Process quality real-time interception method, system, device and storage medium

Abstract

The present application relates to a process quality real-time interception method, system, device and storage medium, the method comprising real-time acquisition of quality detection data output by a sensing device on a production line, obtaining production work order information associated with the quality detection data, and integrating the quality detection data and the production work order information to obtain a quality event object; inputting the quality event object into a multi-rule analysis channel for abnormality calculation to obtain an abnormality confidence; constructing an interception according to the abnormality confidence, the quality event object and a preset interception strategy library to obtain a hierarchical interception instruction; obtaining interception feedback information after executing the hierarchical interception instruction, integrating the interception feedback information with the production work order information to obtain trace record information. The present application generates a quality event object by real-time acquisition and integration of production line data and production context, thereby overcoming the interception delay problem caused by lag comparison in traditional methods.

Description

Technical Field

[0001] This invention relates to the technical field of production line monitoring, and in particular to a method, system, device, and storage medium for real-time interception of process quality. Background Technology

[0002] In the context of the manufacturing industry's transformation towards digitalization and intelligence, real-time monitoring and proactive interception of process quality during production are crucial for ensuring product quality. However, current widely adopted quality control methods largely rely on lagging comparisons with fixed thresholds or periodic statistical analyses, making it difficult to respond instantly to defects, leading to the continued circulation of defective products. Existing technologies typically process detection signals in isolation, failing to effectively correlate them with dynamically changing production tasks, process parameters, and production line resources, resulting in insufficient judgment criteria and limited accuracy. More importantly, most existing solutions stop at issuing alarms, unable to automatically generate precise instructions and drive field equipment to perform differentiated intervention actions, making it difficult to achieve stable and reliable real-time intervention. Summary of the Invention

[0003] The main objective of this invention is to provide a method, system, device, and storage medium for real-time interception of process quality. By collecting and integrating production line data and production context in real time to generate quality event objects, it can realize the response from signal perception to anomaly determination, thereby overcoming the interception delay problem caused by the delayed comparison in traditional methods.

[0004] To achieve the above objectives, the present invention provides a method for real-time interception of process quality, comprising: The system collects quality inspection data output from sensing devices on the production line in real time, obtains production work order information associated with the quality inspection data, and integrates the quality inspection data and the production work order information to obtain a quality event object. The quality event object is input into the multi-rule analysis channel for anomaly calculation to obtain the anomaly confidence level. Based on the anomaly confidence level, the quality event object, and the preset interception strategy library, an interception system is constructed to obtain a hierarchical interception instruction. Obtain the interception feedback information after executing the hierarchical interception command, and integrate the interception feedback information with the production work order information to obtain traceability record information.

[0005] Furthermore, the real-time acquisition of quality inspection data output by sensing devices on the production line, obtaining production work order information associated with the quality inspection data, and integrating the quality inspection data and the production work order information to obtain quality event objects, including: The quality inspection data is read to obtain the device identifier, measurement value, and acquisition timestamp; Using the device identifier as the query key, query the manufacturing execution system for the corresponding work order number and current process code; Based on the work order number, the manufacturing execution system queries the corresponding material batch information and operator information; The work order number, the current process code, the material batch information, and the operator information are integrated into the production work order information; The quality event object is obtained by combining the production work order information, the equipment identifier, the measurement value, and the collection timestamp.

[0006] Further, the step of inputting the quality event object into a multi-rule analysis channel for anomaly calculation to obtain anomaly confidence includes: Identify the real-time measurement value queue and process condition set from the quality event objects; The real-time measurement value queue is matched with each anomaly judgment rule of the multi-rule analysis channel. If the measurement value of the real-time measurement value queue falls into the corresponding anomaly judgment rule, the anomaly type corresponding to the anomaly judgment rule is marked as a violation instance. The number of all the aforementioned violation instances is summarized, and the violation ratio is calculated based on the real-time measurement value queue and the number of instances to obtain the initial violation ratio; The initial violation rate is corrected based on the set of process conditions to obtain the corrected violation rate; The corrected violation ratio is mapped to a preset confidence quantification range to obtain the anomaly confidence level.

[0007] Further, the step of constructing a tiered interception instruction based on the anomaly confidence level, the quality event object, and a preset interception strategy library includes: The anomaly confidence level is compared with a preset first confidence threshold and a second confidence threshold, wherein the first confidence threshold is less than the second confidence threshold. If the anomaly confidence level is less than the first confidence threshold, it is determined that there is no anomaly and no interception command is generated. When the anomaly confidence level is higher than the first confidence threshold and lower than the second confidence threshold, a first-level interception instruction is constructed based on the quality event object and the interception strategy library. When the anomaly confidence level is higher than the second confidence threshold, a second-level interception instruction is constructed based on the quality event object and the interception strategy library. Based on the second-level interception instruction, the quality event object is matched and compared according to the associated abnormal conditions in the interception strategy library, and a third-level interception instruction is generated based on the matching result and the second-level interception instruction; wherein, the hierarchical interception instruction includes the first-level interception instruction, the second-level interception instruction and the third-level interception instruction.

[0008] Furthermore, when the anomaly confidence level is higher than the second confidence threshold, a second-level interception instruction is constructed based on the quality event object and the interception strategy library, including: Extract event feature parameters from the quality event object, and extract the secondary instruction generation template corresponding to the event feature parameters from the interception strategy library; Based on the parameter mapping relationship in the secondary instruction generation template, the event feature parameters are assigned to the instruction parameter items in the secondary instruction generation template to generate a basic instruction parameter set; The instruction parameter set is optimized for interception based on the secondary interception measures in the secondary instruction generation template to obtain the interception instruction parameter set. The interception instruction parameter set is filled into the instruction structure of the secondary instruction generation template to generate a secondary basic instruction unit; Based on the quality event object, all the secondary basic instruction units are encapsulated and combined to obtain the second-level interception instruction.

