An aluminum profile production whole-process quality traceability management system

Abstract

This invention discloses a quality traceability management system for the entire aluminum profile production process, specifically relating to the field of fault prediction and health management in aluminum profile production. It includes a data construction module, a feature fusion module, a risk identification module, an anomaly backtracking module, and a closed-loop optimization module. The data construction module collects process parameters and equipment parameters at each stage of aluminum profile production, aligns them with the material flow cycle time, and extracts key indicators reflecting the trend of state changes. These indicators are then fed into the equipment analysis path and the process analysis path, forming a first dataset and a second dataset. This invention links the dynamic changes in equipment health status with batch quality judgment in real time and combines risk assessment and traceability mechanisms at the cyber-physical system level to identify chronic deviations, limit their impact scope, and dynamically update batch records. This achieves a deep integration of the entire aluminum profile production process quality traceability system with fault prediction and health management.

Description

Technical Field

[0001] This invention relates to the field of fault prediction and health management technology in aluminum profile production, and more specifically, to a quality traceability management system for the entire aluminum profile production process. Background Technology

[0002] In aluminum profile production, existing equipment health management mainly focuses on real-time status. Although it can identify abnormal temperatures, vibrations, or energy consumption fluctuations, it is difficult to capture chronic problems that will only show their effects after a long period of time. Secondly, the quality traceability system only records in chronological order. When the health management system discovers that the equipment has a chronic deviation over a period of time, this information cannot be dynamically reflected in the quality judgment of historical batches, nor can it accurately mark the batches that may be affected. Thirdly, when new evidence comes from feedback from downstream users, this delayed information cannot be linked to the health history of the equipment. The traceability chain cannot be traced back to the specific equipment status and process conditions. Ultimately, when faced with delayed quality hazards, companies either have to follow the initial judgment, resulting in distorted risk assessments and recall scope, or they are forced to modify historical data, thereby compromising the integrity of the system.

[0003] Therefore, the existing quality traceability system for the entire aluminum profile production process lacks the ability to deeply integrate with fault prediction and health management. At the cyber-physical system level, it is impossible to dynamically link changes in equipment health status with batch quality judgment without destroying the original records, and to accurately transmit the scope of impact, thus causing management decisions to remain on outdated information for a long time. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a quality traceability management system for the entire aluminum profile production process. By linking the dynamic changes in equipment health status with batch quality judgment in real time, and combining risk assessment and traceability mechanisms at the cyber-physical system level, the system can identify chronic deviations, limit their scope of influence, and dynamically update batch records. This addresses the problems of existing aluminum profile production process quality traceability systems being unable to deeply integrate with fault prediction and health management, and being unable to dynamically reflect chronic deviations and their scope of influence without destroying original records.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a quality traceability management system for the entire production process of aluminum profiles, comprising a data construction module, a feature fusion module, a risk identification module, an anomaly backtracking module, and a closed-loop optimization module;

[0006] The data construction module collects process parameters and equipment parameters in each stage of aluminum profile production, aligns them with the material flow cycle time, and extracts key indicators that reflect the trend of state changes. These are then sent to the equipment analysis path and the process analysis path to form the first dataset and the second dataset.

[0007] The feature fusion module generates a periodic health curve and a trend decline curve by performing segment health calculations on the first dataset in the equipment analysis path, extracting batch-to-batch deviation patterns by performing consistency calculations on the second dataset in the process analysis path, and generating a third dataset containing health and quality indicators when there is a feature correlation between the results of the two analysis paths.

[0008] The risk identification module binds a third dataset to the batch traceability index, monitors indicator changes during the batch lifecycle, performs risk assessment when preset conditions are met, and generates a set of risk batches by combining spatial distribution differences and decline trend predictions.

[0009] When the anomaly backtracking module receives a downstream quality anomaly signal, it matches the signal with the risk batch set, and uses health curve backtracking and deviation mode backtracking to locate the equipment and process parameters at the time of the anomaly and generate an anomaly snapshot.

[0010] The closed-loop optimization module revises the quality judgment of affected batches based on abnormal snapshots, generates traceability branches and synchronizes them to production scheduling, inventory management and customer management links, and writes the revision results back to the equipment analysis path and process analysis path.

[0011] In a preferred embodiment, in the data construction module, process parameters including raw material composition, process temperature, process pressure and surface treatment parameters, as well as equipment parameters including equipment temperature, vibration, energy consumption and position parameters are obtained in each stage of aluminum profile production. Then, noise removal and outlier screening are performed on the process parameters and equipment parameters respectively to construct a process parameter sequence and equipment parameter sequence with stable numerical characteristics.

