the batch process anomaly sequence, where, This indicates the process deviation of batch b. ZT represents the actual operation duration of batch b. s ΔZθ represents the standard process operation time. (b) Zμ represents the temperature fluctuation range of batch b. θ This represents the average temperature fluctuation range across all batches. ZT represents the ambient temperature at the end of the process section for batch b. m This indicates the ambient temperature at the end of the standard process section.

[0108] Process deviation refers to the overall deviation between the key process parameters (including operation duration, temperature fluctuation range, ambient temperature at the end of the process segment, etc.) of a certain production batch in the coating process of manufacturing and the requirements set by the process standard. Its main function is to identify and screen batches with potential process abnormalities or risks, and to provide data basis for tracing the root cause of abnormalities or optimizing the production process.

[0109] Based on the obtained process operation sequence status, batches that fail to meet both the process standards in terms of temperature fluctuation and operation duration are screened. The operation duration, temperature fluctuation range, and process segment end temperature of each abnormal batch are retrieved and substituted into the formula for calculation:

[0110] Among them, the duration of the operation Data collected from the industrial control system's operation time records; standard operation time ZT s Take 100 minutes, temperature fluctuation amplitude ΔZθ (b) The average fluctuation Zμ of all batches was calculated by measuring the range of temperature measurement nodes for each batch. θ The arithmetic mean of all ΔZθ, and the final ambient temperature. The final temperature of the process segment is collected, and the standard temperature is ZT. m The temperature is fixed at 48.0℃.

[0111] Suppose that for batch B003, the following data is obtained:

[0112] minute;

[0113] ZT s = 100 minutes;

[0114] ΔZθ (b) =4.1℃;

[0115] Zμ θ =4.3℃;

[0116]

[0117] ZT m =48.0℃;

[0118] If we normalize the parameters and then use them, we get:

[0119]

[0120] ZT s =1.0;

[0121] ΔZθ (b) =0.95;

[0122] Zμθ =1.0;

[0123]

[0124] ZT m =1.0;

[0125] Calculate the following items according to the formula: Difference in work duration:

[0126]

[0127] Differences in temperature fluctuation amplitude:

[0128]

[0129] Ambient temperature difference:

[0130]

[0131] Adding the three parts together, we get:

[0132]

[0133] This result indicates that the process anomaly deviation The value of 0.241 represents the overall deviation of batch B003 in terms of operation duration, temperature fluctuation range, and ambient temperature. This value is calculated by comprehensively evaluating the difference between the operation time and the standard operation time, the difference between the temperature fluctuation range and the average temperature fluctuation, and the deviation between the ambient temperature and the standard temperature, reflecting that the process of this batch does not meet the set process standards.

[0134] Please see Figure 6 The parameter traceability module includes:

[0135] The trend recognition submodule analyzes the environmental control parameters before and after the process based on the batch process anomaly sequence, compares the changing trends of temperature, humidity and air pressure in each time period, judges the continuity and consistency of the parameter change direction, and filters the data that show the same changing trend in a period of time to obtain the stage trend sequence.

[0136] First, extract the process segment number and start and end time period corresponding to each abnormal batch. Then, retrieve the temperature, humidity, and air pressure records within that time range from the environmental parameter database. Records are collected in minutes, forming a continuous sequence composed of timestamps and parameter values. For example, the pre-coating stage of batch B20240620 is from 08:30 to 08:55. During this period, the temperature data records are 26.2, 26.5, 26.8, 27.0, 27.4, 27.8, and 28.2, the humidity is 65%, 66%, 67%, 69%, 68%, 66%, and 65%, and the air pressure records are 101.2, 101.3, 101.5, 101.6, 101.7, 101.8, and 101.9 kPa. Construct time series for each of the three types of parameters. Next, determine whether the direction of parameter change in each time period is continuous, such as the temperature value. A continuous increase every minute is considered "continuously rising." If humidity increases in the first half and then decreases in the second half, it is considered "inconsistent in direction." The continuity judgment logic is that the direction of change is consistent for three or more consecutive time points. Each parameter field is processed sequentially, segmented by stage, and the start and end times of the continuous direction interval are recorded. Then, the duration and magnitude of change of each trend direction are calculated. If the three parameters are consistent in direction and all are either rising or falling in a certain segment, the stage is marked as a "consistent trend interval." Otherwise, it is not marked. After filtering the intervals, it is calculated whether the magnitude of the parameter value change in each stage exceeds the process setting value. For example, the temperature change must not exceed 3°C, the humidity must not exceed 10%, and the air pressure must not exceed 0.5 kPa. If any parameter exceeds the range, the trend sequence is excluded, and only the trend stages with compliant magnitudes are retained. All parameter segments that meet the requirements of continuous direction and compliant values ​​are combined into a stage trend sequence.