[0009] Further, the step of matching and comparing the quality event object according to the associated abnormal conditions in the interception strategy library based on the second-level interception instruction, and generating a third-level interception instruction based on the matching result and the second-level interception instruction, includes: Extract the associated anomaly conditions related to the quality event object from the interception policy library, wherein the associated anomaly conditions contain at least one association rule; Traverse each of the association rules in the associated anomaly conditions, and match the event identifier of the quality event object with the historical anomaly event identifiers in the association rules one by one; When a match is successful, the association rule is marked as a valid association rule, and the upgrade action identifier corresponding to the valid association rule is extracted; If the match fails, the second-level interception command will be output directly, and the generation of the third-level interception command will be stopped. If at least one of the valid association rules exists, the second-level interception command is updated with instruction enhancement based on the upgrade action identifier corresponding to all the valid association rules, to obtain the third-level interception command.

[0010] Further, the step of obtaining the interception feedback information after executing the hierarchical interception command, and integrating the interception feedback information with the production work order information to obtain traceability record information includes: The interception feedback information is analyzed to obtain the instruction execution result, the interception execution timestamp, and the execution device identifier; If the result of the instruction execution indicates that the interception failed, then an interception failure status flag is generated; The interception failure status identifier is combined with the work order information to construct an interception failure record, thereby generating a warning record information; If the instruction execution result indicates successful interception, then the interception execution timestamp is compared with the pre-stored standard process time to calculate the interception delay duration. The instruction execution result, the interception execution timestamp, the execution device identifier, the interception delay duration, and the production work order information are associated and bound to generate the traceability record information.

[0011] The present invention also provides a real-time process quality interception system, applied to any of the above-described real-time process quality interception methods, comprising: The acquisition module is used to acquire quality inspection data output by sensing devices on the production line in real time, obtain production work order information associated with the quality inspection data, and integrate the quality inspection data and the production work order information to obtain a quality event object. The analysis module is used to input the quality event object into the multi-rule analysis channel for anomaly calculation and to obtain the anomaly confidence level. The association module is used to construct an interception based on the anomaly confidence level, the quality event object, and a preset interception strategy library, and obtain a hierarchical interception instruction. The processing module is used to obtain the interception feedback information after executing the hierarchical interception instruction, and integrate the interception feedback information with the production work order information to obtain traceability record information.

[0012] The present invention also provides a real-time process quality interception device, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the steps of a real-time interception method for process quality as described in any of the preceding claims.

[0013] The present invention also provides a storage medium storing computer instructions for causing a computer to perform any of the methods described above.

[0014] The present invention provides a method, system, device, and storage medium for real-time interception of process quality, which has the following beneficial effects: By collecting and integrating production line data and production context in real time to generate quality event objects, a response from signal perception to anomaly determination can be achieved, overcoming the interception delay problem caused by lagging comparison in traditional methods. By inputting quality event objects into parallel multi-rule analysis channels for fusion calculation, multi-dimensional information such as static specifications, dynamic processes, and correlation strategies can be comprehensively utilized, significantly improving the accuracy and reliability of anomaly judgment and solving the problems of false alarms and missed alarms caused by isolated data and weak basis. By dynamically constructing hierarchical interception instructions based on anomaly confidence and combining them with a strategy library, differentiated automatic interventions from early warning and locking to process correction can be directly driven on-site equipment, reducing reliance on human experience and achieving standardization and immediacy of interception actions. By integrating interception feedback and production information in real time to generate structured traceability records, the decision-making and execution closed loop can be completely recorded. Attached Figure Description

[0015] Figure 1 This is a flowchart of a real-time interception method for process quality provided by the present invention; Figure 2 This is a structural diagram of a real-time process quality interception device provided by the present invention; Figure 3 This is a structural diagram of a real-time process quality interception system provided by the present invention.

[0016] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] 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.

[0018] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0019] Reference Figure 1 As shown, the present invention provides a method for real-time interception of process quality, comprising: Step S1: Collect quality inspection data output by sensing devices on the production line in real time, obtain production work order information associated with the quality inspection data, and integrate the quality inspection data and the production work order information to obtain a quality event object; Specifically, various sensing devices deployed on the production line, such as high-precision laser measuring instruments, machine vision inspection units, or PLC I / O modules integrated into CNC equipment, continuously generate quality inspection data containing measurement values, unique equipment codes, and timestamps according to a preset sampling frequency. This data stream is aggregated through an industrial IoT edge gateway or a dedicated data acquisition server and transmitted to the processing core via standardized industrial communication protocols.

[0020] Upon receiving quality inspection data, a synchronous context query to the Manufacturing Execution System (MES) is immediately initiated based on the unique equipment code. Using the pre-defined mapping between equipment codes and physical workstations, the specific production station from which the data originated is located. Then, the production task instructions currently actively executing at that station are retrieved from the MES's real-time task queue. The retrieved production work order information is a collection of multiple key production elements, including at least a unique work order tracking number, the code of the standard process being executed, the batch identifier of the currently input materials, and the operator's verified identity information.

[0021] After completing the context binding, perform the data integration operation, logically associate and physically encapsulate the measurement values, collection timestamps, and equipment codes from the perception layer with the work order numbers, process codes, material batches, and operator identifiers from the planning layer, and output a quality event object.

