An on-line quality tracking system for ductile cast iron pipes

Abstract

The application provides an online quality tracking system for nodular cast iron pipes, and belongs to the technical field of industrial quality process control. The system collects process data, equipment state data and detection data in real time in the whole process of blast furnace molten iron, electric furnace temperature adjustment, molten iron spheroidization, casting forming, annealing, zinc spraying and three grinding, generates a pipe body identification for each nodular cast iron pipe and establishes a quality tracking file. According to the conveying beat and position coding, a data correlation window is constructed, multi-station data is written into the file, and unit verification, disorder rearrangement, resampling alignment and missing compensation are performed on the data to form synchronized multi-station data. The system collects execution feedback, establishes a one-to-one correspondence between control instructions and execution feedback, recalculates the risk after correction and determines whether the risk meets the standard. The system realizes multi-station data traceability, risk quantification and correction closed loop, and improves the online quality stability and the controllability of the production process.

Description

Technical Field

[0001] This invention relates to the field of industrial quality process control technology, and in particular to an online quality tracking system for ductile iron pipes. Background Technology

[0002] Ductile iron pipes are widely used in municipal water supply, gas transmission and industrial pipeline systems due to their high strength, good corrosion resistance and long service life. The quality of the product is directly affected by a long process coupling factors, including the tempering state of the molten iron, the stability of the spheroidizing treatment, the fluctuation of the casting and molding, the quality of annealing and surface treatment (such as zinc spraying and three grinding), and subsequent anti-corrosion processes (lining, curing and painting).

[0003] Most existing production quality management methods rely on batch statistical analysis or single-process inspection results for quality judgment. For example, they judge products through final dimensional inspection or sampling mechanical property testing, or they issue alarms based on the parameters of a single workstation exceeding limits. However, since ductile iron pipe quality defects often originate from the cumulative effects of multiple processes, the same anomaly can manifest differently under different conveying cycles, equipment load states, or process combinations. Relying solely on single-point data or offline statistical analysis is insufficient to accurately reflect the true quality evolution of a single pipe throughout the entire production process.

[0004] Therefore, we propose an online quality tracking system for ductile iron pipes. The information disclosed above in the Background section is only for enhancing the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing an online quality tracking system for ductile iron pipes, thereby resolving the technical problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An online quality tracking system for ductile iron pipes includes: Edge IoT acquisition units are deployed at the following stations: blast furnace molten iron, electric furnace heating and tempering, molten iron spheroidization, casting and molding, annealing, zinc spraying, three grinding, water pressure, cement lining, cement curing, cement internal grinding, coating and coding packaging. They are used to collect process data, equipment status data and test data to generate acquisition records. The edge IoT gateway unit is used to generate pipe body identification and establish quality tracking files. Based on the workstation acquisition mapping table, it completes unit verification and timestamp unification. Based on the conveying cycle and position code, it establishes an association window to form multi-workstation data packets and writes them into the file. Further synchronization processing is performed to obtain synchronized multi-workstation data. The feature and risk assessment unit is used to construct a set of quality features and assess the quality risk status for filing. The error correction strategy solving unit is used to determine the set of controllable process parameters and process constraints, and solve to generate the minimum intervention error correction strategy for filing. The control execution and feedback filing unit is used to generate and issue control commands, collect execution feedback, and write it according to the strategy number. The closed-loop verification and secondary adjustment unit is used to recalculate the risk score, determine whether the target is met, trigger secondary adjustment, and record the contribution.

[0007] Furthermore, the edge IoT gateway is configured to generate a pipe body identifier when the ductile iron pipe enters the starting position of the production line, and perform a uniqueness verification on the pipe body identifier; when the pipe body identifier fails to be read or a conflict occurs, the corresponding quality tracking file is set to a pending confirmation state and subsequent multi-station data is prohibited from being written to the quality tracking file until a second read or manual review removes the pending confirmation state.

[0008] Furthermore, the edge IoT gateway unit is configured to establish a workstation acquisition mapping table. The workstation acquisition mapping table includes at least the workstation number, acquisition channel number, field name, field unit, sampling period, effective range, timestamp source and location code source, and performs unit verification, validity verification and unified clock reference timestamp conversion on the acquisition records accordingly.

[0009] Furthermore, the edge IoT gateway is configured to establish a data association window based on the conveyor cycle and the workstation position code trigger point, and to assign the collection record whose timestamp falls into the data association window and whose position code matches to the corresponding pipe body identifier; when the same collection record falls into multiple data association windows, the workstation with the smallest absolute value of the difference between the timestamp and the arrival time is selected as the assigned workstation.

[0010] Furthermore, the edge IoT gateway unit is configured to perform idempotency verification on multi-station data packets. Idempotency verification includes searching for whether the same data packet already exists based on the gateway identifier and packet sequence number; if it already exists, it is discarded and a duplicate write flag is recorded; if it does not exist, it is written and a write success flag is recorded.

[0011] Furthermore, the edge IoT gateway is configured to perform synchronization processing on multi-station data packets. The synchronization processing includes out-of-order reordering, resampling alignment according to a uniform sampling period, and missing value compensation. The missing value compensation uses the previous value to maintain and adds a compensation mark to the compensation value. When there is no synchronized value before the missing interval, the compensation value is set to the reference mean of the field in the reference parameter area.

[0012] Furthermore, the feature construction and risk assessment unit is configured to construct a quality feature set from synchronized multi-station data. The quality feature set includes at least the average level, volatility, and compensation ratio, and is aggregated into feature groups according to risk type. The risk assessment calculates a risk score based on the reference mean, allowable deviation range, and weight in the reference parameter area, determines the risk level and risk type, and uses the field with the largest standardized deviation in the feature group corresponding to the risk type with the highest risk score as the root cause indication information.

