A method and system for online calibration of multi-source sensing data of a steel cable production line

Abstract

The application discloses a kind of steel cable production line multi-source sensing data online calibration method and system, it is related to sensing data online calibration technical field, including: obtaining steel cable production task, process, equipment and detection channel data, determine effective twisting frequency, effective path length between detection stations, image axial length conversion parameter and twist pitch transfer parameter, generate steel cable online calibration basis record;Form twist pitch keeping calibration segment and twist pitch change calibration segment, utilize section boundary marker to correspond multiple source sensing data to the same steel cable entity section;Respectively calculate steel cable transfer time and detection channel acquisition delay, generate candidate calibration parameter using boundary marker average transfer speed, mechanical reference position pulse and image twist pitch measurement result;Select data not involved in parameter calculation for verification, generate candidate calibration parameter verification record and parameter version generation condition record.The application improves the accuracy of cross-station data correspondence and detection channel calibration.

Description

Technical Field

[0001] This invention relates to the field of online calibration technology for sensor data, and in particular to an online calibration method and system for multi-source sensor data in a steel cable production line. Background Technology

[0002] Steel cable production lines are typically equipped with traction speed sensors, twisting component speed sensors, tension sensors, and image measurement devices to collect operating parameters and lay length data during the cable forming process. Routine online calibration generally adjusts the acquisition time, proportional parameters, and bias parameters based on equipment calibration results, standard sample measurements, or the difference between the detection channel and control feedback, and saves the calibration results by production batch.

[0003] However, existing methods still have two limitations: First, the data from different detection stations are mostly directly correlated by timestamps, and it is not easy to distinguish the transmission time of the steel cable between stations from the acquisition delay caused by sensor response, filtering and communication; Second, the fixed working conditions provide limited variation conditions, and it is not easy to identify the sources of deviation between traction speed, twisting component speed and image twist pitch, and candidate parameters lack independent segment verification. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, this invention provides an online calibration method for multi-source sensor data in a steel cable production line, which solves the problem that multi-source sensor data across workstations in a steel cable production line are difficult to accurately correspond and that the deviations of each detection channel are difficult to confirm separately.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an online calibration method for multi-source sensor data of a steel cable production line, comprising: acquiring data on steel cable production tasks, processes, equipment, and detection channels; determining the effective number of twists, the effective path length between detection stations, the image axial length conversion parameter, and the twist pitch transfer parameter; generating a basic record for online calibration of the steel cable; based on the basic record for online calibration of the steel cable, sequentially forming a twist pitch maintenance calibration segment with proportionally varying traction speed and twisting component rotation speed, and a twist pitch variation calibration segment with non-proportionally varying traction speed; using segment boundary markers to map the multi-source sensor data collected at each detection station to the same physical segment of the steel cable, generating a steel cable online calibration segment. The system collects records; it calculates the cable transmission time and detection channel acquisition delay by collecting records from online calibration segments of the steel cable; it verifies the traction speed and twisting component rotation speed detection data using the average transmission speed of boundary markers and the mechanical reference position pulse; and it generates candidate calibration parameters for each detection channel by combining the image twist measurement results of the two types of calibration segments. It selects calibration segments that did not participate in the calculation of candidate calibration parameters and production status recovery sections to verify the candidate calibration parameters for each detection channel. When the independent reference error, twist maintenance relationship error, and twist change relationship error meet the corresponding allowable ranges, it generates candidate calibration parameter verification records and parameter version generation condition records.

[0008] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the determination of the effective twisting count, the effective path length between detection stations, the image axial length conversion parameter, and the twist pitch transmission parameter, and the generation of the basic record for online calibration of the steel cable, includes: reading the structural data and transmission relationship data of the twisting component to determine the effective twisting count per revolution of the twisting component; determining the effective path length between detection stations along the actual operating centerline of the steel cable; determining the image axial length conversion parameter based on the physical length and pixel length of the image calibration component; and determining the twist pitch transmission parameter based on the traction speed, the rotation speed of the twisting component, the effective twisting count, and the offline twist pitch measurement value of the corresponding steel cable section.

