A data management method, system, terminal and storage medium of a factory
By aligning timestamps and sorting processes in multi-source data streams from the factory, and combining virtual process queues and waiting timers for monitoring, process anomalies can be identified and corrected. This solves the problem of insufficient accuracy in anomaly detection in existing factory data management systems, and achieves efficient process anomaly identification and data organization.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NINGBO SIJIU MICROELECTRONICS CO LTD
- Filing Date
- 2026-03-03
- Publication Date
- 2026-06-23
Smart Images
Figure CN122264720A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data management, and in particular to a data management method, system, terminal, and storage medium for a factory. Background Technology
[0002] In the manufacturing system, factory data management systems are key technological supports for improving production and operational efficiency and ensuring product quality.
[0003] In related technologies, factory data management systems are based on predefined rules and threshold configurations. They collect equipment operating status and compare it with preset process routes. When data loss or parameter exceeding limits is detected, an alarm is triggered, and the anomaly is ultimately handled by the operator's experience.
[0004] The aforementioned technologies suffer from insufficient accuracy in identifying production anomalies, leading to frequent false alarms from equipment and delays in process adjustments. Summary of the Invention
[0005] In order to improve production and operation efficiency and ensure product quality, this application provides a data management method, system, terminal and storage medium for factories.
[0006] Firstly, this application provides a data management method for a factory, employing the following technical solution: A data management method for a factory includes: Receive raw data streams from multiple data sources in the factory, including at least one of equipment operating status signals, process setting parameters, and online detection results; The original data stream is timestamped to obtain a standardized data sequence; Extract production batch identifiers and process identifiers from standardized data sequences; Based on the mapping relationship between process identifier and process sequence, a process sorting sequence corresponding to the production batch identifier is generated; Based on the process sorting sequence, data items belonging to the same production batch identifier in the standardized data sequence are grouped according to the process order, and the process-level data records corresponding to the production batch identifier are output. Based on the process-level data records, extract the process parameters that have a transfer relationship between adjacent processes, and calculate the deviation values between the process parameters; When the deviation value exceeds the deviation threshold, record the process position and parameter deviation direction of the abnormal process node; Based on the process location and parameter deviation direction, the parameter correction amount is determined from the preset rule base, and the process parameters of abnormal process nodes are corrected.
[0007] By employing the above technical solution, timestamp alignment is performed on the multi-source raw data streams of the factory, and a process sorting sequence is generated based on the mapping relationship between process identifiers and process sequences. This enables process-level grouping of data from the same production batch, and further extracts process parameters with transitive relationships between adjacent processes, calculates deviation values, and determines and executes parameter corrections when deviations exceed limits, based on a preset rule base. This solution significantly improves the accuracy of process anomaly identification and reduces invalid alarms and false interventions.
[0008] Optionally, a virtual process queue is initialized for the current production batch according to the process sorting sequence, and each position in the virtual process queue corresponds to each process in the process sorting sequence in order; When any data item in the standardized data sequence is received, the steps of extracting the production batch identifier and process identifier from the standardized data sequence are executed. If the production batch identifier matches the current production batch, the data item will be stored in the target location corresponding to the process identifier in the virtual process queue. Determine whether there are data items within the interval from the first process to the target position; If not, repeat the above three steps; If so, the data items within the interval will be output to generate process-level data records.
[0009] By employing the above technical solution, production batch identifiers and process identifiers are extracted from the standardized data sequence. A virtual process queue is initialized for the current production batch based on the process sorting sequence. Asynchronously arriving data items are filled into their corresponding positions according to the process sequence. Process-level data records are only output when the data in the interval from the first process to the current process is complete. This solution significantly improves the accuracy of process data organization and reduces misjudgments caused by data disorder or partial missing data.
[0010] Optionally, start a wait timer; While waiting for the timer to start, receive newly arrived data items and update the virtual process queue; While waiting for the timer to reach the time threshold, monitor whether there are any missing positions within the monitoring interval; If so, mark the missing location; Based on the process type corresponding to the missing location and the historical frequency of missing locations, determine the equipment unit associated with the missing location; Record the equipment unit's identification information into the current production batch's anomaly log and generate an equipment inspection task.
[0011] By adopting the above technical solution, a waiting timer is started in the virtual process queue to continuously receive newly arrived data items. After the timeout, the missing location within the process interval is monitored, and the missing location is associated with the equipment unit by combining the process type and historical missing frequency. The equipment unit is then recorded in the anomaly log of the current production batch, and an equipment inspection task is generated. This solution significantly improves the accuracy of process missing identification and reduces the production risks caused by misjudging data delays as actual missing processes or ignoring hidden equipment faults.
[0012] Optionally, after the waiting timer reaches the time threshold, the product model corresponding to the current production batch is obtained, and the process route configuration information is obtained based on the product model; Based on the product model and process route configuration information, generate the expected process sequence corresponding to the current production batch. The expected process sequence contains one or more processes marked as mandatory. Traverse the virtual process queue and identify the positions where data items are empty as candidate missing positions; Determine whether the candidate missing position contains a necessary process in the expected process sequence; If so, mark the candidate missing location as a valid missing location and trigger the exception handling process; If not, the candidate missing positions are ignored and the exception handling process is not triggered.