[0013] Furthermore, the error correction strategy solution unit is configured to call the risk type station parameter association rule table, determine the candidate effective station based on the root cause indication information, and filter the set of controllable process parameters; when the root cause source station does not have controllable process parameters, it is expanded to the upstream adjacent station set according to the station sequence table to form the candidate effective station extended set.

[0014] Furthermore, the error correction strategy solving unit is configured to construct an error correction decision problem that includes a risk prediction term, a minimum intervention term, a production cost term, and a constraint violation penalty term. The minimum intervention term is the sum of the absolute values ​​of the parameter adjustments, and the constraints include at least parameter window constraints and single adjustment magnitude constraints. By combining and traversing the finite set of candidate adjustment quantities generated by the preset adjustment step size of each parameter, the candidate adjustment quantity vector with the smallest objective function value and the smallest sum of the absolute values ​​of the parameter adjustments is selected as the parameter adjustment quantity set of the minimum intervention error correction strategy.

[0015] Furthermore, the control execution and feedback filing unit is configured to attach a strategy number and instruction sequence number to each control instruction, and establish a one-to-one correspondence between control instructions and execution feedback data based on the strategy number, actuator number, and instruction sequence number; the execution completion determination adopts the execution error and stable duration threshold. When the execution error is not greater than the allowable error threshold within the continuous stable duration, the execution is determined to be complete; otherwise, the reason for failure is recorded; the closed-loop verification and secondary adjustment unit is configured to generate a compliance mark based on the risk score after correction and the preset risk threshold, and update the process constraints or expand the candidate effective workstations according to the execution completion status to trigger secondary adjustment when the compliance is not met. At the same time, the risk improvement amount is calculated and allocated according to the proportion of parameter adjustment amount to obtain the risk improvement contribution of each controllable process parameter and filed.

[0016] The beneficial effects of this invention are as follows: This invention generates a unique pipe body identifier for each ductile iron pipe and establishes a multi-station data association mechanism by combining the conveying cycle and station position coding. This enables accurate binding of multi-source data with a single pipe body object during the production process, avoiding data mismatch problems caused by time offset or cycle fluctuation under continuous production conditions, thereby significantly improving the reliability and consistency of quality traceability results.

[0017] This invention constructs synchronized multi-station data by unifying units, reordering, resampling and aligning, and compensating for missing data collected from multiple workstations. Furthermore, it forms a quality feature set, enabling data from different equipment systems and different sampling periods to participate in risk assessment under a unified time reference, thereby improving the accuracy of quality status identification under complex process conditions.

[0018] This invention calculates the quality risk status based on a set of quality features and locates the risk source workstation through root cause indication information, realizing the transformation from traditional result detection to process risk identification. This allows potential quality anomalies to be identified in advance before the product is completed, thereby reducing the probability of batch quality defects.

[0019] This invention constructs a corrective decision-making mechanism that includes risk prediction, minimum intervention, production cost, and constraint penalty. Under the premise of meeting process constraints, it solves the minimum intervention corrective strategy, which effectively controls the adjustment range of process parameters, avoids large-scale manual experience-based adjustments from disturbing the stability of production cycle time, and improves the stability of production process operation.

[0020] This invention establishes a one-to-one correspondence between control commands and execution feedback data, and recalculates and verifies the risk status after correction, thereby achieving online closed-loop evaluation of the correction effect. When the preset risk requirements are not met, a secondary adjustment is automatically triggered, thus forming a continuously optimized quality control closed loop and improving the system's adaptive adjustment capability.

[0021] This invention calculates and records the contribution of each controllable process parameter to risk improvement, enabling the quality improvement results to be traced back to specific workstations and parameters. This provides a quantitative basis for subsequent process optimization, parameter model updates, and the accumulation of production experience, further enhancing the intelligence and precision of production quality control. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the framework of an online quality tracking system for ductile iron pipes according to the present invention; Figure 2 This is a schematic diagram of an online quality tracking method for ductile iron pipes according to the present invention. Detailed Implementation

[0023] 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.

[0024] Example 1: As Figure 2As shown, this embodiment provides an online quality tracking method for ductile iron pipes, including the following steps: S1. Generate a unique pipe body identifier for each ductile iron pipe in the production line and establish a quality tracking file. Collect process data, equipment status data, and test data at the following stations: blast furnace molten iron, electric furnace heating and tempering, molten iron spheroidization, casting, annealing, zinc spraying, three grinding processes, hydraulic pressing, cement lining, cement curing, cement internal grinding, coating, and inkjet coding and packaging. Link multi-station data and write them into the quality tracking file based on the conveyor cycle time and location code. Specifically, this includes: S110. Pipe body identification generation and binding: When the ductile iron pipe is formed at the casting and molding station and enters the starting position of the subsequent conveying line, the edge IoT edge gateway reads the starting position trigger signal to determine the generation time. The starting position trigger signal is the electrical signal generated when the ductile iron pipe passes the starting position sensor or the starting position encoder.

[0025] At the same time, the edge IoT edge gateway will extract and associate the batch process data and equipment status data generated at the three workstations of the blast furnace molten iron, electric furnace heating and tempering, and molten iron spheroidization of the batch to which the pipe body belongs with the identification of the pipe body.

[0026] The edge IoT edge gateway generates a pipe body identifier at the time of generation. The pipe body identifier is obtained by concatenating the production line number, the generation time, and the shift increment number. The production line number is used to distinguish different production lines, the generation time is used to distinguish different time slices, and the shift increment number is used to distinguish different ductile iron pipes within the same time slice.

[0027] The edge IoT edge gateway performs a uniqueness check on the generated pipe body identifier. The uniqueness check involves searching the quality traceability file to see if the same pipe body identifier already exists.

[0028] When the same pipe identifier is found to exist, the edge IoT edge gateway marks the pipe identifier as a conflict identifier and sets the corresponding quality tracking file to a pending confirmation status.