[0009] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the steps of determining the effective twisting times, the effective path length between detection stations, the image axial length conversion parameter, and the twist transfer parameter to generate the basic record for online steel cable calibration include: saving the twist transfer calibration data according to the steel cable type, rope mold specification, and steel cable tension range; determining the twist transfer ratio parameter and the twist transfer offset parameter when there are at least three traction speeds to calculate the twist and the corresponding offline twist measurement value; and saving the correspondence between the twist calculated by the traction speed and the offline twist measurement value when the fitting result does not meet the twist measurement error requirements.

[0010] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the step of sequentially forming a pitch maintenance calibration segment with proportional changes in traction speed and twisting component speed, and a pitch change calibration segment with non-proportional changes, includes: when the production task, process formula, and equipment operating status meet the conditions for forming the calibration segment, simultaneously adjusting the traction speed and twisting component speed so that the difference between the calculated pitch before and after the adjustment does not exceed the allowable error for pitch maintenance, thus forming a pitch maintenance calibration segment; after restoring the production status before calibration, maintaining the traction speed and adjusting the twisting component speed so that the pitch change reaches the minimum identifiable pitch change in the image and is within the allowable range of the product pitch, thus forming a pitch change calibration segment.

[0011] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the step of mapping the multi-source sensor data collected by each detection station to the same physical steel cable section using section boundary markers includes: after the traction speed control value, twisting component speed control value, steel cable tension, motor current, and frame vibration meet the corresponding stability conditions, forming the start boundary marker and end boundary marker of the calibration segment respectively; and extracting the traction speed, twisting component speed, image twist pitch, and auxiliary verification data located between the same start boundary marker and end boundary marker in each detection station according to the boundary marker number, boundary marker arrival time, and original data sequence number.

[0012] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the calculation of steel cable delivery time and detection channel acquisition delay includes: determining the steel cable delivery time based on the arrival time of the boundary marker of the same section at different detection stations and the effective path length between detection stations; using the steel cable delivery time for the correspondence of steel cable physical sections in different detection station data; determining the detection channel acquisition delay based on the difference between the starting position and the ending position of the detection data change section and the reference change section; and using the detection channel acquisition delay to correct the time offset caused by sensor response, data filtering, data processing, and communication.

[0013] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the method of verifying the traction speed and twisting component rotation speed detection data using the average transmission speed of boundary markers and mechanical reference position pulses includes: calculating the average transmission speed of boundary markers based on the time difference and effective path length of the boundary markers passing through adjacent detection positions in the same section; comparing the average transmission speed of boundary markers with the original detection data of traction speed within the same time range to determine candidate proportional correction parameters for traction speed; calculating the mechanical reference rotation speed based on the time interval between adjacent mechanical reference position pulses; comparing the mechanical reference rotation speed with the original detection data of twisting component rotation speed within the corresponding rotation cycle to determine candidate proportional correction parameters for twisting component rotation speed.

[0014] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the step of generating candidate calibration parameters for each detection channel by combining the image twist measurement results of two types of calibration segments includes: calculating the twist representative value converted from the image measurement position of the twist holding calibration segment and the twist changing calibration segment using the traction speed verified by the average transmission speed of the boundary marker and the twist component rotation speed verified by the mechanical reference position pulse; reading the effective images of the two types of calibration segments, completing the conversion between pixel distance and entity distance, and taking the median of the effective image twist measurement values ​​as the original representative value of the image twist; and determining the image twist candidate ratio correction parameter and the image twist candidate offset correction parameter based on the image measurement position converted representative value and the original representative value of the image twist of the two types of calibration segments.