[0022] Step S2: Input the quality event object into the multi-rule analysis channel for anomaly calculation to obtain the anomaly confidence level; The multi-rule analysis channel comprises at least three independently operating and logically complementary analysis channels: a specification limit determination channel, a statistical process control channel, and a correlation strategy channel. Each analysis channel receives a complete quality event object as input and performs independent calculations and judgments based on its own built-in logic.

[0023] Specifically, the specification limit judgment channel extracts the preset upper and lower tolerance limits of the key quality characteristics of the current process from the process standard domain of the quality event object, and simultaneously extracts the real-time measurement value from the data acquisition domain. The statistical process control channel, based on the work order number and process code in the quality event object, retrieves the most recent dozens of consecutive measurement values ​​under the same production conditions from the historical database, dynamically constructing a process behavior sample window; then, it calculates the statistical center line and control limits of this window, and applies preset trend rules to analyze the position and sequence relationship of the real-time measurement value within the window, aiming to identify potential abnormal trends in the process. The association strategy channel, based on a specific resource combination in the production context domain of the quality event object, such as the combination key "equipment A -- mold B -- material batch C", queries a preset association rule knowledge base; if a specific rule is matched, a secondary evaluation is performed on the measurement value or preliminary judgment according to the additional conditions or correction coefficients defined in the rule.

[0024] After each independent channel completes its analysis, it outputs a binary anomaly determination and a confidence score. The outputs from all channels are aggregated, and based on a preset weight configuration, the three confidence scores are weighted and summed to finally generate an anomaly confidence value.

[0025] Step S3: Based on the anomaly confidence level, the quality event object, and the preset interception strategy library, construct an interception system to obtain a tiered interception instruction; Specifically, the anomaly confidence level is compared with two key thresholds in the interception strategy library: a first confidence threshold and a second confidence threshold that is higher than the first confidence threshold. If the anomaly confidence level is lower than the first threshold, the process terminates and no instructions are generated to avoid unnecessary interference with the production process.

[0026] When the anomaly confidence level falls between the first and second confidence thresholds, it is determined to be a low-risk anomaly. Immediately, a first-level instruction is constructed. Based on the process identifier and anomaly characteristics in the quality event object, a preset first-level instruction template is retrieved from the policy library. This first-level instruction template is an instruction framework whose parameter bits are bound to specific data fields (such as equipment identifiers and Kanban area numbers) in the quality event object through mapping relationships. Specific identifiers and numbers are filled into the template to generate standardized first-level interception instructions such as "Send a yellow flashing warning to the XX Kanban" or "Push a verification prompt to the PDA."

[0027] If the anomaly confidence level exceeds the second confidence threshold, it indicates a high-risk anomaly, initiating the construction of a secondary instruction. A more complex secondary instruction template is invoked to extract the precise device identifier from the quality event object and query the device-controller address mapping table to determine the specific receiving endpoint of the instruction. Event characteristic parameters (such as the specific value of the deviation) and the target controller address are then filled into the template to generate binding secondary basic instruction units such as "Send a station lock signal to PLC XX" or "Stop the operation of conveyor line XX". These units are combined and encapsulated to form a clear, immediately effective secondary interception instruction.

[0028] Building upon this, a correlation check is further performed as a decision point for generating the third-level instruction. Based on information such as material batches in the quality event object, the associated abnormal conditions in the strategy library are traversed and matched with historical interception records. If it is found that the same batch of materials triggers multiple similar interceptions within a short period of time, a systemic risk is determined. At this point, the instruction is strengthened and updated based on the second-level instruction, and a third-level process correction instruction is generated, such as "insert a full inspection station into the preceding process Y" or "suspend the use of batch XX of materials".

[0029] Step S4: Obtain the interception feedback information after executing the hierarchical interception command, and integrate the interception feedback information with the production work order information to obtain traceability record information.

[0030] Specifically, after interception commands at all levels are issued to the production line, the system listens for interception feedback information from the execution endpoints. These endpoints can be programmable logic controllers, Andon systems, or automated guided vehicles, which return interception feedback information containing the execution status code, the action completion timestamp, and their own equipment code after executing the action.

[0031] Upon receiving intercepted feedback information, it is parsed and correlated. By matching the transaction ID in the intercepted feedback information with the characteristics of the original instruction, it is associated back to the original quality event object and the corresponding production work order. If the feedback status code indicates that the instruction execution failed, such as the equipment lock signal not being successfully received, a specific interception failure status identifier is generated, and an immediate warning record is triggered and sent to maintenance personnel.

[0032] If the feedback status code indicates successful interception, a performance evaluation is initiated. The actual interception execution timestamp is compared with the standard cycle time of the process associated with the quality event object to calculate the actual delay time consumed by this interception. Regardless of whether the feedback is successful or failed, this result—including the execution result, timestamp, equipment identifier, calculated delay time, or failure identifier—is deeply integrated with the complete production context information in the quality event object.

[0033] The data is restructured according to a predefined data model. This model forcibly divides the data into an event triggering layer, an analysis and decision-making layer, an instruction execution layer, and a production context layer, ensuring that every traceability record generated is consistent.

[0034] This invention provides a real-time process quality interception method. By collecting and integrating production line data and production context in real time to generate quality event objects, it can achieve a response from signal perception to anomaly determination, thus overcoming the interception delay problem caused by lagging comparison in traditional methods. By inputting quality event objects into parallel multi-rule analysis channels for fusion calculation, it can comprehensively utilize multi-dimensional information such as static specifications, dynamic processes, and correlation strategies, significantly improving the accuracy and reliability of anomaly judgment and solving the problems of false alarms and missed alarms caused by isolated data and weak basis. By dynamically constructing hierarchical interception instructions based on anomaly confidence and combined with a strategy library, it can directly drive field equipment to perform differentiated automatic interventions from early warning and locking to process correction, reducing reliance on human experience and achieving standardization and immediacy of interception actions. By integrating interception feedback and production information in real time to generate structured traceability records, it can completely record the decision-making and execution closed loop.