[0029] When no matching pipe identifier is found, the edge IoT edge gateway marks the pipe identifier as a valid identifier and establishes a one-to-one correspondence between the pipe identifier and the quality tracking file.

[0030] The edge IoT edge gateway outputs the pipe body identification to the identification writing device to complete the physical binding. The identification writing device can be any one of inkjet printing device, laser marking device, or RFID tag writing device.

[0031] If the pipe body identifier read by a subsequent workstation is inconsistent with the valid identifier or the reading fails, the edge IoT edge gateway will set the corresponding quality tracking file to a pending confirmation state and prohibit subsequent multi-workstation data from being written into the quality tracking file until a consistent pipe body identifier is obtained through a second reading or manual verification, at which point the pending confirmation state will be lifted.

[0032] S120. Edge IoT Data Acquisition Configuration and Workstation Mapping: Edge IoT acquisition terminals and edge IoT gateways are deployed at each workstation in the following processes: blast furnace molten iron, electric furnace heating and tempering, molten iron spheroidization, casting, annealing, zinc spraying, three grinding processes, hydraulic pressing, cement lining, cement curing, cement internal grinding, coating, and inkjet printing and packaging. The edge IoT acquisition terminals are used to access signals from sensors or equipment controllers and generate acquisition records. The edge IoT gateways are used to aggregate the acquisition records and perform preprocessing, caching, and command issuance. Each process has a corresponding detection workstation.

[0033] The edge IoT edge gateway establishes a workstation acquisition mapping table for each workstation. The workstation acquisition mapping table includes at least the workstation number, acquisition channel number, field name, field unit, sampling period, data type, effective range, noise reduction method, timestamp source, and location encoding source.

[0034] The edge IoT edge gateway standardizes the units of the collected records based on the workstation data acquisition mapping table, converting field values ​​in the records to their corresponding units and retaining the original unit markers. The edge IoT edge gateway also performs validity checks on the collected records based on their valid range. Field values ​​outside the valid range are marked as outliers and the outlier marker is retained; field values ​​within the valid range are marked as valid.

[0035] The edge IoT edge gateway provides a unified clock reference for this production line. When the timestamp of the collected record comes from the local clock of the acquisition terminal, the local timestamp of the acquisition terminal is converted into a timestamp under the unified clock reference and written into the acquisition record. The timestamp under the unified clock reference is used for establishing the data association window in S130 and for resampling alignment in S210.

[0036] S130, Multi-station data association writing: The edge IoT edge gateway sets a position coding trigger point at each station. The position coding trigger point is a discrete trigger event generated by the conveyor line position encoder or the station entrance detector.

[0037] When the edge IoT edge gateway detects that the ductile iron pipe corresponding to the pipe body identifier has arrived at the position code trigger point of station j, it records the arrival time as follows: Where j is the workstation number index, The arrival time of workstation j.

[0038] The edge IoT edge gateway calculates the data association window of workstation j based on the delivery cycle time T. Where T is the conveying cycle time, and α is... For data association window coefficients, For the data association window of workstation j, calculate according to the following formula: Edge IoT edge gateway will The pipe body identification and workstation number are written into the quality tracking file for subsequent verification and attribution determination.

[0039] The edge IoT edge gateway performs an attribution determination on the data collection records of workstation j. The attribution determination is based on the timestamp of the data collection record. fall into Furthermore, if the location code carried in the data acquisition record matches the location code trigger point of workstation j, the data acquisition record is determined to belong to the data of the pipe body identified at workstation j. To provide timestamps under a unified clock reference.

[0040] When the timestamp of the same collection record When a candidate falls into multiple candidate windows simultaneously, the edge IoT edge gateway performs conflict attribution selection, which determines the selection process. The workstation corresponding to the smallest candidate window is designated as the assigned workstation. ,in The final assigned workstation number is calculated using the following formula: When the quality traceability file is in a pending confirmation state, the edge IoT edge gateway prohibits writing multi-station data packets and writes a prohibition mark to the control execution area for prompting review.

[0041] The edge IoT edge gateway groups the collection records of each workstation belonging to the same pipe body identifier into multi-workstation data packets according to the workstation number. The multi-workstation data packet includes at least the pipe body identifier, packet sequence number, workstation number set, collection record set, collection record timestamp, location code, anomaly mark and unit mark before unit conversion.

[0042] Before writing to the quality traceability file, the edge IoT edge gateway performs idempotency verification on the multi-station data packets. Idempotency verification involves checking whether a data packet with the same tube identifier and sequence number already exists in the quality traceability file. If it exists, the packet is discarded and a duplicate write flag is recorded; otherwise, it is written and a write success flag is recorded. The write success flag serves as the basis for the S210 to confirm the integrity of the input when reading the multi-station data packets.

[0043] S2. Perform unit verification, time synchronization, resampling alignment, and missing data compensation on multi-station data to form synchronized multi-station data, and construct a quality feature set, specifically including: S210. Synchronization processing based on workstation acquisition mapping table: The edge IoT edge gateway reads the multi-workstation data packets corresponding to the same pipe body identifier in the quality tracking file, and extracts the workstation number set, acquisition record set, acquisition record timestamp, anomaly mark and unit mark before unit conversion.

[0044] The edge IoT edge gateway checks the consistency between the unit marker before unit conversion and the field unit for each collection record based on the workstation collection mapping table. If there is a discrepancy, the collection record is marked as having a unit inconsistency and set as an anomaly.

[0045] The edge IoT edge gateway performs validity checks on each field based on its valid range. Field values ​​outside the valid range are marked as abnormal and retained, while field values ​​within the valid range are retained as valid values.

[0046] The edge IoT edge gateway sorts the valid values ​​of each field by timestamp from smallest to largest and then performs random reordering. For duplicate values ​​with the same timestamp, only the first valid value that arrived at the edge IoT edge gateway is retained, and the duplicate collection mark is written into the quality tracking file.