[0015] As a preferred embodiment of the online calibration method for multi-source sensor data of the steel cable production line described in this invention, the following steps are taken: Verifying the candidate calibration parameters for each detection channel, and generating a candidate calibration parameter verification record and a parameter version generation condition record when the independent reference error, twist maintenance relationship error, and twist change relationship error all meet the corresponding allowable ranges, includes: selecting a complete calibration segment whose formation time is later than the candidate calibration parameters and which did not participate in the calculation of the candidate calibration parameters; first verifying the candidate acquisition delay of each detection channel; then using the verified candidate acquisition delay to verify the corresponding candidate proportional correction parameter; using the production state recovery section to verify the stable correspondence after parameter application; sequentially applying the verified traction speed, twisting component speed, and image twist candidate parameters to the same set of original data; and generating a parameter version generation condition record when the independent reference error of traction speed, the independent reference error of twisting component speed, the twist maintenance relationship error, and the twist change relationship error are all within the corresponding allowable ranges.

[0016] Secondly, this invention provides an online calibration system for multi-source sensor data of a steel cable production line, comprising: a basic parameter module for acquiring data on steel cable production tasks, processes, equipment, and testing channels; determining the effective number of twists, the effective path length between testing stations, image axial length conversion parameters, and twist pitch transfer parameters; and generating a basic record for online calibration of the steel cable; a segment acquisition module for forming twist pitch maintenance calibration segments, production state recovery segments, and twist pitch change calibration segments based on the basic record for online calibration of the steel cable; forming segment boundary markers; and mapping the traction speed, twisting component speed, image twist pitch, and auxiliary verification data acquired from each testing station to the same physical segment of the steel cable, generating a record for online calibration segment acquisition of the steel cable; and a calibration calculation module for calculating data based on... The online calibration segment acquisition records for steel cables calculate the cable delivery time and the acquisition delay of the detection channels, verify the traction speed detection data using the average transmission speed of the boundary markers, verify the twisting component speed detection data using the mechanical reference position pulse, and generate candidate calibration parameters for each detection channel by combining the image twist measurement results of the two types of calibration segments. The parameter verification module is used to select complete calibration segments and production state recovery sections that did not participate in the candidate calibration parameter calculation, and sequentially verify the candidate acquisition delay, candidate ratio correction parameters, and image twist candidate offset correction parameters for each detection channel. When the independent reference error, twist maintenance relationship error, and twist change relationship error meet the corresponding allowable ranges, the module generates candidate calibration parameter verification records and parameter version generation condition records.

[0017] The beneficial effects of this invention are as follows: By forming a pitch-maintaining calibration segment and a pitch-change calibration segment, and using segment boundary markers to map the multi-source sensor data of each workstation to the same steel cable physical segment, position matching of cross-workstation data is achieved, which improves the accuracy of multi-source sensor data correspondence; by calculating the steel cable transmission time and detection channel acquisition delay respectively, verifying each detection channel using the average transmission speed of boundary markers, mechanical reference position pulses and image pitch measurement results, and verifying candidate calibration parameters using independent segments, the detection channel deviation is confirmed, which improves the reliability of online calibration results. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart for an online calibration method for multi-source sensor data in a steel cable production line.

[0020] Figure 2This is a schematic diagram of an online calibration system for multi-source sensor data in a steel cable production line.

[0021] Figure 3 Flowchart for online calibration segment acquisition of steel cables.

[0022] Figure 4 A flowchart for generating and verifying candidate calibration parameters for steel cables.

[0023] Figure 5 A comparison chart showing the boundary position errors of steel cable physical sections under different disturbance scenarios.

[0024] Figure 6 A comparison chart of normalized errors before and after the candidate calibration parameters are applied together. Detailed Implementation

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0027] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0028] Reference Figures 1-6 As an embodiment of the present invention, this embodiment provides an online calibration method for multi-source sensor data of a steel cable production line, comprising the following steps:

[0029] S1. Obtain data on steel cable production tasks, processes, equipment, and testing channels; determine the effective number of twists, effective path length between testing stations, image axial length conversion parameters, and twist transfer parameters; and generate basic records for online steel cable calibration.