[0013] By adopting the above technical solution, after the waiting timer expires, the product model of the current production batch is obtained, and a expected process sequence containing mandatory process markers is generated in conjunction with the process route configuration information. Then, empty positions in the virtual process queue are filtered, and when a missing position corresponds to a mandatory process, it is marked as a valid missing position and an exception handling process is triggered. This solution significantly improves the accuracy of process missing determination and reduces invalid alarms and invalid interventions caused by skipping non-critical processes.
[0014] Optionally, query the process adjustment information corresponding to the process route configuration information. The process adjustment information includes optional processes that have been temporarily canceled or skipped. Determine whether the mandatory processes in the expected process sequence include the optional processes in the process adjustment information; If so, remove the optional process from the expected process sequence and record it as a conflict event; Statistical analysis of conflict events is conducted, and it is determined whether the number of statistical occurrences has reached a threshold. If so, then generate suggested modifications to the process route configuration.
[0015] By adopting the above technical solution, the process route configuration information and corresponding process adjustment information of the current product model are queried, the logical consistency between mandatory and optional processes is verified, configuration conflicts caused by incorrectly marking skippable processes as mandatory processes are identified, conflicting processes are removed, and conflict events are recorded. Based on the cumulative statistics of conflict events, process route configuration correction suggestions are generated when a preset threshold is reached. This solution significantly improves the consistency between process configuration and actual production execution, and effectively reduces misjudgments caused by configuration contradictions.
[0016] Optional, identify the data type of the process parameters; Determine if the data type is numeric; If not, obtain the process execution conditions corresponding to the process parameters. The process execution conditions include at least one of the following: the equipment identification, the raw material batch number, and the process specification version. Compare whether the process execution conditions of adjacent processes are consistent; If so, according to the preset hierarchical mapping rules, the non-numerical data of adjacent processes are converted into the first quantization score and the second quantization score respectively, and the deviation value of the first quantization score and the second quantization score is calculated. The hierarchical mapping rules are a correspondence table between non-numerical data and quantization scores. If not, deviation calculations for non-numerical data will not be performed, and process execution condition mismatch events will be recorded.
[0017] By adopting the above technical solution, the data type of process parameters is identified, and when they are determined to be non-numerical, the corresponding process execution conditions are obtained. The system compares the consistency of the execution equipment identification, raw material batch number, or process specification version of adjacent processes. If the conditions are consistent, the non-numerical data is converted into a quantitative score and the deviation is calculated according to a preset hierarchical mapping rule; otherwise, a process execution condition mismatch event is recorded. This solution significantly improves the comparability of non-numerical data and reduces misjudgments caused by ignoring differences in materials, equipment, or versions.
[0018] Optionally, extract multiple attribute dimensions and corresponding attribute values from the non-numerical data. The attribute dimensions include at least two of color, gloss, texture, smell, and touch. The preset hierarchical mapping rules for each attribute dimension are queried separately, and the corresponding attribute values in adjacent processes are converted into multi-dimensional quantization scores, which are then used to construct the first multi-dimensional quantization vector and the second multi-dimensional quantization vector, respectively. Calculate the difference between the first multidimensional quantization vector and the second multidimensional quantization vector to obtain the score deviation corresponding to each attribute dimension; The weighted comprehensive deviation value is obtained by weighting the score deviations of each attribute dimension. If any score deviation exceeds the deviation threshold or the weighted comprehensive deviation value exceeds the global deviation threshold, an alarm indication will be triggered and a multi-dimensional anomaly analysis report will be generated.
[0019] By employing the above technical solution, multiple attribute dimensions such as color, gloss, texture, odor, and tactile feel, along with their corresponding attribute values, are extracted from non-numerical data. Based on pre-defined hierarchical mapping rules for each dimension, the attribute values of adjacent processes are converted into multi-dimensional quantified vectors. The score deviations and weighted comprehensive deviations for each dimension are calculated. When any score deviation or comprehensive deviation exceeds a threshold, an alarm is triggered, and a multi-dimensional anomaly analysis report is generated. This solution significantly reduces reliance on subjective judgment, minimizing missed detections and subjective biases caused by manual visual inspection.
[0020] Secondly, this application provides a factory data management system, which adopts the following technical solution: A factory data management system, comprising: The acquisition module is used to acquire the raw data stream; A memory for storing programs for the data management methods of the factory; The processor and the program in the memory can be loaded and executed by the processor to implement the data management method of the factory.
[0021] By adopting the above technical solution, the acquisition module collects raw data streams from multiple sources in the factory in real time, the processor efficiently executes process-level data organization and anomaly analysis logic, and the memory stores and runs a complete data management program. This achieves fully automated processing from raw data access to the generation of multi-dimensional anomaly alarms and analysis reports. While ensuring the reliability of process monitoring, it significantly improves the detection and response speed of manufacturing process anomalies, providing an efficient and reliable solution for factory operations.
[0022] Thirdly, this application provides a smart terminal, which adopts the following technical solution: A smart terminal includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any of the preceding methods.
[0023] Fourthly, this application provides a computer storage medium capable of storing corresponding programs, which facilitates improvements in production and operational efficiency and ensures product quality. The technical solution adopted is as follows: A computer-readable storage medium storing a computer program that can be loaded by a processor and executed in any of the above-described factory data management methods.