[0012] The process parameter sequence and equipment parameter sequence are aligned by time index using the material flow cycle time at each production node, and a synchronization parameter matrix is ​​formed by combining the production node position mapping relationship.

[0013] The trend feature extraction process is performed using the synchronization parameter matrix. Specifically, the trend feature extraction process includes statistically analyzing the difference between adjacent sampling points to obtain the local rate of change, calculating the correlation coefficient between process parameters and equipment parameters to identify linkage patterns, determining the boundary conditions for the state to transform from stable to abnormal based on the linkage pattern identification method, and generating key indicators that can reflect the trend of state change.

[0014] Key indicators are input into the equipment analysis path and process analysis path according to their source paths, and the correspondence between parameter indexes and source paths is kept consistent during the input process to form the first dataset and the second dataset for subsequent feature fusion and risk identification.

[0015] In a preferred embodiment, in the feature fusion module, a sliding segment window is constructed in the equipment analysis path for the first dataset according to the production segment and material passage time to obtain segment data blocks, and then segment health calculation is performed;

[0016] Section health calculation involves statistically analyzing and aggregating the stability, volatility, and drift of equipment parameters to determine the health value corresponding to each window, and finally outputting a sequence of health indicators arranged in production order.

[0017] By performing periodic analysis and trend decomposition on the health indicator sequence to obtain recurring patterns and continuous changing trends, curve construction is then performed. Curve construction maps the recurring patterns to periodic health curves and decomposes the continuous changing trends into trend decay curves. Finally, a set of health indicators containing both periodic health curves and trend decay curves is output.

[0018] In a preferred embodiment, in the feature fusion module, a control batch group is obtained by processing the second dataset according to the correspondence between production segment and batch in the process analysis path, and then a consistency calculation is performed.

[0019] Consistency calculation statistically analyzes the differences, stability, and continuity of process parameters and identifies deviation boundaries to solve for batch-to-batch deviation patterns. It also transforms the consistency calculation results into quality indicators that reflect the stability of batch processes and finally outputs a set of process characteristics that includes deviation patterns and quality indicators.

[0020] By obtaining corresponding records located in the same production segment and belonging to the same material flow chain from the health indicator set and the process feature set, feature association fusion is then performed. Feature association fusion is filtered by the joint conditions of correlation strength and consistency constraints. Records that meet the preset association conditions are merged to form fused records. Finally, a third dataset containing health indicators and quality indicators is output.

[0021] In a preferred embodiment, in the risk identification module, a batch monitoring index table is constructed by establishing a one-to-one mapping relationship between health indicators and quality indicators in the third dataset according to the batch traceability index. The batch traceability index is used as the primary key, and health indicators and quality indicators are used as fields. When the table is constructed, the spatial location parameters of each batch at the production node are recorded. At the same time, the average value of the health indicators and quality indicators of the batch during the initial production cycle is used as the initial baseline value of the batch and stored in the batch monitoring index table.

[0022] During the batch lifecycle, the latest health and quality indicators from the feature fusion process are received at set time intervals, updated to the batch monitoring index table, and the current offset is calculated based on the initial baseline value. The offset changes within multiple time intervals are continuously recorded to form a dynamic offset sequence, and the offset change rate within each time interval is calculated.

[0023] In a preferred embodiment, in the risk identification module, when the offset and rate of change of health indicators and quality indicators in the dynamic offset sequence both reach the preset trigger conditions, a risk assessment process is executed; during the risk assessment process, the risk coefficient is calculated based on the rate of change of the offset, and the risk diffusion range is derived by combining the spatial location parameters in the batch monitoring index table with the spatial distribution differences of production nodes; when the above conditions are not met, the batch is marked as low-risk and dynamic monitoring updates continue to be executed.

[0024] The risk coefficient and the risk spread range are input into the decline trend prediction model built based on the historical health indicator change sequence, and the future performance decline trend is calculated to obtain the risk level information for each risk batch.

[0025] Based on the risk level information, a risk batch set is selected, and the risk batch set and its risk level information are output to the anomaly backtracking process and the closed-loop optimization process. At the same time, the records of the risk batch set are updated back to the batch monitoring index table.

[0026] In a preferred embodiment, in the anomaly backtracking module, after receiving a quality anomaly signal from the downstream quality inspection stage, the batch traceability index, anomaly occurrence time, and associated quality indicator change characteristics in the quality anomaly signal are analyzed, and a set of target batches that meet the matching requirements are selected based on the joint judgment condition of time overlap and quality indicator feature similarity.