[0035] In some embodiments, the real-time acquisition of quality inspection data output by sensing devices on the production line, obtaining production work order information associated with the quality inspection data, and integrating the quality inspection data and the production work order information to obtain a quality event object includes: The quality inspection data is read to obtain the device identifier, measurement value, and acquisition timestamp; Using the device identifier as the query key, query the manufacturing execution system for the corresponding work order number and current process code; Specifically, the system uses the device identifier as the primary query condition to access the device-workstation configuration mapping table. This table defines which production workstation or workstation each sensing device logically belongs to. The device identifier is converted into a workstation identifier through the mapping table query. After obtaining the workstation identifier, a second key query is initiated, targeting the work order execution status table. This table retrieves the currently scheduled production task records that are being executed at that specific workstation, based on the workstation identifier and status (e.g., "running" or "occupied"). Two fields are extracted from the returned records: work order number and current process code. The query results, along with the original device identifier, are temporarily stored. If the query fails (e.g., the workstation is displayed as idle or there are no valid work orders), the quality inspection data stream is considered invalid background noise and discarded, and the process terminates.

[0036] Based on the work order number, the manufacturing execution system queries the corresponding material batch information and operator information; Specifically, using the work order number, the material consumption association table for work order 1 is accessed first. This table records the planned and actual material batches consumed at each process point in the process route for a specific production work order. The query conditions are the current work order number and the current process code, which retrieve the specific material batch number pre-assigned or actually scanned and put into operation for this process. Using the work order number and the current process code as conditions, the personnel task assignment table for work order 2 is accessed. This table dynamically records the identities of operators authorized to perform specific process tasks at specific workstations within a specific time window. The unique identifier of the operator currently in an active working state is retrieved. These two queries are sequential and both rely on the real-time relational data model maintained by the manufacturing execution system. The obtained material batch number and operator employee number, as raw data, are sent to a lightweight verification step. For example, it verifies whether the material batch has been released for production or whether the operator is qualified to operate the process. After successful verification, the material batch information and operator information are extracted from the query result set and temporarily stored.

[0037] The work order number, the current process code, the material batch information, and the operator information are integrated into the production work order information; The quality event object is obtained by combining the production work order information, the equipment identifier, the measurement value, and the collection timestamp.

[0038] The method provided in this implementation uses equipment identifiers as indexes to query the Manufacturing Execution System in real time, dynamically binding quality data with ongoing production work orders and processes. This ensures that each inspection is linked to a precise production context, preventing data from becoming disconnected from production tasks. Further, by querying material batches and operator information, and integrating all production elements and raw inspection data into a unified quality event object, the method achieves deep integration of perceived data and business information. This provides previously isolated quality signals with clear traceability paths, fundamentally solving the problem of insufficient analytical basis caused by scattered data sources and lack of background information.

[0039] In some embodiments, inputting the quality event object into a multi-rule analysis channel for anomaly calculation to obtain anomaly confidence includes: Identify the real-time measurement value queue and process condition set from the quality event objects; The real-time measurement value queue is matched with each anomaly judgment rule of the multi-rule analysis channel. If the measurement value of the real-time measurement value queue falls into the corresponding anomaly judgment rule, the anomaly type corresponding to the anomaly judgment rule is marked as a violation instance. Specifically, multiple independent and parallel rule analysis channels are activated. Each channel is instantiated and loaded with its corresponding rule set and relevant parameters obtained from the process condition set. For example, the specification limit judgment channel loads the upper and lower tolerance limits; the statistical process control channel loads the historical data window size and control limit calculation coefficients. For each measurement value in the real-time measurement value queue, each channel independently executes its rule matching logic. In the specification limit judgment channel, numerical comparisons are performed directly to determine whether the current measurement value falls outside the acceptable range defined by the upper and lower limits. In the statistical process control channel, the current measurement value is placed in a control chart containing recent historical measurement values, and the value's position relative to the control chart is evaluated. In the association strategy channel, it is checked whether the current resource combination (equipment-mold-material) has special, more stringent tolerance requirements. Each matching calculation produces a judgment result: triggered or not triggered. Once a measurement value triggers a rule in any channel, a violation instance record is immediately created. This record contains the identifier of the triggering channel, the specific type identifier of the triggering rule, the measurement value that caused the trigger or its index, and the core identifier of the object of this quality event for traceability.

[0040] The number of all the aforementioned violation instances is summarized, and the violation ratio is calculated based on the real-time measurement value queue and the number of instances to obtain the initial violation ratio; Specifically, the total number of violation instances is obtained by summing the number of records. The input data—the length of the real-time measurement queue, i.e., the total number of measurements in the queue—is obtained. This value represents the total number of observation samples on which the rule is based. The violation ratio calculation is triggered. The accumulated number of violation instances is used as the numerator, and the total length of the measurement queue is used as the denominator, and a division operation is performed to generate an initial violation ratio. For example, if 2 out of 10 measurements trigger a rule, the ratio is 0.2. This calculation process implicitly assumes that the events triggering violations for each measurement after it is sent to the multi-rule analysis channel are considered independent and equally important. The initial violation ratio is output as an intermediate value.

[0041] The initial violation rate is corrected based on the set of process conditions to obtain the corrected violation rate; The corrected violation ratio is mapped to a preset confidence quantification range to obtain the anomaly confidence level.