[0024] In summary, this application includes at least one of the following beneficial technical effects: 1. The system aligns the timestamps of multi-source raw data streams from the factory and generates a process sorting sequence based on the mapping relationship between process identifiers and process order. This enables process-level grouping of data from the same production batch. Furthermore, it extracts process parameters with transitive relationships between adjacent processes, calculates deviation values, and determines and executes parameter corrections when deviations exceed limits, based on a pre-defined rule base. This solution significantly improves the accuracy of process anomaly identification and reduces invalid alarms and false alarms. 2. A waiting timer is started in the virtual process queue and continuously receives newly arrived data items. After the timeout, the missing location within the process interval is monitored, and the missing location is associated with the equipment unit by combining the process type and historical missing frequency. The equipment unit is then recorded in the current production batch's anomaly log, and an equipment inspection task is generated. This solution significantly improves the accuracy of process missing identification and reduces the production risks caused by misjudging data delays as actual missing processes or ignoring hidden equipment faults. 3. Identify the data type of process parameters and, if determined to be non-numerical, obtain the corresponding process execution conditions. Compare the execution equipment identification, raw material batch number, or process specification version of adjacent processes to see if they are consistent. If the conditions are consistent, convert the non-numerical parameters into quantitative scores and calculate the deviation according to the preset hierarchical mapping rules; otherwise, record the process execution condition mismatch event. This solution significantly improves the comparability of non-numerical process parameters and reduces misjudgments caused by ignoring differences in materials, equipment, or versions. Attached Figure Description
[0025] Figure 1 This is a flowchart illustrating a factory data management method provided in an embodiment of this application.
[0026] Figure 2 This is a flowchart illustrating a data alignment method based on a virtual process queue provided in an embodiment of this application.
[0027] Figure 3 This is a flowchart illustrating a process defect identification method based on timeout monitoring provided in an embodiment of this application.
[0028] Figure 4 This is a flowchart illustrating a method for determining the absence of a necessary step, as provided in an embodiment of this application.
[0029] Figure 5 This is a flowchart illustrating a process configuration conflict detection and correction method provided in an embodiment of this application.
[0030] Figure 6 This is a flowchart illustrating a method for calculating the deviation of non-numerical process parameters provided in an embodiment of this application.
[0031] Figure 7This is a flowchart illustrating a method for quantization and comparison of multidimensional non-numerical parameters provided in an embodiment of this application.
[0032] Figure 8 This is a schematic diagram of the structure of a factory data management system provided in an embodiment of this application. Detailed Implementation
[0033] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1 to 8 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.
[0034] This application discloses a data management method for a factory. (Refer to...) Figure 1 The method includes: Step S101: Receive raw data streams from multiple data sources in the factory. The raw data streams include at least one of equipment operating status signals, process setting parameters, and online detection results.
[0035] Multiple data sources refer to two or more independent data generating units, such as different types of sensors or detection devices.
[0036] Raw data streams refer to the collection of unprocessed data output from various data collection points on the factory floor in real time or near real time. These data come from diverse sources, have different formats, and may have different time bases.
[0037] Equipment operating status signals refer to electrical or digital signals that reflect the current working status of production equipment, such as the main motor current value of an injection molding machine.
[0038] Process setting parameters refer to process control instructions pre-configured by the operator, such as a heating temperature setting of 220℃.
[0039] Online inspection results refer to process quality data acquired in real time by inspection devices during the production process, such as a vision system detecting a scratch area of 3.2 mm² on the product surface.
[0040] Step S102: Perform timestamp alignment on the original data stream to obtain a standardized data sequence.
[0041] Timestamp alignment refers to rearranging data items from different data sources, with different sampling frequencies or transmission delays, according to their actual occurrence time and grouping them into a unified time grid, so that all related data at the same moment can be correlated and analyzed.
[0042] A standardized data sequence refers to a set of data that has been processed by timestamp alignment, arranged in chronological order, and with each data item having a unified time reference. For example, device A reports temperature every 100ms, while device B reports dimensions every 500ms. At time T=1000ms, the temperature and dimension values at that time are grouped into the same record. For missing dimension data between T=1100ms and T=1400ms, a forward padding method can be used. That is, the dimension value of 25.3mm measured at the most recent valid reporting time, T=1000ms, is used to fill in the missing time points sequentially until the next valid data arrives at T=1500ms, ultimately forming a standardized data sequence with a step size of 100ms.
[0043] Step S103: Extract the production batch identifier and process identifier from the standardized data sequence.
[0044] Production batch identifiers are codes used to uniquely identify a batch of products, and are usually assigned by the MES (Manufacturing Execution System). Process identifiers are codes used to identify a specific processing step, such as WELD-03 indicating the third welding process.
[0045] In standardized data sequences, production batch identifiers and process identifiers are usually embedded in the metadata or payload information of each data item as structured fields. The corresponding field values can be read directly by parsing the data protocol format. For data sources that do not explicitly contain identifiers, they are derived by associating the device ID, workstation number, and pre-stored mapping table, thereby completing the extraction of production batch identifiers and process identifiers.
[0046] Step S104: Based on the mapping relationship between process identifier and process sequence, generate the process sorting sequence corresponding to the production batch identifier.
[0047] A process sequence is an ordered list that arranges all process identifiers corresponding to the current production batch according to the order of process execution, based on the actual processing logic. It is used to reflect the process paths that the batch of products should go through in the factory in sequence.