[0027] For the target batch set, historical health indicator data corresponding to the batch traceability index are retrieved from the batch monitoring index table. Taking the time of the abnormality as the starting point for backtracking, the health indicators and corresponding offsets and offset change rates of each time interval are extracted in the order of monitoring time, and a health indicator backtracking sequence is generated.

[0028] For the target batch set, historical quality index data corresponding to the batch traceability index is retrieved from the batch monitoring index table. Taking the production segment where the anomaly occurred as the starting point for backtracking, the quality index and corresponding offset and offset change rate of each time interval are extracted in the order of production segments, and a deviation pattern backtracking sequence is generated.

[0029] The health indicator backtracking sequence and the deviation pattern backtracking sequence are aligned according to the dual correspondence between timestamp and production segment. The intersection segment that simultaneously meets the health indicator abnormality and quality indicator abnormality at the time of the abnormality is identified. The equipment parameter and process parameter status of the intersection segment are analyzed to generate an abnormal snapshot containing the abnormality triggering conditions, abnormal segment parameter characteristics and potential cause paths.

[0030] In a preferred embodiment, the closed-loop optimization module receives an anomaly snapshot output from the anomaly backtracking module. Based on the anomaly triggering conditions, anomaly segment parameter characteristics, and potential cause paths in the anomaly snapshot, the quality judgment result of the target batch corresponding to the batch monitoring index table is revised, and the revised quality judgment result and the original judgment result are combined to form a revision record.

[0031] Based on the revised quality judgment results, the risk coefficient, risk spread range and risk level information of the corresponding batch in the batch monitoring index table are called to generate a traceability branch containing the revision record, risk level and impact range, and the traceability branch is synchronized to the production scheduling, inventory management and customer management links;

[0032] The revised quality judgment results and corresponding traceability branches are written back to the equipment analysis path and process analysis path associated with the batch in the batch monitoring index table, and the equipment parameters and process parameters in the path are updated.

[0033] After the batch monitoring index table is updated, the dynamic correction mechanism of the risk assessment model and the decline trend prediction model is invoked. The parameter data of the revised equipment analysis path and process analysis path are input into the model, the parameter weight adjustment and prediction curve retraining are performed, and the optimized prediction results are output.

[0034] The technical effects and advantages of this invention are as follows:

[0035] 1. This solution establishes a dynamic mapping between long-term monitoring data of equipment health management and batch quality judgment results, enabling the real-time mapping of chronic deviation information to relevant batches at the cyber-physical system level without modifying the original records. This allows for the marking of affected batches and updating of risk assessment results after potential quality hazards are discovered.

[0036] 2. This system supports linking delayed quality information from downstream user feedback with historical equipment health data at the cyber-physical system level to complete the original traceability chain and achieve the ability to trace back from end-user feedback to specific equipment status and process conditions, thus avoiding information fragmentation.

[0037] 3. This solution introduces a long-term monitoring and trend analysis mechanism to capture and quantify chronic fault symptoms that only show their impact after multiple production cycles, providing a data foundation for early intervention and traceability.

[0038] 4. This solution also calculates the correlation between health indicators and quality results in real time. When the equipment status deviates from the normal range, the system updates the quality status of the affected batches, realizing dynamic correction of the judgment results without the need for manual modification of historical data.

[0039] 5. This system limits the specific set of risky batches by combining changes in health status with batch production records, ensuring that the scope of risky batch determination strictly corresponds to the actual affected production records, thereby maintaining a match between the scope of determination in recall and risk control. Attached Figure Description

[0040] Figure 1 This is a system module diagram of the present invention.

[0041] Figure 2 This is a flowchart of the data acquisition and preprocessing process of the present invention.

[0042] Figure 3 This is a flowchart of the health status analysis and chronic deviation identification process of the present invention.

[0043] Figure 4 This is a flowchart illustrating the risk batch limitation and traceability process of the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Refer to the instruction manual appendix Figure 1-4 An embodiment of the present invention provides a quality traceability management system for the entire production process of aluminum profiles, comprising a data construction module, a feature fusion module, a risk identification module, an anomaly backtracking module, and a closed-loop optimization module.

[0046] The data construction module collects process parameters and equipment parameters at each stage of aluminum profile production. Process parameters include raw material composition, process temperature, process pressure and surface treatment, while equipment parameters include equipment temperature, vibration, energy consumption and location. Based on the material flow cycle, time alignment is performed and key indicators reflecting the trend of state changes are extracted. These are then sent to the equipment analysis path and process analysis path respectively to form the first dataset and the second dataset.