[0042] The method provided in this implementation, by running multiple rule analysis channels in parallel, independently determines real-time measured values ​​from multiple dimensions using specifications, statistics, and correlation strategies. It integrates absolute standards, process trends, and specific production contexts, significantly improving the comprehensiveness and accuracy of anomaly identification and avoiding the limitations of a single rule perspective. By aggregating violation instances triggered by each channel and calculating the initial violation ratio, discrete anomaly events are transformed into continuous risk frequency indicators, providing a quantifiable preliminary assessment basis for quality status and enhancing the objectivity of the judgment. By mapping the corrected ratio to a preset confidence quantification interval, a standardized anomaly confidence score is finally output, unifying the results of analyses from different dimensions.

[0043] In some embodiments, the step of constructing a tiered interception instruction based on the anomaly confidence level, the quality event object, and a preset interception strategy library includes: The anomaly confidence level is compared with a preset first confidence threshold and a second confidence threshold, wherein the first confidence threshold is less than the second confidence threshold. If the anomaly confidence level is less than the first confidence threshold, it is determined that there is no anomaly and no interception command is generated. When the anomaly confidence level is higher than the first confidence threshold and lower than the second confidence threshold, a first-level interception instruction is constructed based on the quality event object and the interception strategy library. Specifically, the interception strategy library is accessed based on key attributes (such as process code, equipment identifier, and defect feature code) in the quality event object as a composite query key. The library pre-stores corresponding first-level instruction templates for different processes and risk types. This template is a parameterized instruction framework containing the instruction type (such as a Kanban alert), the target address (such as the ID or network address of a Kanban screen), and several placeholder parameters to be filled. The first-level instruction template is parsed, and the placeholders are instantiated and assigned values ​​based on the specific information in the quality event object. For example, the specific value of the equipment identifier is filled into the target workstation field of the instruction, and the specific value of the measurement deviation is filled into the text template of the prompt message. Lightweight instruction optimization is performed, such as adjusting the list of personnel receiving the prompt message based on the current shift information. One or more first-level instruction data packets are output.

[0044] When the anomaly confidence level is higher than the second confidence threshold, a second-level interception instruction is constructed based on the quality event object and the interception strategy library. Based on the second-level interception instruction, the quality event object is matched and compared according to the associated abnormal conditions in the interception strategy library, and a third-level interception instruction is generated based on the matching result and the second-level interception instruction; wherein, the hierarchical interception instruction includes the first-level interception instruction, the second-level interception instruction and the third-level interception instruction.

[0045] The method provided in this implementation accurately classifies quality risks by comparing anomaly confidence levels with preset first and second thresholds. This effectively distinguishes between different scenarios such as no anomalies, low-risk warnings, and high-risk situations requiring immediate intervention, avoiding unnecessary interference with the production process and ensuring the accuracy and efficiency of interception decisions. By constructing differentiated first- and second-level interception instructions based on quality event objects and a strategy library, a tiered response from non-intrusive visual warnings to mandatory equipment locking is achieved. This allows low-risk events to be alerted without interrupting production, while high-risk events are immediately physically intercepted, balancing production continuity and quality safety. By matching and comparing quality events based on associated anomaly conditions and generating extended third-level interception instructions accordingly, potential systemic risks behind a single anomaly can be identified. Control measures are extended from the point of occurrence to related processes or logistics links, achieving an upgrade from point-based interception to line-based preventative control, significantly improving the depth and foresight of quality control.

[0046] In some embodiments, when the anomaly confidence level is higher than the second confidence threshold, a second-level interception instruction is constructed based on the quality event object and the interception strategy library to obtain a second-level interception instruction, including: Extract event feature parameters from the quality event object, and extract the secondary instruction generation template corresponding to the event feature parameters from the interception strategy library; Specifically, based on predefined extraction rules, the device identifier and measured value are obtained from the data acquisition domain, the process code and work order number are obtained from the production context domain, and the anomaly confidence level and the main rule channel identifier and specific anomaly type code that triggered this judgment are obtained from the associated analysis result domain. These discrete data items are combined into an event feature parameter. Based on this event feature parameter as a composite query key, the secondary instruction template index in the interception strategy library is accessed. For example, the feature parameter "process code: welding-001" and "anomaly type: current exceeds upper limit" are mapped to a template ID named "welding current anomaly - equipment lock" in the index. Based on this ID, the corresponding secondary instruction generation template is loaded from the template storage area. The content of this template clearly defines the fixed parts (such as instruction type code, target device communication protocol header) and variable parts (i.e., the instruction parameter items to be filled and their data format) required for subsequent instruction construction.

[0047] Based on the parameter mapping relationship in the secondary instruction generation template, the event feature parameters are assigned to the instruction parameter items in the secondary instruction generation template to generate a basic instruction parameter set; The instruction parameter set is optimized for interception based on the secondary interception measures in the secondary instruction generation template to obtain the interception instruction parameter set. Specifically, the basic instruction parameter set is used as the main input, and the basic instruction parameter set is parsed again with reference to the secondary interception measures defined in the secondary instruction generation template, and pre-configured optimization measures are executed. The optimization measures are targeted at specific scenarios and include the following types: The first type is parameter verification and correction. For example, the measure may stipulate that if the device lock-up duration parameter exceeds the preset safety maximum value, it will be automatically corrected to that maximum value to prevent prolonged production line stoppage due to missetting. The second type is condition supplementation. For example, the measure may stipulate that when the target device is in automatic mode, an additional instruction parameter requesting mode switching to manual mode needs to be added to the parameter set; if the device is already in manual mode, this parameter is not needed. The third type is derived parameter generation. For example, the measure may determine the calibration process code to be executed after device reset by using a lookup table method based on the difference between abnormal measurement values ​​and standard upper limits, and add this code as a new parameter to the set. The fourth type is dependent parameter parsing. For example, the measure stipulates that the target device identifier in the basic parameters needs to be further parsed into two lower-level parameters: the master controller address and the slave module slot number.