[0048] Step S105: Based on the process sorting sequence, group the data items belonging to the same production batch identifier in the standardized data sequence according to the process order, and output the process-level data record corresponding to the production batch identifier.
[0049] Process-level data records refer to the data set corresponding to each process, which includes all relevant parameters collected during the execution of that process.
[0050] Grouping is used to separate and collect data generated in different processes within the same production batch according to the actual processing sequence.
[0051] Step S106: Based on the process-level data records, extract the process parameters that have a transfer relationship between adjacent processes, and calculate the deviation value between the process parameters.
[0052] Process parameters refer to quantifiable variables that have physical or logical dependencies between adjacent processes and can affect the execution of subsequent processes or product quality. These include equipment operating status signals, process setting parameters, and online detection results.
[0053] Deviation value refers to the numerical difference of corresponding process parameters in adjacent processes. It is usually the absolute value of the difference between the parameter value of the later process and the parameter value of the earlier process, and is used to quantify the degree of deviation in parameter transfer between processes.
[0054] Step S107: When the deviation value exceeds the deviation threshold, record the process position and parameter deviation direction of the abnormal process node.
[0055] An abnormal process node refers to a process in the process sequence where the process parameters deviate from the previous process by a preset threshold, and is therefore judged to be abnormal.
[0056] The process location refers to the sequential number or identifier of the abnormal process node in the process sequence corresponding to the current production batch, which is used to clarify the specific link in the process flow where the abnormality occurred.
[0057] The direction of parameter deviation refers to whether the deviation of process parameters between adjacent processes is too high or too low, that is, whether the parameter value of the later process is increased or decreased relative to the parameter value of the earlier process.
[0058] Step S108: Based on the process position and parameter deviation direction, determine the parameter correction amount from the preset rule base, and correct the process parameters of abnormal process nodes.
[0059] The parameter correction amount refers to the specific value of the process parameters of abnormal process nodes, which is obtained from the preset rule base according to the process position and the direction of parameter deviation. Its function is to compensate for the deviation and restore process stability.
[0060] By employing the above technical solution, timestamp alignment is performed on the multi-source raw data streams of the factory, and a process sorting sequence is generated based on the mapping relationship between process identifiers and process sequences. This enables process-level grouping of data from the same production batch, and further extracts process parameters with transitive relationships between adjacent processes, calculates deviation values, and determines and executes parameter corrections when deviations exceed limits, based on a preset rule base. This solution significantly improves the accuracy of process anomaly identification and reduces invalid alarms and false interventions.
[0061] This application discloses a data alignment method based on a virtual process queue. (Refer to...) Figure 2 The method includes: Step S201: Initialize a virtual process queue for the current production batch according to the process sorting sequence. Each position in the virtual process queue corresponds to a process in the process sorting sequence in order.
[0062] A virtual process queue is an ordered data container created in memory for the current production batch, where each position corresponds to a process in the process sequence. The purpose of establishing a virtual process queue is to ensure that even when the data arrival order is uncertain or delayed, multi-source asynchronous data can still be collected and its integrity verified according to the process sequence specified by the technology, avoiding analysis errors caused by out-of-order data.
[0063] Step S202: When any data item in the standardized data sequence is received, perform the step of extracting the production batch identifier and process identifier from the standardized data sequence.
[0064] A data item is a basic data unit in a standardized data sequence.
[0065] Step S203: If the production batch identifier matches the current production batch, then store the data item in the target position corresponding to the process identifier in the virtual process queue.
[0066] The data item will then be stored in the target location in the virtual process queue determined by the process identifier.
[0067] The target location refers to the storage location pre-allocated for the process identifier in the virtual process queue according to the process sorting sequence.
[0068] Data items belonging to the current production batch are placed in the corresponding positions in the virtual process queue according to the process identifier, ensuring that the data is stored in the process order.
[0069] Step S204: Determine whether there are data items in the interval from the first process to the target position.
[0070] The purpose of this judgment is to ensure the continuity and completeness of the output process-level data records, and to avoid subsequent deviations due to missing data in intermediate processes.
[0071] Step S205: If not, repeat the above three steps.
[0072] If there is missing process data within the interval, it means that continuous process data has not yet been collected. It is necessary to continue receiving and processing subsequent data items until all data items from the first process to the target position are stored.
[0073] Step S206: If yes, output the data items within the interval to generate process-level data records.
[0074] After confirming that all data items from the first process to the target location exist, the data items of these consecutive processes are packaged and output to form a complete process-level data record.
[0075] By employing the above technical solution, production batch identifiers and process identifiers are extracted from the standardized data sequence. A virtual process queue is initialized for the current production batch based on the process sorting sequence. Asynchronously arriving data items are filled into their corresponding positions according to the process sequence. Process-level data records are only output when the data in the interval from the first process to the current process is complete. This solution significantly improves the accuracy of process data organization and reduces misjudgments caused by data disorder or partial missing data.
[0076] This application discloses a method for identifying process defects based on timeout monitoring. (Refer to...) Figure 3 The method includes: Step S301: Start the wait timer.
[0077] Setting a wait timer is to set a reasonable waiting time limit during the process of receiving data items within a range, to prevent the indefinite postponement of subsequent data output and anomaly judgment due to the delay of individual data, thereby ensuring the timeliness and predictability of the processing flow.