[0047] Edge IoT edge gateways operate under a unified clock reference and a unified sampling period. Establish a continuous time interval series, where To ensure a consistent sampling period, the larger of the sampling period for this field and the preset minimum alignment period is used.

[0048] The edge IoT edge gateway calculates the synchronized values ​​for the i-th unified time interval. Where i is the index of the unified time interval number, and k is the index of the sample number. For the kth valid field value, This is the set of valid value indices falling within the i-th uniform time interval. For set The number of elements is calculated using the following formula: when When the value is 0, the edge IoT edge gateway determines that the unified time interval is a missing interval and performs missing compensation. The missing compensation adopts the previous value preservation method, which uses the most recent synchronized value before the missing interval as the compensation value for the missing interval and adds a compensation mark; when there is no synchronized value before the missing interval, the compensation value is set to the field reference mean and a compensation mark is added. The field reference mean is taken from the reference parameter area.

[0049] The edge IoT edge gateway aggregates the synchronized timestamp sequence and synchronized numerical sequence of each field to form synchronized multi-station data, and writes the synchronized multi-station data into the quality traceability file as S220 input.

[0050] S220. Construction of Quality Feature Set for Pipe Identification: The edge IoT edge gateway reads the synchronized multi-station data in the quality tracking file, groups the synchronized multi-station data according to the station number and field name, and obtains the synchronized numerical sequence of each field under each station.

[0051] The edge IoT edge gateway removes synchronized values ​​marked with anomalies from the synchronized value sequence of each field, and retains synchronized values ​​marked with compensation to participate in the calculation. At the same time, it calculates the compensation ratio and writes it into the quality tracking file. The compensation ratio is the ratio of the number of compensation marks to the total number of synchronized values.

[0052] The edge IoT edge gateway calculates the average level and fluctuation level of the synchronized numerical sequence for each field. The average level is calculated by "summing the synchronized values ​​and dividing by the quantity", and the fluctuation level is calculated by "summing the squares of the differences between the synchronized values ​​and the average level and dividing by the quantity and taking the square root". Synchronized values ​​marked with anomalies are ignored during the calculation.

[0053] The edge IoT edge gateway aggregates the average level, fluctuation degree and compensation ratio of each field into feature groups according to the risk type based on the quality correlation tags in the workstation acquisition mapping table. The risk types include at least the risk of fluctuation in front-end molten iron composition, abnormal risk of electric furnace heating and tempering, risk of molten iron spheroidization decay, risk of casting and forming defects, risk of deviation in annealing temperature curve, risk of surface treatment (zinc spraying / three grinding) quality, risk of leakage in water pressure test, risk of cracking in cement lining and curing, and risk of abnormal coating adhesion.

[0054] The edge IoT edge gateway merges the characteristic groups of each risk type according to the pipe body identifier to form a quality characteristic set, and writes the quality characteristic set into the quality tracking file as an input of S230.

[0055] S230, Quality Risk Assessment and Filing: The edge IoT edge gateway reads the quality feature set in the quality tracking file and reads the corresponding reference mean value for each feature from the reference parameter area. Permissible deviation range With weight , where p is the quality feature number index.

[0056] The edge IoT edge gateway calculates the standardized deviation for each quality feature. ,in The p-th quality characteristic value for the current pipe body is calculated using the following formula: when When the value is 0 or the feature is unavailable, the edge IoT edge gateway marks the feature as reference unavailable and does not participate in the risk scoring calculation. At the same time, the reference unavailable mark is written into the quality tracking file.

[0057] The edge IoT edge gateway calculates a risk score for the feature group corresponding to the risk type. The risk score is a weighted sum of the features contained in that risk type, calculated according to... Calculate, where M is the total number of features involved in the calculation within the feature group of this risk type.

[0058] The edge IoT edge gateway compares the risk score with the classification threshold in the reference parameter area to determine the risk level. The risk level includes at least low risk, medium risk, and high risk. The edge IoT edge gateway determines the risk type with the highest risk score as the current pipe identification risk type, and selects the feature with the largest standardized deviation from the feature group of this risk type as the root cause indication information. The root cause indication information includes at least the feature field name and the source workstation number.

[0059] Preferred weight Based on statistical data of quality defects in historical batches of products, and pre-determined using the Hierarchical Analysis of Experts (AHP), this data characterizes the significance of the combined impact of different quality characteristics on this type of risk. The edge IoT edge gateway combines the risk type, risk level, risk score, and root cause indication information into a quality risk status and writes it into the quality tracking file as an S310 input.

[0060] S3. Conduct risk assessment based on the quality feature set to obtain the quality risk status, which includes risk type, risk level, and root cause indication information, and write it into the quality tracking file. Specifically, this includes: S310. Determination of controllable process parameter set and process constraints: The edge IoT edge gateway reads the quality risk status in the quality tracking file and extracts the risk type, risk level, risk score and root cause indication information.

[0061] The edge IoT edge gateway calls the risk type-workstation-parameter association rule table stored in the reference parameter area. The risk type-workstation-parameter association rule table must at least include the risk type, candidate effective workstation, controllable process parameter name, parameter adjustable direction, and parameter lower limit. Parameter upper limit Maximum adjustment range in a single instance Maximum allowed execution time Preset adjustment step size Actuator number, permissible error threshold And the rhythm influence factor.

[0062] The edge IoT edge gateway determines the priority workstation in the workstation sequence table based on the source workstation number in the root cause indication information. The priority workstation is the workstation with the same source workstation number.

[0063] When a priority workstation does not have a controllable process parameter in the risk type-workstation-parameter association rule table, the edge IoT edge gateway generates an upstream adjacent workstation set based on the workstation sequence table and uses it as an extended set of candidate effective workstations.

[0064] The edge IoT edge gateway filters the set of controllable process parameters corresponding to the risk type from the risk type-workstation-parameter association rule table, and at the same time determines the set of corresponding candidate effective workstations and corresponding execution agency numbers.