[0030] Acquire data on steel cable production tasks, structure, process, equipment, production control channels, calibration and testing channels, auxiliary verification channels, testing station installation, twist transfer calibration, and historical calibration data. Establish the correspondence between production tasks, equipment, testing channels, and testing stations. Determine the effective twist count, effective path length, image axial length conversion parameters, control parameters for the two types of calibration segments, parameters for the production state recovery section, and calibration judgment error. Generate basic records for online steel cable calibration.

[0031] Production task data includes task number, production batch, cable type, nominal diameter, core type, number of strands, cable lay direction, design lay length, allowable lay length range, take-up reel number, and task execution status. Cable structure data includes strand structure, number and arrangement of outer strands, spool specifications, exit location, cable axis direction, and the object for image lay length measurement. The image lay length measurement object is selected from the same outer strand boundary or inter-strand groove trajectory that can be continuously identified along the cable axis. The repeated positions after one complete arrangement cycle are extracted, and the axial physical distance between the repeated positions is used as the image lay length measurement value. When the same measurement object cannot be continuously tracked, the distance between adjacent strand boundaries or different grooves is not used to form a complete lay length.

[0032] Production process data includes allowable ranges for traction speed, twisting component speed, cable tension, design lay length, traction speed and twisting component speed, allowable rate of change, control quantity stability, allowable tension difference during calibration segments, allowable deviation for production status recovery, and production states where calibration segments are prohibited. Production states where calibration segments are prohibited include production line startup, shutdown, routine speed increases / decreases, product switching, rope die replacement, take-up reel replacement, splice handling, wire breakage handling, equipment maintenance, and equipment alarms. Each process range reads the corresponding production process card, equipment control parameter table, and product inspection procedure for the current task and records the document version.

[0033] The equipment data includes the serial numbers and operating status of the traction device, twisting component, rope-binding mold, tension adjustment device, take-up device, boundary marking device, marking detector, image twist measurement device, mechanical reference position switch, and edge processing unit. The system reads the twisting component's structural diagram, transmission ratio, and equipment debugging records to determine the effective twist count per revolution. If the twisting component completes one full twist per revolution, the effective twist count is recorded as one; if it completes two or more full twists, the corresponding count is recorded according to the equipment structure and transmission relationship. The effective twist count is linked to the twisting component's serial number, structural version, and transmission parameter version. If the effective twist count cannot be determined, an incomplete record of online calibration basic parameters is generated, the traction speed is not calculated, the twist is not calculated, and no two types of calibration segments are generated.

[0034] The traction speed and lay pitch are calculated based on the axial traction speed of the steel cable, the rotation speed of the twisting component, and the effective number of twists.

[0035] ;

[0036] in, Calculate the pitch for the traction speed; This refers to the axial traction speed of the steel cable; The number of revolutions per unit time for the twisting component; The effective number of twists per revolution of the twisting component. and Use corresponding time units, and and All are greater than zero.

[0037] The production control channel data includes traction drive speed feedback, twist drive speed feedback, two control target values, control command transmission time, completion mark, and control allowable deviation. The calibration and testing channel data includes traction speed, twist component speed, and image twist measurement data, and records the sensor number, acquisition unit number, sampling period, resolution, calibration allowable error, filtering time, communication delay limit, timestamp source, and current calibration parameters.

[0038] When the production control channel and the calibration detection channel read the same physical sensor signal, a channel homology record is generated, and the corresponding detection channel is not individually verified using production control feedback. The traction speed detection channel is now verified by the actual transmission speed formed by boundary markers, and the twisting component speed detection channel is now verified by the mechanical reference position switch and the number of pulses per encoder revolution. When an independent reference signal is unavailable, the corresponding detection channel is written into a channel set where parameter updates are not allowed.