[0078] Step S302: While waiting for the timer to start, receive newly arrived data items and update the virtual process queue.
[0079] During the waiting timer's operation, new data items are continuously received and filled into the corresponding positions of the virtual process queue according to the process identifier.
[0080] Step S303: When the waiting timer reaches the time threshold, monitor whether there are any missing positions within the monitoring interval.
[0081] A missing position refers to a process in the virtual process queue from the first process to the target position where no data item has been filled.
[0082] Monitoring is used to promptly determine whether there are any unreceived data items in the virtual process queue after a timeout, thus avoiding the impact on data processing efficiency due to long waiting times.
[0083] Step S304: If yes, mark the missing positions.
[0084] Marking missing locations is to clearly distinguish between cases where data has not arrived and cases where the process does not need to be executed, in subsequent processing.
[0085] If there are no missing locations within the monitoring interval, it indicates that the data within the interval is complete and there is no need to mark it as missing. The process-level data record can be directly output.
[0086] Step S305: Determine the equipment unit associated with the missing location based on the process type and historical missing frequency corresponding to the missing location.
[0087] Process type refers to the classification name defined by the process route, which indicates the functional attributes of a certain process, such as coating or slitting.
[0088] Historical missing frequency refers to the number or proportion of times data fails to arrive on time within a preset statistical period for a process, and is used to reflect the data stability of related equipment.
[0089] A device unit refers to a physical device or its data acquisition terminal that performs a specific process, such as a temperature sensor on a coating machine.
[0090] Step S306: Record the identification information of the equipment unit into the anomaly log of the current production batch and generate an equipment inspection task.
[0091] Recording the identification information of equipment units in the anomaly log is to establish a correlation between production batches and equipment anomalies, which facilitates subsequent quality traceability and fault analysis.
[0092] Equipment inspection tasks refer to inspection instructions generated for maintenance personnel for specific equipment units, used to promptly identify the causes of missing data.
[0093] By adopting the above technical solution, a waiting timer is started in the virtual process queue to continuously receive newly arrived data items. After the timeout, the missing location within the process interval is monitored, and the missing location is associated with the equipment unit by combining the process type and historical missing frequency. The equipment unit is then recorded in the anomaly log of the current production batch, and an equipment inspection task is generated. This solution significantly improves the accuracy of process missing identification and reduces the production risks caused by misjudging data delays as actual missing processes or ignoring hidden equipment faults.
[0094] This application discloses a method for determining the absence of a necessary process. (Refer to...) Figure 4 The method includes: Step S401: After the waiting timer reaches the time threshold, obtain the product model corresponding to the current production batch, and obtain the process route configuration information based on the product model.
[0095] Product model refers to a unique code used to distinguish different product specifications, such as battery model NCM-18650.
[0096] Process route configuration information refers to the predefined sequence of operations and their attributes, which are bound to a product model. It includes the name of each operation, its execution order, and the corresponding workstation or equipment information. The process route configuration information is retrieved from a pre-stored process database based on the product model.
[0097] Step S402: Based on the product model and process route configuration information, generate the expected process sequence corresponding to the current production batch. The expected process sequence includes one or more processes marked as mandatory.
[0098] The expected process sequence refers to the list of processes that the current production batch should go through in sequence according to the product model and process route configuration information requirements.
[0099] Step S403: Traverse the virtual process queue and determine the positions where data items are empty as candidate missing positions.
[0100] Candidate missing positions refer to the positions in the virtual process queue where no data items have been filled in yet. Further determination is needed to determine whether the position corresponds to a mandatory process that must be executed.
[0101] Step S404: Determine whether the candidate missing position is a necessary process in the expected process sequence.
[0102] The purpose of this judgment is to confirm whether the candidate missing position corresponds to an essential process that cannot be skipped, so as to avoid triggering abnormal handling for non-critical gaps.
[0103] Step S405: If yes, mark the candidate missing location as a valid missing location and trigger the exception handling process.
[0104] The exception handling process refers to the operation of steps S304-S306.
[0105] Step S406: If not, ignore the candidate missing positions and do not trigger the exception handling process.
[0106] Once it is confirmed that the candidate missing position does not correspond to a necessary process, it is considered a normal process skip or a non-critical gap. The position is ignored directly, without marking it or initiating the exception handling process.
[0107] By adopting the above technical solution, after the waiting timer expires, the product model of the current production batch is obtained, and a expected process sequence containing mandatory process markers is generated in conjunction with the process route configuration information. Then, empty positions in the virtual process queue are filtered, and when a missing position corresponds to a mandatory process, it is marked as a valid missing position and an exception handling process is triggered. This solution significantly improves the accuracy of process missing determination and reduces invalid alarms and invalid interventions caused by skipping non-critical processes.
[0108] This application discloses a method for detecting and correcting process configuration conflicts. (Refer to...) Figure 5 The method includes: Step S501: Query the process adjustment information corresponding to the process route configuration information. The process adjustment information includes optional processes that have been temporarily canceled or skipped.
[0109] Process adjustment information refers to temporary changes to the process route configuration information during production, used to record optional processes that have been canceled or skipped. Generally, related process adjustment information is located based on the current product model or production batch identifier.
[0110] The purpose of the query is to obtain temporary changes to the standard process route in the current production process, so as to exclude the absence of processes caused by normal adjustments in subsequent judgments and avoid misjudgments.