[0048] The system sequentially applies all relevant rules to iterate through and process the basic instruction parameter set. Each rule application may modify, delete, or add new key-value pairs to parameter values. The entire process ensures logical consistency between parameters. Upon completion, the output is a set of interception instruction parameters.

[0049] The interception instruction parameter set is filled into the instruction structure of the secondary instruction generation template to generate a secondary basic instruction unit; Based on the quality event object, all the secondary basic instruction units are encapsulated and combined to obtain the second-level interception instruction.

[0050] The method provided in this implementation extracts specific event feature parameters from quality event objects and matches them with corresponding secondary instruction generation templates. This allows the instruction construction process to be highly customized based on specific contexts such as anomaly type, equipment, and process, ensuring that the generated interception instructions accurately correspond to the actual on-site situation and improving the instructions' relevance and effectiveness. By converting feature parameters into specific instruction parameter sets based on the parameter mapping relationships in the template, and further applying interception optimization rules for parameter verification, supplementation, and adjustment, the final instruction parameters not only accurately reflect anomaly characteristics but also adapt to the real-time status of the equipment and safety constraints, significantly enhancing the accuracy, security, and context adaptability of the instructions. By filling the optimized instruction parameter sets into a predefined instruction structure to generate standardized basic instruction units, it ensures that each output instruction strictly adheres to the communication protocol specifications of the target industrial controller, allowing it to be directly parsed and executed by the equipment. This eliminates the protocol conversion step and improves the reliability and timeliness of instruction execution.

[0051] In some embodiments, the step of matching and comparing the quality event object according to the associated anomaly conditions in the interception strategy library based on the second-level interception instruction, and generating a third-level interception instruction based on the matching result and the second-level interception instruction, includes: Extract the associated anomaly conditions related to the quality event object from the interception policy library, wherein the associated anomaly conditions contain at least one association rule; Traverse each of the association rules in the associated anomaly conditions, and match the event identifier of the quality event object with the historical anomaly event identifiers in the association rules one by one; When a match is successful, the association rule is marked as a valid association rule, and the upgrade action identifier corresponding to the valid association rule is extracted; If the match fails, the second-level interception command will be output directly, and the generation of the third-level interception command will be stopped. If at least one of the valid association rules exists, the second-level interception command is updated with instruction enhancement based on the upgrade action identifier corresponding to all the valid association rules, to obtain the third-level interception command.

[0052] Specifically, all collected upgrade action identifiers are parsed and integrated. When multiple identifiers exist, such as simultaneously needing to freeze material batches and insert pre-checks, the logical relationships between these identifiers are handled to avoid conflicts. Based on each upgrade action identifier, a pre-defined upgrade action implementation module is invoked. Each module is responsible for transforming the abstract intent represented by the identifier into a specific extended instruction unit. For example, for inserting pre-checks, the module generates a formatted transaction request sent to the production scheduling system, which includes the affected material batches, the suggested inspection procedure code, and its location. Then, the core update operation is executed, packaging the extended unit as an independent instruction block within the same transaction package as the original instruction; it can also modify certain parameters of the original instruction, such as extending the equipment lock time until the associated batch of materials has been checked; it can also add new control metadata to the instruction package, instructing the actuator to process multiple instructions within this package in a specific order. After integration, the generated composite instruction package undergoes final consistency verification and formatting, and the output is a third-level interception instruction. The method provided in this implementation extracts relevant anomaly conditions from the interception strategy library and matches them with quality event objects. Based on preset rules, it can identify whether a current single anomaly event is related to historical data, thereby effectively identifying systemic risks caused by specific material batches, equipment components, or process parameters. This enhances the ability to prevent potential common problems, moving beyond simply handling isolated events. By directly outputting a second-level interception command and stopping the escalation process when matching fails, the method ensures that broader interventions are only implemented when there is clear evidence of correlation. This avoids unnecessary interference with the production process due to unfounded over-control, reflecting the principles of precise and prudent decision-making.

[0053] In some embodiments, obtaining the interception feedback information after executing the hierarchical interception instruction, and integrating the interception feedback information with the production work order information to obtain traceability record information includes: The interception feedback information is analyzed to obtain the instruction execution result, the interception execution timestamp, and the execution device identifier; If the result of the instruction execution indicates that the interception failed, then an interception failure status flag is generated; The interception failure status identifier is combined with the work order information to construct an interception failure record, thereby generating a warning record information; Specifically, the process retrieves the interception failure status identifier and work order information. Based on a predefined warning record data template, a new empty record structure is created. This template defines the fields the record should contain, including event summary, fault details, production context, and time metadata. The failure reason code, executing device identifier, failure timestamp, and associated instruction ID from the interception failure status identifier are extracted and populated into the fault details section of the warning record. Simultaneously, fields such as production work order number, product model, process code, and material batch from the work order information are populated into the production context section of the record. Furthermore, the builder generates a human-readable warning summary, for example, by concatenating key information using a template to state that the interception instruction [ID] in process [code] failed at [time] due to [reason]. A globally unique warning record ID and timestamp are also generated and populated. After all fields are populated, the record undergoes a logical consistency check to ensure no key fields are missing. Ultimately, the warning log information is immediately sent to two parallel channels: one is inserted into the quality event and interception traceability database for long-term storage and analysis; the other is published to a high-priority message bus or notification service to trigger actions in real time, such as pushing alarm messages to the production supervisor's mobile terminal, displaying a red alert on the central dashboard in the workshop, or automatically creating an emergency work order for the maintenance system.