[0039] The auxiliary verification channel data includes cable tension, current of both motors, frame vibration, and take-up device operation data. It is only used to determine whether a calibration segment can be formed in the production section. If the cable tension exceeds the product's allowable range, the motor current experiences a sudden load change, the frame vibration exceeds the stable operating range, or the take-up device performs additional speed compensation, a record prohibiting the formation of a calibration segment will be generated.

[0040] The installation data for the inspection station includes the rope mold exit, boundary mark formation, mark detection, image lay length measurement, tension measurement, and traction meter position. The effective path length between adjacent inspection positions is measured along the actual operating centerline of the steel cable. The effective path length includes the straight running section and the envelope sections on the guide wheel, tension wheel, and traction wheel. The effective path length is determined through installation measurement, known length calibration rope testing, or calibration using a metering device, and the measurement error and version are recorded. If the path structure changes or the version is inconsistent with the current production line structure, the original path length data is discontinued.

[0041] The marker detector and the sensor at the same workstation are connected to the same data acquisition unit, recording the marker arrival time and the sensor's original data sequence number. When different workstations use different data acquisition units, all acquisition units are connected to a unified clock. If the maximum synchronization error exceeds the allowable error corresponding to cross-workstation data, the acquisition delay is not calculated using the timestamps of different acquisition units.

[0042] The image twist measurement device includes an industrial camera, lens, light source, trigger unit, and image processing unit. A calibration ruler or a calibration piece of known length is placed on the cable axial measurement plane. The axial pixel length of the calibration piece is read from the start, middle, and end regions of the effective field of view. The image axial length conversion parameter for each region is obtained by dividing the physical length of the calibration piece by the corresponding pixel length. The lens distortion correction residual and image edge extraction repetition error are converted from pixel units to length units, added to the calibration piece length error, and then divided by the smallest calibration piece pixel length in the three regions to obtain the camera calibration allowable error. If the difference between the maximum and minimum values ​​of the conversion parameters for each region does not exceed the camera calibration allowable error, the median of the conversion parameters for each region is taken as the image axial length conversion parameter; otherwise, a calibration failure record is generated.

[0043] The lens distortion correction residual and the error from repeated extraction of axial pixel length of the calibration component are converted into a solid length error and added to the length error of the calibration component. The allowable error for camera calibration is expressed as:

[0044] ;

[0045] in, The camera is calibrated to allowable error, in millimeters per pixel; The upper limit of the length error given in the calibration or verification certificate for the calibration part, in millimeters; The maximum residual between the calibration point correction position and the corresponding theoretical position along the axial direction of the steel cable after lens distortion correction is completed, in pixels; The maximum difference in axial pixel length of the calibration component is obtained by repeatedly extracting the calibration component while keeping the positions of the industrial camera, lens, light source, and calibration component unchanged. The unit is pixels. This is the maximum value among the three image axial length conversion parameters: the start region, the middle region, and the end region of the effective field of view, expressed in millimeters per pixel. This is the minimum axial pixel length of the calibration element in the starting, middle, and ending regions of the effective field of view of the image, expressed in pixels.

[0046] The image axial length conversion parameters for the start, middle, and end regions of the effective field of view are obtained by dividing the physical length of the calibration component by the axial pixel length of the calibration component in the corresponding region. The difference between the maximum and minimum values ​​of the image axial length conversion parameters for the start, middle, and end regions of the effective field of view is compared:

[0047] ;

[0048] in, This is the minimum value among the image axial length conversion parameters for the start, middle, and end regions of the effective field of view. If satisfied, the median of the image axial length conversion parameters for the start, middle, and end regions of the effective field of view is taken as the image axial length conversion parameter; otherwise, a calibration failure record is generated.