[0111] Step S502: Determine whether the mandatory processes in the expected process sequence include the optional processes in the process adjustment information.
[0112] The purpose of this judgment is to check whether there are any processes that are set as mandatory processes in the expected process sequence, but are listed as optional processes and skipped in the process adjustment information, in order to find inconsistencies between the process configuration and the actual execution.
[0113] Step S503: If so, remove the optional process from the expected process sequence and record it as a conflict event.
[0114] A conflict event refers to a situation where the same process is defined as a mandatory process in the process route configuration, but is marked as optional and skipped in the process adjustment information, creating a contradiction in the configuration logic.
[0115] If the required processes in the expected process sequence do not include the optional processes in the process adjustment information, it means that the optional processes in the process adjustment information have not been set as required processes, there is no configuration conflict, the expected process sequence remains unchanged, and no conflict event is recorded.
[0116] Step S504: Statistically analyze the conflict events and determine whether the number of times the statistics are counted has reached the threshold.
[0117] The purpose of counting the number of conflict events is to identify whether there are recurring logical contradictions in the process route configuration of the product model, and to provide a basis for whether the configuration needs to be modified.
[0118] Step S505: If so, generate a suggested modification for the process route configuration.
[0119] When the number of conflict events reaches a preset threshold, a correction suggestion for the process route configuration of the product model is generated, prompting the process that frequently conflicts to be changed from a mandatory process to an optional process or vice versa.
[0120] By adopting the above technical solution, the process route configuration information and corresponding process adjustment information of the current product model are queried, the logical consistency between mandatory and optional processes is verified, configuration conflicts caused by incorrectly marking skippable processes as mandatory processes are identified, conflicting processes are removed, and conflict events are recorded. Based on the cumulative statistics of conflict events, process route configuration correction suggestions are generated when a preset threshold is reached. This solution significantly improves the consistency between process configuration and actual production execution, and effectively reduces misjudgments caused by configuration contradictions.
[0121] This application discloses a method for calculating the deviation of non-numerical process parameters. (Refer to...) Figure 6 The method includes: Step S601: Identify the data type of the process parameters.
[0122] Data type refers to the data format category used for process parameters, including numerical and non-numerical types, used to distinguish between directly calculable quantities and label-type information that needs to be processed through mapping or conditional judgment.
[0123] The data type of process parameters is identified by whether the parameter value consists of pure numbers or numerical values with units. If the parameter value is a number, a decimal, or a string that can be parsed as a number, it is determined to be a numeric type; otherwise, it is determined to be a non-numeric type.
[0124] Step S602: Determine whether the data type is numeric.
[0125] The purpose of this judgment is to differentiate the processing paths. For numerical data, the deviation can be calculated directly, while for non-numerical data, a quantitative evaluation is required in conjunction with the process execution conditions and mapping rules.
[0126] Step S603: If not, obtain the process execution conditions corresponding to the process parameters. The process execution conditions include at least one of the following: the equipment identifier, the raw material batch number, and the process specification version.
[0127] Process execution conditions refer to the specific execution elements bound to process parameters during the production process, such as which equipment was used, which batch of raw materials was used, and which version of the process document was followed.
[0128] The execution equipment identifier refers to the number or code used to uniquely identify specific equipment on the production line, such as the temperature sensor TS-087.
[0129] The raw material batch number refers to the unique number of a batch of raw materials, such as the cathode powder batch number CAM20260115.
[0130] The process specification version refers to the version number of the process document on which the current process is based, such as Liquid Injection Operation Instruction_V3.2.
[0131] The process execution conditions can be obtained by querying the associated equipment number, batch label of the raw materials used, and version number of the process document being executed in the current process record.
[0132] If the data type is numerical, then the calculation of the deviation of parameters between adjacent processes can be performed directly without obtaining the process execution conditions.
[0133] Step S604: Compare whether the process execution conditions of adjacent processes are consistent.
[0134] Comparing process execution conditions is to confirm whether adjacent processes are executed under the same process execution conditions, and to avoid the misjudgment of abnormalities due to differences in non-numerical parameters caused by different equipment, materials or process versions.
[0135] Step S605: If yes, according to the preset hierarchical mapping rule, convert the non-numerical data of adjacent processes into the first quantization score and the second quantization score respectively, and calculate the deviation value of the first quantization score and the second quantization score. The hierarchical mapping rule is a correspondence table between non-numerical data and quantization scores.
[0136] The first quantitative score refers to the value obtained by converting the non-numerical data of the previous process in an adjacent process according to the hierarchical mapping rule, which is used for subsequent deviation calculation.
[0137] The second quantitative score refers to the value obtained by converting the non-numerical data of the next process in an adjacent process according to the same hierarchical mapping rule, and is used to calculate the deviation with the first quantitative score.
[0138] The deviation value refers to the difference between two scores obtained after converting the non-numerical data of the preceding and following processes in adjacent processes. It is used to measure the degree of consistency between the two in process execution.
[0139] Step S606: If not, do not perform deviation calculation for non-numerical data, and record the process execution condition mismatch event.
[0140] When the process execution conditions of adjacent processes are inconsistent, skip the deviation calculation of non-numerical data and record the process execution condition mismatch event.