[0047] The feature fusion module generates a periodic health curve and a trend decline curve by performing segment health calculations on the first dataset in the equipment analysis path, extracting batch-to-batch deviation patterns by performing consistency calculations on the second dataset in the process analysis path, and generating a third dataset containing health and quality indicators when there is a feature correlation between the results of the two analysis paths.

[0048] The risk identification module binds a third dataset to the batch traceability index, monitors indicator changes during the batch lifecycle, performs risk assessment when preset conditions are met, and generates a set of risk batches by combining spatial distribution differences and decline trend predictions.

[0049] When the anomaly backtracking module receives a downstream quality anomaly signal, it matches the signal with the risk batch set, and uses health curve backtracking and deviation mode backtracking to locate the equipment and process parameters at the time of the anomaly and generate an anomaly snapshot.

[0050] The closed-loop optimization module revises the quality judgment of affected batches based on abnormal snapshots, generates traceability branches containing revision records, risk levels and impact scope, and synchronizes them to production scheduling, inventory management and customer management links. At the same time, the revision results are written back to the equipment analysis path and process analysis path.

[0051] In the data construction module, process parameters including raw material composition, process temperature, process pressure and surface treatment parameters are obtained in each stage of aluminum profile production, as well as equipment parameters including equipment temperature, vibration, energy consumption and position parameters. Then, noise removal and outlier screening are performed on the process parameters and equipment parameters respectively to construct a process parameter sequence and equipment parameter sequence with stable numerical characteristics.

[0052] The process parameter sequence and equipment parameter sequence are aligned by time indexing using the material flow cycle time at each production node, and a synchronization parameter matrix is ​​formed by combining the production node position mapping relationship, thereby ensuring that the process parameters and equipment parameters have a one-to-one data association at the same production node.

[0053] The trend feature extraction process is performed using the synchronization parameter matrix. Specifically, the trend feature extraction process includes statistically analyzing the difference between adjacent sampling points to obtain the local rate of change, calculating the correlation coefficient between process parameters and equipment parameters to identify linkage patterns, determining the boundary conditions for the state to transform from stable to abnormal based on the linkage pattern identification method, and generating key indicators that can reflect the trend of state change.

[0054] Key indicators are input into the equipment analysis path and process analysis path according to their source paths, and the correspondence between parameter indexes and source paths is kept consistent during the input process to form the first dataset and the second dataset for subsequent feature fusion and risk identification.

[0055] In the feature fusion module, the segment data blocks are obtained by constructing a sliding segment window in the equipment analysis path for the first dataset according to the production segment and material passage time, and then the segment health calculation is performed.

[0056] Section health calculation involves statistically analyzing and aggregating the stability, volatility, and drift of equipment parameters to determine the health value corresponding to each window, and finally outputting a sequence of health indicators arranged in production order.

[0057] By performing periodic analysis and trend decomposition on the health indicator sequence to obtain recurring patterns and continuous changing trends, curve construction is then performed. Curve construction maps the recurring patterns to periodic health curves and decomposes the continuous changing trends into trend decay curves. Finally, a set of health indicators containing both periodic health curves and trend decay curves is output.

[0058] In the feature fusion module, a control batch group is obtained by analyzing the second dataset in the process analysis path according to the correspondence between production sections and batches, and then a consistency calculation is performed.

[0059] Consistency calculation statistically analyzes the differences, stability, and continuity of process parameters and identifies deviation boundaries to solve for batch-to-batch deviation patterns. It also transforms the consistency calculation results into quality indicators that reflect the stability of batch processes and finally outputs a set of process characteristics that includes deviation patterns and quality indicators.

[0060] By obtaining corresponding records located in the same production segment and belonging to the same material flow chain from the health indicator set and the process feature set, feature association fusion is then performed. Feature association fusion is filtered by the joint conditions of correlation strength and consistency constraints. Records that meet the preset association conditions are merged to form fused records. Finally, a third dataset containing health indicators and quality indicators is output.

[0061] In the risk identification module, a one-to-one mapping relationship is established between health indicators and quality indicators in the third dataset according to the batch traceability index. A batch monitoring index table is constructed with the batch traceability index as the primary key and health indicators and quality indicators as fields. When the table is built, the spatial location parameters of each batch at the production node are recorded. At the same time, the average value of the health indicators and quality indicators of the batch in the initial production cycle is used as the initial baseline value of the batch and stored in the batch monitoring index table.