[0065] The edge IoT edge gateway reads back the setting register from the actuator controller to obtain the current actual set value of each controllable process parameter. and the This serves as the input for the target setpoint calculated by S410; this input is compared with the actual setpoint in the execution feedback data. The two are consistent, and no further distinction will be made between their respective standards.

[0066] The edge IoT edge gateway establishes process constraints for each controllable process parameter. These constraints include at least parameter window constraints and single adjustment range constraints. The parameter window constraints are determined by… and The single adjustment range constraint is given by Provided.

[0067] The edge IoT edge gateway will control the set of process parameters, candidate active workstations, and current actual set values. Maximum allowed execution time Preset adjustment step size The process constraints are written into the quality tracking file as S320 input.

[0068] S320. Problem Construction for Corrective Decision-Making: The edge IoT edge gateway constructs a parameter adjustment vector by combining the single adjustment amounts of each controllable process parameter. Each dimension is .

[0069] Risk prediction items for building edge IoT edge gateways Risk prediction items are based on the pre-correction risk score. Based on this, a risk sensitivity coefficient is introduced from the reference parameter area to estimate risk changes. The risk sensitivity coefficient is the linear coefficient of each controllable process parameter to the risk score change.

[0070] Edge IoT edge gateway will perform risk scoring before correction. The risk score is taken as S230 and written into the quality tracking file.

[0071] Edge IoT edge gateways calculate risk prediction items using a linear approximation , calculated as in The risk sensitivity coefficient is stored in the reference parameter area; when the adjustable parameter direction requires risk reduction, the edge IoT edge gateway only allows... Take Non-positive direction, otherwise it will be Set to 0 to treat it as unadjustable.

[0072] Specifically, risk sensitivity coefficient It is pre-determined by performing multiple linear regression analysis on the adjustment records of each controllable process parameter in the historical production data of ductile iron pipes and the corresponding changes in risk scores. Its physical meaning is the expected change rate of risk score caused by each unit adjustment of the process parameter.

[0073] Edge IoT edge gateway construction production cost item The production cost item is calculated by summing the absolute value of the cycle time impact factor and the adjustment range. The cycle time impact factor comes from the risk type-workstation-parameter association rule table.

[0074] Edge IoT edge gateway construction constraint violation penalty item When any constraint is violated, let When all constraints are satisfied, it means ,in A preset large penalty value is set and stored in the reference parameter area.

[0075] Edge IoT edge gateway construction objective function And with the goal of minimization, it is calculated using the following formula: in, The minimum intervention item weight coefficient, For the weighting coefficient of the production cost item, To constrain the weight coefficients of the violation penalty items, they are all pre-stored in the reference parameter area.

[0076] In practical applications, to ensure the absolute priority of process constraints, the weight coefficient of the constraint violation penalty term is determined. Set to much greater than and Positive numbers (e.g., values ​​of 10000); and and Values ​​are assigned based on the current production strategy. When the production environment prioritizes ensuring pipeline cycle stability, the minimum intervention item is assigned a weight coefficient. Larger values.

[0077] The edge IoT edge gateway adds parameter window constraints and single adjustment range constraints to each controllable process parameter, established in the following constraint form: The edge IoT edge gateway combines the objective function and constraints to form a corrective decision problem and writes it into the quality traceability file as an input to the S330.

[0078] S330, Minimal Intervention Correction Strategy Output and Input: The edge IoT edge gateway generates a finite set of candidate adjustment values ​​for each controllable process parameter. The finite set of candidate adjustment quantities is composed of The values ​​that violate the adjustable direction of the parameters are constructed and eliminated. This is the preset adjustment step size for the m-th controllable process parameter.

[0079] The edge IoT edge gateway performs a combination operation on the finite set of candidate adjustment values ​​for each controllable process parameter to generate a set of candidate adjustment value vectors. , where q is the index of the candidate adjustment vector.

[0080] Edge IoT edge gateway for each candidate adjustment vector If any constraint is violated, discard the candidate vector.

[0081] The edge IoT edge gateway calculates the objective function value for each feasible candidate vector. The candidate vector with the smallest objective function value is selected as the optimal adjustment vector. When multiple candidate vectors have the same minimum objective function value, the edge IoT edge gateway selects... The smallest one as To ensure minimal intervention.

[0082] Edge IoT edge gateway based on Generate a minimum intervention correction strategy, which should include at least a strategy number, candidate effective workstations, a set of parameter adjustment amounts, a target effective time period, safety verification conditions, and a maximum allowed execution time. .

[0083] The strategy number is generated by the edge IoT edge gateway and is guaranteed to be unique in the quality traceability file. It is used to establish the correspondence between control commands and execution feedback data in S410–S430. The edge IoT edge gateway writes the minimum intervention correction strategy into the quality traceability file as an input to S410.

[0084] S4. Based on the quality risk status, determine the candidate effective workstations and the set of controllable process parameters. Under the condition of satisfying process constraints, solve for the minimum intervention correction strategy and generate control command execution parameter adjustments. Specifically, this includes: S410, Control Command Generation and Edge IoT Issuance: The edge IoT edge gateway reads the minimum intervention correction strategy in the quality tracking file and extracts the strategy number, candidate effective workstation, parameter adjustment set, target effective time period, security verification conditions and maximum allowed execution time.

[0085] The edge IoT edge gateway splits the parameter adjustment set into parameter instruction entries according to the actuator number. Each parameter instruction entry contains at least the policy number, candidate effective workstation number, actuator number, target parameter name, and current actual set value. Parameter adjustment amount Maximum allowed execution time With preset adjustment step size .