[0049] During production line installation and commissioning, replacement of the image-based pitch measurement device, replacement of the assemblies, or the first production of a particular cable model, pitch transfer calibration records are obtained. Cable sections with stable production conditions and passing inspection are selected, and the traction speed, twisting component speed, effective twisting counts, and cable tension are recorded. The traction speed is calculated to determine the pitch. Image-based pitch measurement values ​​are repeatedly collected for the same cable section at the image-based pitch measurement location. After winding, the offline pitch measurement value for the corresponding section is measured using a calibrated offline pitch measurement instrument. The pitch calculated by the traction speed and the offline pitch measurement values ​​are used to generate pitch transfer ratio parameters, pitch transfer offset parameters, or a corresponding table. The image-based pitch measurement values ​​are used to calculate the repeated measurement error of the image-based pitch and are compared with the offline pitch measurement values ​​to verify the image axial length conversion results. They are not used in fitting the pitch transfer ratio parameters and pitch transfer offset parameters. Three types of pitch data, section numbers, and their respective uses are stored according to cable model, assemblies specification, and tension range.

[0050] When there are at least three independent calibration sections with different calculated lay lengths for the same cable type, rope mold specifications, and tension range, and each calculated lay length for the traction speed has a corresponding offline lay length measurement value, the lay length transfer ratio parameter and lay length transfer bias parameter are calculated using the weighted least squares method. The maximum absolute difference between the offline lay length measurement value and the fitted calculation value for each calibration section is used as the linear fitting residual. If the linear fitting residual does not exceed the upper limit of the absolute value of the lay length error for the traction speed calculation and the offline lay length measurement error, the lay length transfer ratio parameter, lay length transfer bias parameter, and effective interval are saved. The effective interval is taken as the minimum to maximum calculated lay length for the traction speed of each calibration section; extrapolation outside the interval is prohibited. Calibration sections with equipment alarms, interrupted raw data, or offline repeated measurement differences exceeding the allowable range are excluded, and calibration points are not deleted separately based on the fitted residual. A corresponding table is saved when the linear fitting residual exceeds the allowable range. When there are only two different calibrated pitches, the corresponding records of the two points are saved, and an independent calibration segment that was not involved in the formation of the record and whose pitch for traction speed calculation lies between the two points is selected for verification. When the difference between the interpolation result and the offline pitch measurement value does not exceed the sum of the upper limit of error, and the pitch change direction is consistent, the pitch transfer parameters for the corresponding interval are formed. The image pitch repeatability measurement error is only used for the validity judgment of the image pitch measurement channel and the determination of the image pitch judgment error, and does not participate in the fitting residual judgment of the pitch transfer relationship.

[0051] Historical calibration records are read and categorized by cable type, rope mold specification, traction speed, twisting component speed, and cable tension range. Records with complete production tasks, no equipment alarms, continuous raw data, complete boundary marker identification, and qualified product inspection are selected. Each category has at least ten records, covering at least three independent production cycles, with each production cycle containing at least two records. The following parameters are calculated: stable fluctuation range of traction speed and twisting component speed, image twist repeatability measurement error, stable fluctuation range of cable tension, repeatability time difference of marker detection, pulse repeatability error of mechanical reference position, and historical variation range of calibration parameters. If the required quantity or cycle is not met, corresponding parameters are formed using sensor calibration allowable error, equipment control resolution, and process allowable deviation, and the source is recorded. The control quantity stability allowable deviation is determined based on equipment control resolution and process allowable deviation, serving as the maximum allowable range for forming a calibration segment. The historical stability fluctuation range reflects the actual fluctuation level within the corresponding category. When forming a calibration segment, the smaller of the two values ​​is taken as the stability judgment range. If the historical stability fluctuation range exceeds the control quantity stability allowable deviation, the control quantity stability allowable deviation is used, and a historical stable state record to be reviewed is generated.