[0062] During the batch lifecycle, the latest health and quality indicators from the feature fusion process are received at set time intervals, updated to the batch monitoring index table, and the current offset is calculated based on the initial baseline value. The offset changes within multiple time intervals are continuously recorded to form a dynamic offset sequence, and the offset change rate within each time interval is calculated. Specifically, the offset of the health indicator is obtained by subtracting the initial baseline value of the current batch from the health indicator value of the current batch, and the offset of the quality indicator is obtained by subtracting the initial baseline value of the current batch from the quality indicator value of the current batch. Furthermore, the offset difference between two adjacent sampling times is divided by the interval length between the two time points to obtain the offset change rate.

[0063] In the risk identification module, when the offset and rate of change of health indicators and quality indicators in the dynamic offset sequence both reach the preset trigger conditions, a risk assessment process is executed. During the risk assessment process, the risk coefficient is calculated based on the rate of change of the offset, and the risk diffusion range is deduced by combining the spatial location parameters in the batch monitoring index table with the spatial distribution differences of production nodes. When the above conditions are not met, the batch is marked as low-risk and dynamic monitoring updates continue. The process of executing dynamic monitoring involves the system continuing to receive and update health indicators and quality indicators at set time intervals to form a rolling monitoring record.

[0064] The risk coefficient and the risk spread range are input into the decline trend prediction model built based on the historical health indicator change sequence, and the future performance decline trend is calculated to obtain the risk level information for each risk batch.

[0065] Based on the risk level information, a risk batch set is selected, and the risk batch set and its risk level information are output to the anomaly backtracking process and the closed-loop optimization process. At the same time, the records of the risk batch set are updated back to the batch monitoring index table for dynamic correction and optimization of the risk assessment model and the decline trend prediction model.

[0066] It should be noted that in the formula structure involved in this scheme, dimensionless terms can be used as proportional or structural adjustment factors. When combined with quantities with units, they only play a role in numerical scaling and do not introduce new physical dimensions. Therefore, they will not change or confuse the overall unit system. This combination of "dimensionless terms and terms with units" can be understood as a composite structural expression commonly used in mathematical physics modeling. It conforms to the principle of dimensional consistency and has a clear physical interpretation basis.

[0067] Secondly, in the formula structure of this scheme, if multiple variables with different physical units are involved, including but not limited to time, mass or energy variables, their joint appearance is to express the collaborative modeling relationship of multiple physical mechanisms. Each variable can form a unified structure through function mapping, ratio combination or normalization adjustment, with clear units and clear meaning. The overall expression conforms to the principle of dimensional consistency and the conventional formula of engineering modeling.

[0068] In this solution, constants, weights, adjustment factors, threshold parameters, proportional coefficients, etc., are all adjustable control parameters for different application environments. Their values ​​depend on the target equipment configuration, data input characteristics, and performance optimization goals. During the implementation phase, they are set to converge within a reasonable range through model verification, performance constraints, or engineering calibration. Although these parameters do not have a unique preset value, they have clear adjustment logic and calculation paths. They belong to the deterministic setting process in engineering implementation. The purpose of this setting is to ensure that the solution is both universally adaptable and reproducible and operable, without affecting its technical clarity and feasibility.

[0069] definition For health indicators in the first To the The rate of change within each monitoring interval For quality indicators in the first To the Rate of change within each monitoring interval:

[0070]

[0071]

[0072]

[0073]

[0074]

[0075] in In the first Health indicator values ​​at each monitoring time point; In the first Health indicator values ​​at each monitoring time point; In the first The quality indicator values ​​at each monitoring time point; In the first The quality indicator values ​​at each monitoring time point; For the first Timestamps for each monitoring time point; For the first Timestamps for each monitoring time point; For health indicators in the first To the Rate of change within each monitoring interval; For quality indicators in the first To the Rate of change within each monitoring interval; This refers to the batch currently being evaluated; In order to be consistent with batch Another batch for comparison; For batch Spatial diffusion factor; The number of batches participating in spatial computation; For another batch The weighting parameters (obtained by weighting factors such as batch output and production priority); For batch With another batch Spatial distance between them;

[0076] For batch The overall risk coefficient; For batch The rate of change of health indicators; For batch The rate of change of quality indicators; This is a risk mapping function for the rate of change of health indicators; This is a risk mapping function for the rate of change of quality indicators; For batch In the future Internal performance degradation trend value; It is a multivariate trend prediction function; For batch The historical health indicator decline curve fitting function; For batch In time Historical values ​​of health indicators The time sampling interval;

[0077] Furthermore, through health indicators With quality indicators rate of change of offset , Quantitative calculations were performed, and the spatial diffusion factor was incorporated. Derive the batch comprehensive risk coefficient Then Compared with historical health indicators decline curve Input to recession trend prediction function This will lead to a future performance degradation trend. Its purpose is to establish a closed-loop calculation link from dynamic monitoring trigger condition determination, risk coefficient construction, risk diffusion range derivation to decline trend prediction, so that the risk assessment results are quantifiable, traceable and optimizable, thereby providing a basis for risk decision-making for production scheduling, inventory management and customer management.