[0137] The reversal calculation submodule analyzes the changes in the direction of parameter change within each continuous interval based on the phased trend sequence, identifies the nodes where the parameter direction reverses, calculates the number of times the trend reversal occurs, and obtains the set of direction change nodes.

[0138] First, read the parameter direction labels for each trend segment. Then, compare the changes in temperature, humidity, and air pressure within adjacent trend segments to see if there is a reversal from rising to falling or vice versa. Use this direction change as the reversal criterion. Check the combination of parameter direction labels before and after each trend segment. For example, if segment 1 is trending upwards and segment 2 is trending downwards, mark the starting time of segment 2 as a "temperature reversal node." Similarly, perform the same judgment for humidity and air pressure. Record the first, second, and cumulative reversal time nodes for each type of parameter. Next, calculate the cumulative number of direction reversals for each type of parameter throughout the entire trend sequence and categorize them. The total reversal frequency data, for example, shows that in batch B20240620, temperature reversal occurred 3 times, humidity reversal occurred 2 times, and air pressure reversal occurred 1 time. Based on the set threshold, it is judged whether the number of times is too high. The threshold standard is 40% of the total number of continuous trend segments. If there are 8 trend segments, more than 3 times is marked as "frequent reversal". If it does not meet the standard, it is marked as "relatively stable trend". The time position, parameter type and direction change type of each identified reversal node are recorded. At the same time, additional information such as the corresponding trend segment number and the value range before and after the reversal is packaged to generate a complete set of direction change nodes for traceability judgment in subsequent process stages.

[0139] The process aggregation submodule determines the process segment where the reverse node is located based on the set of direction change nodes, identifies the process segment identifier associated with the node, and aggregates the associated process segments to obtain the parameter trend traceability segment.

[0140] First, the timestamp of each reversal node is queried to locate the process execution segment in which that time point falls. Its affiliation is determined by comparing the node time with the start and end times of the process segment in the process schedule. For example, if the reversal time is 09:17, and it falls within the planned time period of the spraying segment from 08:50 to 09:30, then the node belongs to the spraying segment. Subsequently, the parameter type, reversal direction, and previous / next values ​​associated with the node are added to the additional information of that process segment. Finally, all process segments to which the marked reversal nodes belong are uniquely assigned, removing duplicate segment numbers and retaining only the record of the first identification. Simultaneously, the number of reversal nodes in each process segment is counted. If the number of reversal nodes in a certain process segment reaches 3 or more, the segment is determined to be a high-frequency fluctuation segment. The judgment criteria for this condition are set as follows: the threshold is 3 when the total number of nodes is greater than 5, and the threshold is adjusted to 2 when it is less than 5. After the classification is completed, the number, the type of reversal parameter involved, the number of reversal time points and the fluctuation trend type of each process segment are recorded. After being sorted in chronological order, they are merged into a group of traceable structures to form a parameter trend traceability segment. This segment contains the node evolution path, parameter change direction and reversal frequency and location required for visualization.

[0141] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A manufacturing big data management system based on supply chain, characterized in that: The system includes: The event attribution module analyzes changes in business processes based on warehouse transfer instructions, combines material inbound labels and logistics outbound task records to identify changes in action status fields, and attributes each status change to each business event by dynamically counting the task record sequence, thus obtaining the event trigger feature set. The data integration module analyzes the distribution of temperature and humidity monitoring node parameters under task tags based on the event trigger feature set, records real-time inventory quantity changes, monitors vehicle location information collected by the locator, and obtains a multi-source joint dataset. Based on the multi-source joint dataset, the inventory trend module compares the changing trends of inventory data for each category, calculates the synchronicity between the dynamic inventory level of the finished goods warehouse and the periodic changes, analyzes the changes in the inventory of the same category during the same period, determines the correlation between the changes in the turnover inventory of the transit warehouse and the overall trend, and obtains the key fluctuation characteristics of inventory. Based on the key fluctuation characteristics of the inventory, the batch discrimination module analyzes the temperature fluctuations in the coating process parameter records, judges the degree of matching between the equipment operation time and the process standard, compares the parameter fluctuations with the process boundary differences, archives the associated parameters, and obtains the batch process anomaly sequence.

2. The supply chain-based manufacturing big data management system according to claim 1, characterized in that, The event trigger feature set includes event classification coding, event priority labels, and node change descriptions. The multi-source joint dataset includes environmental perception data structures, inventory monitoring records, and logistics tracking entries. The key inventory fluctuation features include category fluctuation indicators, periodic change mapping, and associated status attributes. The batch process anomaly sequence includes batch identification information, abnormal process features, and deviation source annotations.