[0052] A calibration segment is generated based on current production data to form a permission record. The task number and process formula version remain unchanged. The production line is not in a state where calibration segment generation is prohibited. The traction speed control value, twisting component speed control value, and cable tension are within the corresponding process range, and the difference between the maximum and minimum values ​​within the stable observation time does not exceed the corresponding stable fluctuation range. The record is then written as permitted. The stable observation time is the maximum value among the traction device control response time, twisting component control response time, and tension adjustment response time. The sampling quantity for each channel is calculated by dividing the stable observation time by the sampling period and rounding up. If the sampling quantity is insufficient, the observation time is extended or insufficient data is written.

[0053] Determine the control parameters for maintaining the twist pitch in the calibration segment. Read the traction speed control value before the calibration segment is formed. and the speed control value of the twisting component Set the traction speed adjustment amount And calculate the adjustment amount of the twisting component speed:

[0054] ;

[0055] in, Adjust the rotational speed of the twisting component in the calibration segment to maintain the twist pitch; To calibrate the rotation speed control value of the twisting component before segment formation; This is the adjustment amount for traction speed; To calibrate the traction speed control value before segment formation, and Greater than zero.

[0056] According to the resolution of the twisting component speed control Take the values ​​and recalculate the adjusted traction speed and twist pitch. If the difference in twist pitch before and after adjustment does not exceed the allowable error for twist pitch maintenance, the adjustment of traction speed is not less than the minimum identifiable change in traction speed, the adjustment of twisting component speed is not less than the minimum identifiable change in speed, and the adjusted speed, rotational speed, tension, and twist pitch are all within the allowable range of the process, retain the corresponding combination. The two minimum identifiable changes are taken as the maximum values ​​of the corresponding sensor resolution, historical stable fluctuation, and control resolution, respectively.

[0057] When multiple combinations exist, they are first sorted by the difference between the two relative adjustment values ​​from smallest to largest. If the differences are the same, they are then sorted by the larger of the two relative adjustment values, and the combination that appears first in the sorted list is selected. If no combination meets the conditions, no twist pitch maintenance calibration segment record is generated for the current production task.

[0058] Determine the control parameters for the twist pitch variation calibration segment. Maintain the traction speed control value as follows: The rotation speed adjustment amounts for the positive and negative twisting components are set separately. And calculate the traction speed of the calibration segment to calculate the twist pitch:

[0059] ;

[0060] in, Calculate the twist pitch for the traction speed of the calibration segment to adjust the twist pitch variation; To calibrate the rotation speed control value of the twisting component before segment formation; This refers to the adjustment amount of the twisting component's rotation speed; For the effective number of twists, and Greater than zero.

[0061] Will The twist pitch is converted to the image measurement position. If the difference between this value and the image measurement position-converted twist pitch before the calibration segment is formed is not less than the minimum identifiable change in image twist pitch, and is within the allowable range of the product twist pitch, the corresponding twisting component speed adjustment is retained. The minimum identifiable change in image twist pitch is the maximum value between the image axial length conversion resolution and the image twist pitch repeatability measurement error. When multiple adjustment values ​​meet the conditions, they are sorted in ascending order of absolute value of the twisting component speed adjustment, and the adjustment value with the smallest absolute value is selected. If neither positive nor negative adjustments meet the conditions, a twist pitch change calibration segment record is generated for the current production task.

[0062] The lengths of the two types of calibration segments are determined based on the effective field of view length of the image, the acquisition cycle, the cable traction speed, the control stabilization time, and the effective path length from the boundary marker formation position to the image measurement position. The exclusion length inside the boundary marker is determined by the sum of the marker width, the effective field of view length of the image, the converted length of the marker detection error, and the movement length corresponding to the camera trigger delay. The length of the production state recovery section is calculated based on the time required for speed, rotational speed, and tension stabilization, along with the current traction speed.