[0078] In the anomaly backtracking module, after receiving a quality anomaly signal from the downstream quality inspection link, the batch traceability index, anomaly occurrence time, and associated quality indicator change characteristics in the quality anomaly signal are analyzed. Based on the joint judgment condition of time overlap and quality indicator feature similarity, a set of target batches that meet the matching requirements is selected.

[0079] For the target batch set, historical health indicator data corresponding to the batch traceability index are retrieved from the batch monitoring index table. Taking the time of the abnormality as the starting point for backtracking, the health indicators and corresponding offsets and offset change rates of each time interval are extracted in the order of monitoring time, and a health indicator backtracking sequence is generated.

[0080] For the target batch set, historical quality index data corresponding to the batch traceability index is retrieved from the batch monitoring index table. Taking the production segment where the anomaly occurred as the starting point for backtracking, the quality index and corresponding offset and offset change rate of each time interval are extracted in the order of production segments, and a deviation pattern backtracking sequence is generated.

[0081] The backtracking sequences of health indicators and deviation patterns are aligned according to the dual correspondence between timestamps and production segments. The alignment process involves first aligning by timestamps and then performing a second alignment by production segment codes. This identifies the intersection segments that simultaneously satisfy both health indicator and quality indicator anomalies at the time of anomaly occurrence. The equipment and process parameter states of the intersection segments are analyzed to generate an anomaly snapshot containing anomaly triggering conditions, anomaly segment parameter characteristics, and potential causal paths. The anomaly snapshot is then output to the closed-loop optimization process for dynamic optimization of quality judgment revision and risk prediction models.

[0082] In the closed-loop optimization module, an anomaly snapshot output from the anomaly backtracking module is received. Based on the anomaly triggering conditions, anomaly segment parameter characteristics, and potential cause paths in the anomaly snapshot, the quality judgment result of the target batch corresponding to the batch monitoring index table is revised, and the revised quality judgment result and the original judgment result are combined to form a revision record.

[0083] Based on the revised quality judgment results, the risk coefficient, risk spread range and risk level information of the corresponding batch in the batch monitoring index table are called to generate a traceability branch containing the revision record, risk level and impact range, and the traceability branch is synchronized to the production scheduling, inventory management and customer management links;

[0084] The revised quality judgment results and corresponding traceability branches are written back to the equipment analysis path and process analysis path associated with the batch in the batch monitoring index table, and the equipment parameters and process parameters in the path are updated for dynamic optimization and anomaly prevention in the subsequent production process.

[0085] After the batch monitoring index table is updated, the dynamic correction mechanism of the risk assessment model and the decline trend prediction model is invoked. The parameter data of the revised equipment analysis path and process analysis path are input into the model, the parameter weight adjustment and prediction curve retraining are performed, and the optimized prediction results are output to guide the production control and risk management strategies of subsequent batches.

[0086] In practice, the risk identification mechanism is first established by collecting the offsets and rates of change of batch health and quality indicators in real time. This avoids misjudgments caused by a single parameter. When the triggering conditions are met, a risk assessment process is executed to calculate the risk coefficient. By combining the spatial location parameters in the batch monitoring index table with the spatial distribution differences of production nodes, the risk diffusion range is deduced, enabling spatial propagation prediction of potential faults. When the conditions are not met, the batch is marked as low-risk and continuously monitored dynamically, thus ensuring the real-time nature of health management.

[0087] Then, based on this, the risk coefficient and the diffusion range are input into the decline trend prediction model constructed based on the historical health indicator change sequence to predict the future performance decline and classify the risk level for each risk batch; then the risk batch set and its level information are output to the anomaly backtracking and closed-loop optimization process, and updated back to the batch monitoring index table to realize the dynamic correction of model parameters and judgment criteria.