[0086] The target setpoint for the edge IoT edge gateway is calculated using the following formula. : The edge IoT edge gateway verifies the security conditions before distribution. Whether the parameter window constraint and the single adjustment range constraint are met. If either constraint is not met, it is marked as a constraint conflict state and the issuance is prohibited. At the same time, the constraint conflict state is written into the control execution area as the basis for S520 non-compliance type judgment and constraint update.

[0087] The edge IoT edge gateway encapsulates the issueable parameter command items into control commands and sends them to the corresponding actuator controller of the candidate effective workstation. The control command includes at least the strategy number, command sequence number, actuator number, target parameter name, and target setpoint. Maximum allowed execution time With the effective start time.

[0088] The instruction sequence number is a sequence number generated by incrementing the issuance time under the same policy number, and is used for S430 idempotent writing. The edge IoT edge gateway writes the control instruction issuance time, policy number, executor number, and instruction sequence number into the control execution area, which serves as the basis for the S430 index field and the starting point of the S510 recalculation window.

[0089] S420, Actuator Correction Execution and Acknowledgment Generation: After receiving the control command, the actuator controller sets the target setpoint... Write the settings into the controller's register and drive the actuator to adjust the corresponding controllable process parameters.

[0090] The edge IoT acquisition terminal collects and reports execution feedback data during execution. The execution feedback data includes at least the strategy number, instruction sequence number, actuator number, and target set value. Actual set value Actual measurement value sequence The data collection process includes collecting timestamp sequences, execution status words, and fault codes. The edge IoT edge gateway calculates the execution error based on the actual measurement sequence. Calculate using the following formula: The edge IoT edge gateway reads the allowable error threshold from the risk type-workstation-parameter association rule table. And read the stable duration threshold from the reference parameter area. ,in As the allowable error threshold, This is the threshold for the stable duration.

[0091] Edge IoT edge gateway for continuous satisfaction The duration is accumulated, and when the accumulated duration reaches... When the execution is completed, the status will be set to "completed" and the execution completion time will be recorded as the first time the cumulative duration is reached. At that moment.

[0092] When the maximum allowed execution time is exceeded The cumulative duration has not yet been reached. When the execution is completed, the edge IoT edge gateway will set the execution status to failure and record the reason for failure. The reason for failure can be determined from the fault code, execution status word, or long-term non-convergence of execution error.

[0093] The edge IoT edge gateway will execute the status, completion time, and failure reason. and Write execution feedback data as input for S430 filing and S520 non-compliance type determination.

[0094] S430, Write execution feedback data into quality tracking file: The edge IoT edge gateway establishes the correspondence between control command and execution feedback data based on the policy number, execution mechanism number and command sequence number and writes it into the control execution area.

[0095] Before writing execution feedback data, the edge IoT edge gateway performs idempotency verification. Idempotency verification involves checking whether there is already an execution feedback record with the same policy number, the same actuator number, and the same instruction sequence number in the quality tracking file.

[0096] When an existing record already exists, the edge IoT edge gateway retains only the latest execution feedback record in the collection timestamp sequence and marks the earlier records as historical records.

[0097] When it does not exist, the edge IoT edge gateway will write the execution feedback record to the execution data area and add a write success flag.

[0098] Edge IoT edge gateway will set target values Actual set value Actual measurement value sequence Execution error sequence The execution status, completion time, and reasons for failure are recorded in the quality tracking file, including the execution error sequence. It is obtained by calculating point by point from S420.

[0099] The correspondence between control commands and execution feedback data serves as the index for determining the S510 recalculation window, updating S520 constraints, and calculating the traceable contribution of S530.

[0100] S5. Collect execution feedback data and establish the correspondence between control commands and execution results; recalculate and determine the compliance status after corrective action; trigger secondary adjustments when compliance is not met; and calculate the contribution of parameters to risk improvement and write it into the quality tracking file. Specifically, this includes: S510. Recalculation of quality risk status and calculation of risk improvement after correction: The edge IoT edge gateway reads the execution feedback data corresponding to the strategy number from the quality tracking file, extracts the execution completion time and execution status, and uses the execution completion time as the starting time of the recalculation.

[0101] The edge IoT edge gateway takes the start time of the recalculation as the starting point, extracts the newly added workstation data of the next continuous conveying cycle T as the recalculation data window after correction, and writes the window identifier into the control execution area to record the recalculation caliber.

[0102] The edge IoT edge gateway generates synchronized multi-station data after correction by adding new multi-station data in the corrected recalculation data window according to S210, and generates a set of quality features after correction according to S220.

[0103] The edge IoT edge gateway calculates the post-correction risk score for the post-correction quality feature set according to the same calculation caliber and the same reference parameter area of ​​S230, determines the post-correction quality risk status, and writes the post-correction quality risk status into the quality tracking file.

[0104] Edge IoT edge gateways read pre-correction risk scores from quality traceability files as... And read the corrected risk score as .

[0105] The risk improvement amount for edge IoT edge gateways is calculated using the following formula. : Edge IoT edge gateway will , and Write it to the control execution area as an input to S520 and S530.

[0106] S520, Preset Risk Threshold Determination and Secondary Adjustment Trigger: The edge IoT edge gateway reads the preset risk threshold from the reference parameter area. The preset risk threshold is the risk scoring threshold configured for the risk type after correction.

[0107] Edge IoT edge gateways will correct risk scores Comparison, when The target is then set to "achieved" and written to the control execution area.

[0108] when > The edge IoT gateway sets the compliance flag to non-compliance and writes it to the control execution area. When the compliance flag is non-compliance, the edge IoT gateway reads the execution status and failure reason and determines the non-compliance type. When the execution status is failure, the non-compliance type is set to execution incomplete. When the execution status is complete but still non-compliance, the non-compliance type is set to risk not sufficiently mitigated.

[0109] When the non-compliance type is "execution incomplete," the edge IoT edge gateway updates the process constraints based on the failure reason and writes them into the quality tracking file. The update method includes at least adjustments to the corresponding parameters. Lower, to The candidate workstation may be moved up or switched to the next candidate workstation in the upstream adjacent workstation set.