[0054] If the instruction execution result indicates successful interception, then the interception execution timestamp is compared with the pre-stored standard process time to calculate the interception delay duration. The instruction execution result, the interception execution timestamp, the execution device identifier, the interception delay duration, and the production work order information are associated and bound to generate the traceability record information.

[0055] Specifically, all necessary input elements are collected: instruction execution result, interception execution timestamp, execution device identifier, interception delay duration (if successful) or warning record information (if failed), and production work order information. For both success and failure scenarios, a unified traceability record data model is adopted, but with internal fields that can be flexibly adapted. The work order number and quality event ID in the production work order information are used as the primary key and anchor point for the record. Next, it begins to populate data according to the structured format defined by the model: the instruction execution result, execution timestamp, device identifier, and delay duration (or warning record ID) are filled in as execution feedback layer data; all fields in the production work order information are filled in as production context layer data; origin layer data of this event is associated and introduced from a broader context, such as the core characteristics of the original quality event object that triggered this interception (e.g., measurement value, anomaly confidence, triggering rule). All this data is tightly bound together through a shared event ID. Metadata for the record is generated, such as creation time and record version number, and the record checksum is calculated to ensure integrity. Ultimately, this self-describing traceability record information object, containing end-to-end information, is serialized and persistently stored in an end-to-end quality traceability database or data lake optimized for analysis.

[0056] The method provided in this implementation, through instruction execution analysis of interception feedback information, can accurately parse the instruction execution results, precise timestamps, and specific device identifiers. This ensures that the final state, occurrence time, and executing entity of each interception action can be clearly captured and recorded, achieving monitorability and auditability of the interception execution process. By branching processing based on the instruction execution results, a failure flag containing detailed reasons is generated and a warning record is constructed when interception fails. This can immediately drive manual intervention and maintenance response, effectively preventing quality risks from spiraling out of control due to equipment failure or communication anomalies. When interception is successful, the interception delay duration is calculated, which can accurately quantify the system response efficiency from decision-making to execution, providing key indicators for evaluating and optimizing real-time control performance.

[0057] Reference Figure 2 As shown, the present invention also provides a real-time process quality interception system, applied to any of the above-described real-time process quality interception methods, comprising: The acquisition module is used to acquire quality inspection data output by sensing devices on the production line in real time, obtain production work order information associated with the quality inspection data, and integrate the quality inspection data and the production work order information to obtain a quality event object. The analysis module is used to input the quality event object into the multi-rule analysis channel for anomaly calculation and to obtain the anomaly confidence level. The association module is used to construct an interception based on the anomaly confidence level, the quality event object, and a preset interception strategy library, and obtain a hierarchical interception instruction. The processing module is used to obtain the interception feedback information after executing the hierarchical interception instruction, and integrate the interception feedback information with the production work order information to obtain traceability record information.

[0058] This invention provides a real-time process quality interception system. By collecting and integrating production line data and production context in real time to generate quality event objects, it can achieve a response from signal perception to anomaly determination, thus overcoming the interception delay problem caused by lagging comparison in traditional methods. By inputting quality event objects into parallel multi-rule analysis channels for fusion calculation, it can comprehensively utilize multi-dimensional information such as static specifications, dynamic processes, and correlation strategies, significantly improving the accuracy and reliability of anomaly judgment and solving the problems of false alarms and missed alarms caused by isolated data and weak basis. By dynamically constructing hierarchical interception instructions based on anomaly confidence and combined with a strategy library, it can directly drive field equipment to perform differentiated automatic interventions from early warning and locking to process correction, reducing reliance on human experience and achieving standardization and immediacy of interception actions. By integrating interception feedback and production information in real time to generate structured traceability records, it can completely record the decision-making and execution closed loop.

[0059] Reference Figure 3 As shown, the present invention also provides a real-time process quality interception device, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the steps of a real-time interception method for process quality as described in any of the preceding claims.

[0060] In this embodiment, the processor and memory can be connected via a bus or other means. The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, digital signal processor, application-specific integrated circuit, or one or more integrated circuits configured to implement embodiments of the present invention.

[0061] The present invention also provides a storage medium storing computer instructions for causing a computer to perform the method according to any one of the preceding claims.

[0062] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the system and each module described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0063] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for real-time interception of process quality, characterized in that, include: The system collects quality inspection data output from sensing devices on the production line in real time, obtains production work order information associated with the quality inspection data, and integrates the quality inspection data and the production work order information to obtain a quality event object. The quality event object is input into the multi-rule analysis channel for anomaly calculation to obtain the anomaly confidence level. Based on the anomaly confidence level, the quality event object, and the preset interception strategy library, an interception system is constructed to obtain a hierarchical interception instruction. Obtain the interception feedback information after executing the hierarchical interception command, and integrate the interception feedback information with the production work order information to obtain traceability record information.

2. The method for real-time interception of process quality according to claim 1, characterized in that, The process involves real-time acquisition of quality inspection data output from sensing devices on the production line, obtaining production work order information associated with the quality inspection data, and integrating the quality inspection data and the production work order information to obtain quality event objects, including: The quality inspection data is read to obtain the device identifier, measurement value, and acquisition timestamp; Using the device identifier as the query key, query the manufacturing execution system for the corresponding work order number and current process code; Based on the work order number, the manufacturing execution system queries the corresponding material batch information and operator information; The work order number, the current process code, the material batch information, and the operator information are integrated into the production work order information; The quality event object is obtained by combining the production work order information, the equipment identifier, the measurement value, and the collection timestamp.