[0088] This solution embeds fault prediction and health management directly into the quality traceability system, forming a closed-loop linkage mechanism oriented towards cyber-physical systems. This allows the system to link this information in real time to the historical batch quality records of already produced batches when it detects chronic equipment deviations, receives feedback from downstream, or identifies potential hazards, and marks potentially affected batches. This enables enterprises to directly use the latest equipment status and process conditions to assess risks and determine the scope of recalls when facing delayed quality issues, without having to modify existing data or rely on outdated information. Thus, while maintaining the integrity of the traceability chain, it improves the timeliness of quality management.

[0089] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A quality traceability management system for the entire production process of aluminum profiles, characterized in that, It includes a data construction module, a feature fusion module, a risk identification module, an anomaly backtracking module, and a closed-loop optimization module; The data construction module collects process parameters and equipment parameters in each stage of aluminum profile production, aligns them with the material flow cycle time, and extracts key indicators that reflect the trend of state changes. These are then sent to the equipment analysis path and the process analysis path to form the first dataset and the second dataset. The feature fusion module generates a periodic health curve and a trend decline curve by performing segment health calculations on the first dataset in the equipment analysis path, extracting batch-to-batch deviation patterns by performing consistency calculations on the second dataset in the process analysis path, and generating a third dataset containing health and quality indicators when there is a feature correlation between the results of the two analysis paths. The risk identification module binds a third dataset to the batch traceability index, monitors indicator changes during the batch lifecycle, performs risk assessment when preset conditions are met, and generates a set of risk batches by combining spatial distribution differences and decline trend predictions. When the anomaly backtracking module receives a downstream quality anomaly signal, it matches the signal with the risk batch set, and uses health curve backtracking and deviation mode backtracking to locate the equipment and process parameters at the time of the anomaly and generate an anomaly snapshot. The closed-loop optimization module revises the quality judgment of affected batches based on abnormal snapshots, generates traceability branches and synchronizes them to production scheduling, inventory management and customer management links, and writes the revision results back to the equipment analysis path and process analysis path.

2. The quality traceability management system for the entire aluminum profile production process according to claim 1, characterized in that: In the data construction module, process parameters including raw material composition, process temperature, process pressure and surface treatment parameters are obtained in each stage of aluminum profile production, as well as equipment parameters including equipment temperature, vibration, energy consumption and position parameters. Then, noise removal and outlier screening are performed on the process parameters and equipment parameters respectively to construct a process parameter sequence and equipment parameter sequence with stable numerical characteristics. The process parameter sequence and equipment parameter sequence are aligned by time index using the material flow cycle time at each production node, and a synchronization parameter matrix is ​​formed by combining the production node position mapping relationship. The trend feature extraction process is performed using the synchronization parameter matrix. Specifically, the trend feature extraction process includes statistically analyzing the difference between adjacent sampling points to obtain the local rate of change, calculating the correlation coefficient between process parameters and equipment parameters to identify linkage patterns, determining the boundary conditions for the state to transform from stable to abnormal based on the linkage pattern identification method, and generating key indicators that can reflect the trend of state change. Key indicators are input into the equipment analysis path and process analysis path according to their source paths, and the correspondence between parameter indexes and source paths is kept consistent during the input process to form the first dataset and the second dataset for subsequent feature fusion and risk identification.

3. The aluminum profile production process quality traceability management system according to claim 2, characterized in that: In the feature fusion module, the segment data blocks are obtained by constructing a sliding segment window in the equipment analysis path for the first dataset according to the production segment and material passage time, and then the segment health calculation is performed. Section health calculation involves statistically analyzing and aggregating the stability, volatility, and drift of equipment parameters to determine the health value corresponding to each window, and finally outputting a sequence of health indicators arranged in production order. By performing periodic analysis and trend decomposition on the health indicator sequence to obtain recurring patterns and continuous changing trends, curve construction is then performed. Curve construction maps the recurring patterns to periodic health curves and decomposes the continuous changing trends into trend decay curves. Finally, a set of health indicators containing both periodic health curves and trend decay curves is output.

4. The quality traceability management system for the entire aluminum profile production process according to claim 3, characterized in that: In the feature fusion module, a control batch group is obtained by analyzing the second dataset in the process analysis path according to the correspondence between production sections and batches, and then a consistency calculation is performed. Consistency calculation statistically analyzes the differences, stability, and continuity of process parameters and identifies deviation boundaries to solve for batch-to-batch deviation patterns. It also transforms the consistency calculation results into quality indicators that reflect the stability of batch processes and finally outputs a set of process characteristics that includes deviation patterns and quality indicators. By obtaining corresponding records located in the same production segment and belonging to the same material flow chain from the health indicator set and the process feature set, feature association fusion is then performed. Feature association fusion is filtered by the joint conditions of correlation strength and consistency constraints. Records that meet the preset association conditions are merged to form fused records. Finally, a third dataset containing health indicators and quality indicators is output.