3. The supply chain-based manufacturing big data management system according to claim 1, characterized in that, The event attribution module includes: The instruction deconstruction submodule, based on warehouse transfer instruction information, analyzes material inbound labels and logistics outbound task content through operation type, target location and time records, judges the logical consistency and process integrity between related information, and calculates the task transfer path between nodes in combination with transfer business links to obtain the transfer status node set. The state discrimination submodule calls the allocation state node set to judge the changes in the action state field, analyze the continuity of the state before and after the task execution, identify the state switching between each business action by comparing the task time and the allocation start node time, identify the key task execution stage, and obtain the state evolution trajectory index. The event mapping submodule analyzes the action state content corresponding to each stage based on the state evolution trajectory index. By analyzing the action type, state switching, and task time information, it determines the business event attribution of each data item, filters priority and classification features, and obtains the event trigger feature set.

4. The supply chain-based manufacturing big data management system according to claim 1, characterized in that, The data integration module includes: The data acquisition and distribution submodule determines the monitoring nodes and environmental parameters corresponding to the task tags based on the event trigger feature set, optimizes the correspondence between the monitoring node numbers and task tags, calculates the distribution of environmental parameters of each node under the different task stages, filters out nodes with fluctuation characteristics in the monitoring node data, and obtains the node parameter distribution density. The inventory trajectory submodule calls the node parameter distribution density and compares it with the inventory change records under each task label to determine the increase or decrease trend of inventory quantity, calculates the inventory fluctuation amplitude in the continuous time series, and obtains the inventory trend structure. The location information integration submodule analyzes vehicle location data within the time period corresponding to the task tag based on the inventory trend structure, filters key spatial trajectory points during inventory fluctuation phases, optimizes the matching relationship between task time and vehicle trajectory, judges the integrity of spatial movement trajectory, and obtains a multi-source joint dataset.

5. The supply chain-based manufacturing big data management system according to claim 1, characterized in that, The inventory trend module includes: The category trend analysis submodule analyzes the inventory monitoring content and task tags based on the multi-source joint dataset, compares the changing trends of the inventory quantity of each category in the same period, judges the frequency and growth of fluctuations, and obtains the category inventory fluctuation frequency. The finished goods synchronization calculation submodule compares the inventory fluctuation frequency of the product category with the finished goods inventory change process to obtain the finished goods inventory trend synchronization degree. The transit correlation judgment submodule determines the correlation between transit warehouse inventory changes and finished goods warehouse trends within key periods based on the finished goods inventory trend synergy. It filters time periods with the same fluctuation direction and synchronous inventory changes, and identifies key category characteristics of the changes to obtain key inventory fluctuation characteristics.

6. The supply chain-based manufacturing big data management system according to claim 1, characterized in that, The batch discrimination module includes: The temperature fluctuation judgment submodule analyzes the coating process parameter records based on the key fluctuation characteristics of the inventory, judges the temperature change trend of each batch at the temperature measurement node, compares it with the allowable range of the process standard, judges whether the actual temperature fluctuation meets the process requirements, and obtains the process temperature fluctuation status. The time sequence comparison submodule analyzes the start and end times of each batch of equipment operation based on the process temperature fluctuation status, calculates the continuous running time of the equipment, compares the operation process with the operation range specified by the process standard, determines whether the equipment operation process meets the process specifications, and obtains the process operation time sequence status. The process deviation archiving submodule filters batches that fail to meet process standard requirements during temperature fluctuations and equipment operation based on the process operation sequence status. It calculates the deviation performance of the operation process, temperature fluctuations and ambient temperature, obtains the process abnormality deviation degree, archives the parameter information that meets the judgment conditions, and obtains the batch process abnormality sequence.

7. The supply chain-based manufacturing big data management system according to claim 1, characterized in that, The system also includes: Based on the batch process anomaly sequence, the parameter traceability module traces the changes in environmental control parameters before and after the process, analyzes equipment operating efficiency and operation duration, judges the continuous trend changes of environmental parameters, collects related process segments, and obtains parameter trend traceability segments. The parameter trend tracing segment includes trend variation nodes, tracing path indexes, and process stage identifiers.

8. The supply chain-based manufacturing big data management system according to claim 7, characterized in that, The parameter traceability module includes: The trend recognition submodule analyzes the environmental control parameters before and after the process based on the batch process anomaly sequence, compares the changing trends of temperature, humidity and air pressure in each time period, judges the continuity and consistency of the parameter change direction, and filters out data that show the same changing trend in a period of time to obtain the stage trend sequence. The reversal calculation submodule analyzes the changes in the direction of parameter change within each continuous interval based on the stage trend sequence, identifies the nodes where the parameter direction reverses, calculates the number of times the trend reversal occurs, and obtains the set of direction change nodes. The process aggregation submodule determines the process segment where the reverse node is located based on the set of directional change nodes, identifies the process segment identifier associated with the node, and aggregates the associated process segments to obtain the parameter trend traceability segment.

Images