[0110] When the non-compliance type is that the risk has not been sufficiently improved, the edge IoT edge gateway will expand the candidate effective workstations to the set of upstream adjacent workstations and enable more controllable process parameter sets. At the same time, the root cause indication information after correction will be written into the quality tracking file as the root cause input for secondary adjustment.

[0111] After updating process constraints or expanding candidate active workstations, the edge IoT gateway triggers secondary adjustments. This secondary adjustment involves returning to execution S310 to redefine the set of controllable process parameters and process constraints, and resolving the minimum intervention correction strategy. The edge IoT gateway increments the number of secondary adjustments and records the triggering reason in the control execution area, serving as input for subsequent auditing and strategy effectiveness evaluation.

[0112] S530, Risk Improvement Contribution Calculation and Closed-Loop Result Consolidation: The edge IoT edge gateway reads the minimum intervention correction strategy corresponding to the strategy number in the quality tracking file and extracts the set of parameter adjustment amounts for the current non-zero adjustment. The corresponding set of actuator numbers is used, and the risk improvement amount ΔR written by S510 is read. The edge IoT edge gateway calculates the total adjustment range. The total adjustment is the sum of the absolute values ​​of all non-zero adjustments in this adjustment, i.e. Here, N represents the number of parameters that underwent non-zero adjustments in this instance. When =0, the edge IoT edge gateway sets the risk improvement contribution of all parameters to 0 and writes it to the unadjusted flag. When the value is >0, the edge IoT edge gateway calculates the risk improvement contribution for each parameter that undergoes a non-zero adjustment. Calculate using the following formula: Edge IoT edge gateway will store each parameter With the actuator number, target parameter name, and target set value The execution status, execution completion time, and failure reasons are all written into the quality tracking file, and a correlation is established between the strategy number and the control command-execution feedback data to ensure that the contribution is traceable.

[0113] The edge IoT edge gateway will display the quality risk status before and after the correction. The compliance markers, non-compliance types, number of secondary adjustments, and triggering reasons are solidified and written into the quality tracking file. The solidified results are used as historical reference inputs for subsequent batch S230 risk assessments and as training samples for offline updating of reference parameter area weights and risk sensitivity coefficients, thus forming an online quality tracking closed loop of tracking-correction-verification-recorrection.

[0114] Example 2: Figure 1 As shown, this embodiment provides an online quality tracking system for ductile iron pipes, including: The edge IoT acquisition unit includes edge IoT acquisition terminals deployed at multiple workstations, including blast furnace molten iron, electric furnace heating and tempering, molten iron spheroidization, casting and molding, annealing, zinc spraying, three grinding, water pressure, cement lining, cement curing, cement internal grinding, coating and coding packaging. It is used to collect process data, equipment status data and test data related to the quality of ductile iron pipes, and generate acquisition records containing field values, timestamps and location codes. The edge IoT gateway unit, including an edge IoT edge gateway that is communicatively connected to the edge IoT acquisition terminal, is used for: Generate a pipe body identifier corresponding to each ductile iron pipe and establish a quality tracking file with the pipe body identifier as the primary key; perform unit unification, validity verification, and unified clock reference timestamp conversion on the collected records based on the workstation acquisition mapping table; establish a data association window based on the conveyor cycle and workstation position code trigger point, assign multi-workstation acquisition records to the corresponding pipe body identifier and form multi-workstation data packets, perform idempotency verification and write them into the quality tracking file; perform out-of-order rearrangement, resampling alignment and missing data compensation on the multi-workstation data packets to form synchronized multi-workstation data. The feature construction and risk assessment unit is used to: calculate a quality feature set based on synchronized multi-workstation data. The quality feature set includes at least the average level, volatility, and compensation ratio, and retains the field names and source workstation numbers; aggregate the quality feature set by risk type and calculate the risk score, determine the risk type, risk level, and root cause indication information, form the quality risk status, and write it into the quality tracking file. The minimum intervention correction strategy solution unit is used to: determine candidate effective workstations, controllable process parameter sets, and process constraints based on the quality risk status and risk type-workstation-parameter association rule table; Construct a corrective decision problem that includes risk prediction, minimum intervention, production cost, and constraint violation penalty, and solve it to obtain the parameter adjustment vector. Generate a minimum intervention corrective strategy that includes strategy number, parameter adjustment set, target effective period, and safety verification conditions, and write it into the quality tracking file. The control execution and acknowledgment filing unit is used to: generate control instructions based on the minimum intervention correction strategy, including strategy number, instruction number, actuator number, target parameter name, target set value and maximum allowable execution time, and issue them to the actuator controllers of candidate effective workstations; Collect and generate execution feedback data that includes actual set values, actual measured value sequences, execution errors, execution status, execution completion time and failure reasons. Establish the correspondence between control commands and execution feedback data based on "strategy number + executor number + command sequence number" and perform idempotent writing. Write the execution feedback data into the quality tracking file. The closed-loop verification and secondary adjustment unit is used to: extract the post-correction recalculation data window starting from the execution completion time; recalculate the post-correction risk score and calculate the risk improvement amount according to the same caliber as the risk assessment; compare the post-correction risk score with the preset risk threshold to generate a compliance mark; when the compliance mark is not met, update the process constraints or expand the candidate effective workstations based on the execution status and failure reasons, and trigger secondary adjustment to return to the minimum intervention correction strategy solution unit for re-solution; calculate the risk improvement contribution of each controllable process parameter based on the risk improvement amount and parameter adjustment amount, and write it into the quality tracking file. The quality tracking file includes at least a reference parameter area, a control execution area, and an execution data area. The reference parameter area is used to store the reference mean, allowable deviation range, weighting coefficient, grading threshold, preset risk threshold, stability judgment threshold, and preset maximum penalty value. The control execution area is used to store control command issuance records, execution status, and number of secondary adjustments. The execution data area is used to store the target set value, actual set value, actual measurement value sequence, and execution error sequence.