[0141] By adopting the above technical solution, the data type of process parameters is identified, and when they are determined to be non-numerical, the corresponding process execution conditions are obtained. The system compares the consistency of the execution equipment identification, raw material batch number, or process specification version of adjacent processes. If the conditions are consistent, the non-numerical data is converted into a quantitative score and the deviation is calculated according to a preset hierarchical mapping rule; otherwise, a process execution condition mismatch event is recorded. This solution significantly improves the comparability of non-numerical data and reduces misjudgments caused by ignoring differences in materials, equipment, or versions.
[0142] This application discloses a method for quantization comparison of multidimensional non-numerical parameters. (Refer to...) Figure 7 The method includes: Step S701: Extract multiple attribute dimensions and corresponding attribute values from the non-numerical data. The attribute dimensions include at least two of color, gloss, texture, smell, and touch.
[0143] Attribute dimensions refer to the qualitative description of a certain type of perceptible feature of a product, such as the state recorded from different aspects such as color and texture.
[0144] Attribute values refer to a specific description of the actual state of a product under a certain attribute dimension, such as dark gray under the color dimension.
[0145] The product is observed or measured through manual inspection or specialized testing equipment to obtain the attribute dimensions and their corresponding specific descriptions as attribute values.
[0146] Step S702: Query the preset hierarchical mapping rules for each attribute dimension, convert the corresponding attribute values in adjacent processes into multi-dimensional quantization scores, and form the first multi-dimensional quantization vector and the second multi-dimensional quantization vector respectively.
[0147] The first multidimensional quantization vector refers to a set of quantized scores formed by transforming the attribute values of each attribute dimension in the previous process of adjacent processes according to the corresponding hierarchical mapping rules and arranging them in dimensional order. The first quantization score is a numerical value obtained by transforming a single non-numerical parameter, while the first multidimensional quantization vector contains the quantization results of multiple attribute dimensions and is used for multidimensional joint analysis.
[0148] The second multidimensional quantization vector refers to a set of quantized scores formed by transforming the attribute values of each attribute dimension in the subsequent process of an adjacent process according to the same hierarchical mapping rule and arranging them in the same dimensional order. Similarly, the second multidimensional quantization vector is a quantized combination of multidimensional attributes, while the second quantization score only applies to a single parameter; the two have different dimensions.
[0149] Step S703: Calculate the difference between the first multidimensional quantization vector and the second multidimensional quantization vector to obtain the score deviation corresponding to each attribute dimension.
[0150] Score deviation refers to the difference in quantified scores between adjacent processes on a certain attribute dimension, reflecting the degree of change in the process result under that dimension. For example, if the previous process has color and gloss scores of 70 and 60, and the next process has scores of 75 and 85, then the score deviations for color and gloss are 5 and 25, respectively.
[0151] Step S704: Calculate the weighted comprehensive deviation value by weighting the score deviations of each attribute dimension.
[0152] The weighted average deviation value is the overall deviation index obtained by summing the scores of each attribute dimension according to the importance of each attribute dimension to product quality. For example, if the color deviation is 5 and the weight is 0.4, and the gloss deviation is 20 and the weight is 0.3, then the weighted average deviation value is 5 × 0.4 + 20 × 0.3 = 8.
[0153] Step S705: If any score deviation exceeds the deviation threshold or the weighted comprehensive deviation value exceeds the global deviation threshold, an alarm indication is triggered and a multi-dimensional anomaly analysis report is generated.
[0154] A multi-dimensional anomaly analysis report is an analysis record generated when the score deviation or weighted comprehensive deviation exceeds its respective threshold. It includes the deviation of multiple attribute dimensions and their corresponding process information.
[0155] By employing the above technical solution, multiple attribute dimensions such as color, gloss, texture, odor, and tactile feel, along with their corresponding attribute values, are extracted from non-numerical data. Based on pre-defined hierarchical mapping rules for each dimension, the attribute values of adjacent processes are converted into multi-dimensional quantified vectors. The score deviations and weighted comprehensive deviations for each dimension are calculated. When any score deviation or comprehensive deviation exceeds a threshold, an alarm is triggered, and a multi-dimensional anomaly analysis report is generated. This solution significantly reduces reliance on subjective judgment, minimizing missed detections and subjective biases caused by manual visual inspection.
[0156] Based on the same inventive concept, embodiments of this application provide a factory data management system, see reference. Figure 8 The system includes: Module 801 is used to acquire the raw data stream; Memory 802 is used to store the program for the data management method of the factory; The processor 803 can load and execute programs in memory to implement the data management method of the factory.
[0157] By adopting the above technical solution, the acquisition module collects raw data streams from multiple sources in the factory in real time, the processor efficiently executes process-level data organization and anomaly analysis logic, and the memory stores and runs a complete data management program. This achieves fully automated processing from raw data access to the generation of multi-dimensional anomaly alarms and analysis reports. While ensuring the reliability of process monitoring, it significantly improves the detection and response speed of manufacturing process anomalies, providing an efficient and reliable solution for factory operations.
[0158] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0159] This application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as a data management method for a factory.
[0160] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.
[0161] Based on the same inventive concept, embodiments of this application provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed to implement a factory data management method.
[0162] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0163] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.