3. The method for real-time interception of process quality according to claim 1, characterized in that, The step of inputting the quality event object into a multi-rule analysis channel for anomaly calculation to obtain anomaly confidence includes: Identify the real-time measurement value queue and process condition set from the quality event objects; The real-time measurement value queue is matched with each anomaly judgment rule of the multi-rule analysis channel. If the measurement value of the real-time measurement value queue falls into the corresponding anomaly judgment rule, the anomaly type corresponding to the anomaly judgment rule is marked as a violation instance. The number of all the aforementioned violation instances is summarized, and the violation ratio is calculated based on the real-time measurement value queue and the number of instances to obtain the initial violation ratio; The initial violation rate is corrected based on the set of process conditions to obtain the corrected violation rate; The corrected violation ratio is mapped to a preset confidence quantification range to obtain the anomaly confidence level.

4. The method for real-time interception of process quality according to claim 1, characterized in that, The step of constructing a tiered interception instruction based on the anomaly confidence level, the quality event object, and a preset interception strategy library includes: The anomaly confidence level is compared with a preset first confidence threshold and a second confidence threshold, wherein the first confidence threshold is less than the second confidence threshold. If the anomaly confidence level is less than the first confidence threshold, it is determined that there is no anomaly and no interception command is generated. When the anomaly confidence level is higher than the first confidence threshold and lower than the second confidence threshold, a first-level interception instruction is constructed based on the quality event object and the interception strategy library. When the anomaly confidence level is higher than the second confidence threshold, a second-level interception instruction is constructed based on the quality event object and the interception strategy library. Based on the second-level interception instruction, the quality event object is matched and compared according to the associated abnormal conditions in the interception strategy library, and a third-level interception instruction is generated based on the matching result and the second-level interception instruction; wherein, the hierarchical interception instruction includes the first-level interception instruction, the second-level interception instruction and the third-level interception instruction.

5. The real-time interception method for process quality according to claim 4, characterized in that, When the anomaly confidence level is higher than the second confidence threshold, a second-level interception instruction is constructed based on the quality event object and the interception strategy library, including: Extract event feature parameters from the quality event object, and extract the secondary instruction generation template corresponding to the event feature parameters from the interception strategy library; Based on the parameter mapping relationship in the secondary instruction generation template, the event feature parameters are assigned to the instruction parameter items in the secondary instruction generation template to generate a basic instruction parameter set; The instruction parameter set is optimized for interception based on the secondary interception measures in the secondary instruction generation template to obtain the interception instruction parameter set. The interception instruction parameter set is filled into the instruction structure of the secondary instruction generation template to generate a secondary basic instruction unit; Based on the quality event object, all the secondary basic instruction units are encapsulated and combined to obtain the second-level interception instruction.

6. The real-time interception method for process quality according to claim 4, characterized in that, The process involves matching and comparing the quality event object based on the associated anomaly conditions in the interception strategy library, and generating a third-level interception instruction based on the matching result and the second-level interception instruction, including: Extract the associated anomaly conditions related to the quality event object from the interception policy library, wherein the associated anomaly conditions contain at least one association rule; Traverse each of the association rules in the associated anomaly conditions, and match the event identifier of the quality event object with the historical anomaly event identifiers in the association rules one by one; When a match is successful, the association rule is marked as a valid association rule, and the upgrade action identifier corresponding to the valid association rule is extracted; If the match fails, the second-level interception command will be output directly, and the generation of the third-level interception command will be stopped. If at least one of the valid association rules exists, the second-level interception command is updated with instruction enhancement based on the upgrade action identifier corresponding to all the valid association rules, to obtain the third-level interception command.

7. The method for real-time interception of process quality according to claim 1, characterized in that, The step of obtaining interception feedback information after executing the hierarchical interception command, and integrating the interception feedback information with the production work order information to obtain traceability record information includes: The interception feedback information is analyzed to obtain the instruction execution result, the interception execution timestamp, and the execution device identifier; If the result of the instruction execution indicates that the interception failed, then an interception failure status flag is generated; The interception failure status identifier is combined with the work order information to construct an interception failure record, thereby generating a warning record information; If the instruction execution result indicates successful interception, then the interception execution timestamp is compared with the pre-stored standard process time to calculate the interception delay duration. The instruction execution result, the interception execution timestamp, the execution device identifier, the interception delay duration, and the production work order information are associated and bound to generate the traceability record information.

8. A real-time process quality interception system, characterized in that, The method for real-time interception of process quality according to any one of claims 1-7 includes: The acquisition module is used to acquire quality inspection data output by sensing devices on the production line in real time, obtain production work order information associated with the quality inspection data, and integrate the quality inspection data and the production work order information to obtain a quality event object. The analysis module is used to input the quality event object into the multi-rule analysis channel for anomaly calculation and to obtain the anomaly confidence level. The association module is used to construct an interception based on the anomaly confidence level, the quality event object, and a preset interception strategy library, and obtain a hierarchical interception instruction. The processing module is used to obtain the interception feedback information after executing the hierarchical interception instruction, and integrate the interception feedback information with the production work order information to obtain traceability record information.

9. A real-time process quality interception device, characterized in that, include: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the real-time interception method for process quality as described in any one of claims 1-7.

10. A storage medium, characterized in that, The computer contains computer instructions for causing the computer to perform the method according to any one of claims 1 to 7.

Images