5. The aluminum profile production process quality traceability management system according to claim 4, characterized in that: In the risk identification module, a one-to-one mapping relationship is established between health indicators and quality indicators in the third dataset according to the batch traceability index. A batch monitoring index table is constructed with the batch traceability index as the primary key and health indicators and quality indicators as fields. When the table is built, the spatial location parameters of each batch at the production node are recorded. At the same time, the average value of the health indicators and quality indicators of the batch in the initial production cycle is used as the initial baseline value of the batch and stored in the batch monitoring index table. During the batch lifecycle, the latest health and quality indicators from the feature fusion process are received at set time intervals, updated to the batch monitoring index table, and the current offset is calculated based on the initial baseline value. The offset changes within multiple time intervals are continuously recorded to form a dynamic offset sequence, and the offset change rate within each time interval is calculated.

6. The quality traceability management system for the entire aluminum profile production process according to claim 5, characterized in that: In the risk identification module, when the offset and rate of change of health indicators and quality indicators in the dynamic offset sequence both reach the preset trigger conditions, the risk assessment process is executed. During the risk assessment process, the risk coefficient is calculated based on the rate of change of offset, and the risk diffusion range is derived by combining the spatial location parameters in the batch monitoring index table with the spatial distribution differences of production nodes. If the above conditions are not met, the batch will be marked as low-risk and dynamic monitoring and updates will continue. The risk coefficient and the risk spread range are input into the decline trend prediction model built based on the historical health indicator change sequence, and the future performance decline trend is calculated to obtain the risk level information for each risk batch. Based on the risk level information, a risk batch set is selected, and the risk batch set and its risk level information are output to the anomaly backtracking process and the closed-loop optimization process. At the same time, the records of the risk batch set are updated back to the batch monitoring index table.

7. The quality traceability management system for the entire aluminum profile production process according to claim 1, characterized in that: In the anomaly backtracking module, after receiving a quality anomaly signal from the downstream quality inspection link, the batch traceability index, anomaly occurrence time, and associated quality indicator change characteristics in the quality anomaly signal are analyzed. Based on the joint judgment condition of time overlap and quality indicator feature similarity, a set of target batches that meet the matching requirements is selected. For the target batch set, historical health indicator data corresponding to the batch traceability index are retrieved from the batch monitoring index table. Taking the time of the abnormality as the starting point for backtracking, the health indicators and corresponding offsets and offset change rates of each time interval are extracted in the order of monitoring time, and a health indicator backtracking sequence is generated. For the target batch set, historical quality index data corresponding to the batch traceability index is retrieved from the batch monitoring index table. Taking the production segment where the anomaly occurred as the starting point for backtracking, the quality index and corresponding offset and offset change rate of each time interval are extracted in the order of production segments, and a deviation pattern backtracking sequence is generated. The health indicator backtracking sequence and the deviation pattern backtracking sequence are aligned according to the dual correspondence between timestamp and production segment. The intersection segment that simultaneously meets the health indicator abnormality and quality indicator abnormality at the time of the abnormality is identified. The equipment parameter and process parameter status of the intersection segment are analyzed to generate an abnormal snapshot containing the abnormality triggering conditions, abnormal segment parameter characteristics and potential cause paths.

8. The quality traceability management system for the entire production process of aluminum profiles according to claim 7, characterized in that: In the closed-loop optimization module, an anomaly snapshot output from the anomaly backtracking module is received. Based on the anomaly triggering conditions, anomaly segment parameter characteristics, and potential cause paths in the anomaly snapshot, the quality judgment result of the target batch corresponding to the batch monitoring index table is revised, and the revised quality judgment result and the original judgment result are combined to form a revision record. Based on the revised quality judgment results, the risk coefficient, risk spread range and risk level information of the corresponding batch in the batch monitoring index table are called to generate a traceability branch containing the revision record, risk level and impact range, and the traceability branch is synchronized to the production scheduling, inventory management and customer management links; The revised quality judgment results and corresponding traceability branches are written back to the equipment analysis path and process analysis path associated with the batch in the batch monitoring index table, and the equipment parameters and process parameters in the path are updated. After the batch monitoring index table is updated, the dynamic correction mechanism of the risk assessment model and the decline trend prediction model is invoked. The parameter data of the revised equipment analysis path and process analysis path are input into the model, the parameter weight adjustment and prediction curve retraining are performed, and the optimized prediction results are output.

Images