Claims
1. A data management method for a factory, characterized in that, include: Receive raw data streams from multiple data sources in the factory, including at least one of equipment operating status signals, process setting parameters, and online detection results; The original data stream is timestamped to obtain a standardized data sequence; Extract production batch identifiers and process identifiers from standardized data sequences; Based on the mapping relationship between process identifier and process sequence, a process sorting sequence corresponding to the production batch identifier is generated; Based on the process sorting sequence, data items belonging to the same production batch identifier in the standardized data sequence are grouped according to the process order, and the process-level data records corresponding to the production batch identifier are output. Based on the process-level data records, extract the process parameters that have a transfer relationship between adjacent processes, and calculate the deviation values between the process parameters; When the deviation value exceeds the deviation threshold, record the process position and parameter deviation direction of the abnormal process node; Based on the process location and parameter deviation direction, the parameter correction amount is determined from the preset rule base, and the process parameters of abnormal process nodes are corrected.
2. The factory data management method according to claim 1, characterized in that, After generating the process sequence corresponding to the production batch identifier based on the mapping relationship between process identifier and process sequence, the method further includes: Based on the process sorting sequence, initialize a virtual process queue for the current production batch. Each position in the virtual process queue corresponds to a process in the process sorting sequence in order. When any data item in the standardized data sequence is received, the steps of extracting the production batch identifier and process identifier from the standardized data sequence are executed. If the production batch identifier matches the current production batch, the data item will be stored in the target location corresponding to the process identifier in the virtual process queue. Determine whether there are data items within the interval from the first process to the target position; If not, repeat the above three steps; If so, the data items within the interval will be output to generate process-level data records.
3. The factory data management method according to claim 2, characterized in that, The method further includes: Start the wait timer; While waiting for the timer to start, receive newly arrived data items and update the virtual process queue; While waiting for the timer to reach the time threshold, monitor whether there are any missing positions within the monitoring interval; If so, mark the missing location; Based on the process type corresponding to the missing location and the historical frequency of missing locations, determine the equipment unit associated with the missing location; Record the equipment unit's identification information into the current production batch's anomaly log and generate an equipment inspection task.
4. The factory data management method according to claim 3, characterized in that, The method further includes: After the waiting timer reaches the time threshold, obtain the product model corresponding to the current production batch, and obtain the process route configuration information based on the product model; Based on the product model and process route configuration information, generate the expected process sequence corresponding to the current production batch. The expected process sequence contains one or more processes marked as mandatory. Traverse the virtual process queue and identify the positions where data items are empty as candidate missing positions; Determine whether the candidate missing position contains a necessary process in the expected process sequence; If so, mark the candidate missing location as a valid missing location and trigger the exception handling process; If not, the candidate missing positions are ignored and the exception handling process is not triggered.
5. The factory data management method according to claim 4, characterized in that, Before generating the expected process sequence corresponding to the current production batch based on product model and process route configuration information, the process also includes: Query the process adjustment information corresponding to the process route configuration information. The process adjustment information includes optional processes that have been temporarily canceled or skipped. Determine whether the mandatory processes in the expected process sequence include the optional processes in the process adjustment information; If so, remove the optional process from the expected process sequence and record it as a conflict event; Statistical analysis of conflict events is conducted, and it is determined whether the number of statistical occurrences has reached a threshold. If so, then generate suggested modifications to the process route configuration.
6. The factory data management method according to claim 1, characterized in that, Before calculating the deviation between process parameters, the method further includes: Identify the data type of process parameters; Determine if the data type is numeric; If not, obtain the process execution conditions corresponding to the process parameters. The process execution conditions include at least one of the following: the equipment identification, the raw material batch number, and the process specification version. Compare whether the process execution conditions of adjacent processes are consistent; If so, according to the preset hierarchical mapping rules, the non-numerical data of adjacent processes are converted into the first quantization score and the second quantization score respectively, and the deviation value of the first quantization score and the second quantization score is calculated. The hierarchical mapping rules are a correspondence table between non-numerical data and quantization scores. If not, deviation calculations for non-numerical data will not be performed, and process execution condition mismatch events will be recorded.
7. A factory data management method according to claim 6, characterized in that, Before converting the non-numerical data at both ends of adjacent processes into first and second quantization scores respectively according to the preset hierarchical mapping rules, the process further includes: Extract multiple attribute dimensions and their corresponding attribute values from non-numerical data. The attribute dimensions include at least two of the following: color, gloss, texture, smell, and touch. The preset hierarchical mapping rules for each attribute dimension are queried separately, and the corresponding attribute values in adjacent processes are converted into multi-dimensional quantization scores, which are then used to construct the first multi-dimensional quantization vector and the second multi-dimensional quantization vector, respectively. Calculate the difference between the first multidimensional quantization vector and the second multidimensional quantization vector to obtain the score deviation corresponding to each attribute dimension; The weighted comprehensive deviation value is obtained by weighting the score deviations of each attribute dimension. If any score deviation exceeds the deviation threshold or the weighted comprehensive deviation value exceeds the global deviation threshold, an alarm indication will be triggered and a multi-dimensional anomaly analysis report will be generated.
8. A factory data management system, characterized in that, The system is used to execute the factory data management method as described in any one of claims 1 to 7, including: The acquisition module is used to acquire the raw data stream; A memory for storing programs for the data management methods of the factory; The processor and the program in the memory can be loaded and executed by the processor to implement the data management method of the factory.
9. A smart terminal, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and execute the method as described in any one of claims 1 to 7.