[0115] The above formulas are all dimensionless calculations. The preset parameters, grading thresholds, and stability judgment thresholds in the formulas are specific values ​​obtained by those skilled in the art based on the specific specifications of the ductile iron pipes produced (such as DN800 pipes), the response accuracy of the equipment controller, and the allowable tolerance range of the corresponding metallurgical process standards, combined with historical trial production data.

[0116] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0117] In conclusion, the above are merely preferred embodiments of the present invention and are 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. An online quality tracking system for ductile iron pipes, characterized in that, include: Edge IoT acquisition units are deployed at the following stations: blast furnace molten iron, electric furnace heating and tempering, molten iron spheroidization, casting and molding, annealing, zinc spraying, three grinding, water pressure, cement lining, cement curing, cement internal grinding, coating and coding packaging. They are used to collect process data, equipment status data and test data to generate acquisition records. The edge IoT gateway unit is used to generate pipe body identification and establish quality tracking files. Based on the workstation acquisition mapping table, it completes unit verification and timestamp unification. According to the conveying cycle and position code, it establishes an association window to form multi-workstation data packets and writes them into the file. Synchronization processing is used to obtain synchronized multi-workstation data. The feature and risk assessment unit is used to construct a set of quality features and assess the quality risk status for filing. The error correction strategy solving unit is used to determine the set of controllable process parameters and process constraints, and solve to generate the minimum intervention error correction strategy for filing. The control execution and feedback filing unit is used to generate and issue control commands, collect execution feedback, and write it according to the strategy number. The closed-loop verification and secondary adjustment unit is used to recalculate the risk score, determine whether the target is met, trigger secondary adjustment, and record the contribution.

2. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The edge IoT gateway is configured to generate a pipe body identifier when the ductile iron pipe enters the starting position of the production line and to perform a uniqueness check on the pipe body identifier; When the pipe body identification fails to be read or a conflict occurs, the corresponding quality tracking file will be set to a pending confirmation state and subsequent multi-station data will be prohibited from being written to the quality tracking file until a second read or manual review removes the pending confirmation state.

3. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The edge IoT gateway is configured to establish a workstation acquisition mapping table. The workstation acquisition mapping table includes at least the workstation number, acquisition channel number, field name, field unit, sampling period, valid range, timestamp source and location code source, and performs unit verification, validity check and unified clock reference timestamp conversion on the acquisition records accordingly.

4. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The edge IoT gateway is configured to establish a data association window based on the conveyor cycle and the workstation location code trigger point, and to assign the collection record whose timestamp falls into the data association window and whose location code matches to the corresponding pipe body identifier; when the same collection record falls into multiple data association windows, the workstation with the smallest absolute value of the difference between the timestamp and the arrival time is selected as the assigned workstation.

5. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The edge IoT gateway unit is configured to perform idempotency verification on multi-station data packets. Idempotency verification includes searching for whether the same data packet already exists based on the gateway identifier and packet sequence number; if it already exists, it is discarded and a duplicate write flag is recorded; if it does not exist, it is written and a write success flag is recorded.

6. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The edge IoT gateway is configured to perform synchronization processing on multi-station data packets. The synchronization processing includes out-of-order reordering, resampling alignment according to a uniform sampling period, and missing value compensation. The missing value compensation uses the previous value to maintain and adds a compensation mark to the compensation value. When there is no synchronized value before the missing interval, the compensation value is set to the reference mean of the field in the reference parameter area.

7. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The feature construction and risk assessment unit is configured to construct a set of quality features from synchronized multi-workstation data. The set of quality features includes at least the average level, volatility, and compensation ratio, and is aggregated into feature groups according to risk type. Risk assessment calculates risk scores based on the reference mean, allowable deviation range, and weights in the reference parameter area, determines the risk level and risk type, and uses the field with the largest standardized deviation in the feature group corresponding to the risk type with the highest risk score as the root cause indication information.

8. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The error correction strategy solution unit is configured to call the risk type station parameter association rule table, determine the candidate effective station based on the root cause indication information, and filter the set of controllable process parameters. When the root cause source station does not have controllable process parameters, it is expanded to the upstream adjacent station set according to the station sequence table to form the candidate effective station extended set.

9. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The error correction strategy solving unit is configured to construct an error correction decision problem that includes a risk prediction term, a minimum intervention term, a production cost term, and a constraint violation penalty term. The minimum intervention term is the sum of the absolute values ​​of the parameter adjustments, and the constraints include at least parameter window constraints and single adjustment magnitude constraints. By combining and traversing the finite set of candidate adjustment quantities generated by the preset adjustment step size of each parameter, the candidate adjustment quantity vector with the smallest objective function value and the smallest sum of the absolute values ​​of the parameter adjustments is selected as the parameter adjustment quantity set of the minimum intervention error correction strategy.

10. The online quality tracking system for ductile iron pipes according to claim 1, characterized in that, The control execution and feedback filing unit is configured to add a strategy number and instruction sequence number to each control instruction, and establish a one-to-one correspondence between control instructions and execution feedback data based on the strategy number, the execution mechanism number and the instruction sequence number; The execution status is determined by the execution error and the stability duration threshold. When the execution error is not greater than the allowable error threshold within the continuous stability duration, the execution is considered successful; otherwise, the reason for failure is recorded. The closed-loop verification and secondary adjustment unit is configured to generate a compliance mark based on the risk score after correction and the preset risk threshold. When the compliance is not met, the process constraints are updated or the candidate effective workstations are expanded according to the execution status to trigger secondary adjustment. At the same time, the risk improvement amount is calculated and allocated according to the parameter adjustment amount to obtain the risk improvement contribution of each controllable process parameter and file it.

Images