[0063] The permissible error for traction speed calibration is determined by summing the speed error converted from path length measurement error, the speed error converted from marker detection time error, and the upper limit of the absolute value of the permissible error for traction speed sensor calibration. The permissible error for twisting component speed calibration is determined by converting the mechanical reference position detection error and encoder pulse count error into speed errors based on the effective measurement duration of the calibration segment, and then adding this to the upper limit of the absolute value of the permissible error for speed sensor calibration. The image twist judgment error is determined by summing the upper limit of the absolute values ​​of the image axial length calibration error, the image twist repeat measurement error, and the offline twist measurement error. The single change in the candidate proportional correction parameter must not exceed the correction amount converted from the corresponding sensor calibration permissible error, and must not exceed the upper limit of the absolute value of parameter changes in historical qualified calibration records.

[0064] Write the production task number, cable structure parameters, production process range, equipment number, channel correspondence, effective twist count, effective path length, unified clock parameters, image axial length conversion parameters, twist transfer calibration parameters, calibration segment formation permission conditions, two types of calibration segment control parameters, calibration segment length, production status recovery section length, calibration judgment error, current calibration parameter version and parameter source into the basic record of online cable calibration.

[0065] S2. Based on the basic records of online calibration of steel cables, calibration segments with proportional changes in traction speed and twisting component speed and non-proportional changes in twist pitch are formed in sequence. Multi-source sensor data collected by each detection station are mapped to the same physical section of the steel cable using section boundary markings, and online calibration segment collection records of steel cables are generated.

[0066] Read the basic records of online cable calibration. When the production task meets the formation conditions, adjust the traction speed and twisting component speed according to the control parameters of the calibration segment with the twist pitch maintenance and the calibration segment with the twist pitch change. After the production control value, cable tension, motor current and frame vibration meet the corresponding stability conditions, form the section boundary mark. Collect the detection data of traction speed and twisting component speed during cable formation, as well as the image twist pitch and boundary mark data formed when the same cable section passes through the image measurement position and the mark detection position. Establish the relationship between each detection data and the actual cable section according to the boundary mark number and the original data sequence number, and generate the online calibration segment acquisition record of the cable.

[0067] The processor reads the online calibration baseline record of the steel cable according to the production task number, and compares the current production batch, steel cable model, rope-binding mold number, twisting component number, traction device number, process formula version, effective path length version, image calibration version, lay length transfer calibration version, and current calibration parameter version. If any number or version is inconsistent, a calibration baseline record mismatch mark is generated, and no control command is sent; if all items are consistent, the processor reads the control parameters of the two types of calibration segments, relevant time parameters, segment length, minimum effective length, boundary mark configuration, and sampling parameters of each detection channel.

[0068] The system continuously collects data on production task status, control and detection values ​​of traction speed and twisting component speed, cable tension, motor current, frame vibration, take-up device operating status, and equipment alarm status, and writes the collection time and original data sequence number according to the sampling cycle. If the production task number and process formula version remain unchanged, and the production line has not undergone start-up / stop, routine speed increase / decrease, product switching, mold change, reel change, joint handling, wire breakage handling, or maintenance, and the traction speed control value, twisting component speed control value, and cable tension are within the process allowable range, and the difference between the maximum and minimum values ​​within the stable observation time before calibration does not exceed the corresponding stable fluctuation range, a calibration segment is generated to form an authorized record. If there is a sudden change in motor current load, frame vibration exceeding limits, the take-up device performing additional speed compensation, equipment alarms, or discontinuous necessary channel data sequence numbers, a prohibited formation status is written.

[0069] When the calibration segment formation permission record is written, a twist holding control command is sent to the production line controller. The control command includes the calibration cycle number, target traction speed, target twisting component speed, allowable control deviation, allowable rate of change of the control quantity, and maximum time required to reach the control target. The production line controller adjusts the traction device and twisting component according to the allowable rate of change and maintains the current process target value of the tension regulating device. The processor records the command transmission time, the start response time of both drives, the time when both control values ​​enter the target allowable range, and the command execution completion marker.

[0070] Once both control values ​​enter the target allowable range, the traction speed and pitch are calculated based on the feedback value from the production control channel.

[0071] ;

[0072] in, Calculate the pitch for the traction speed corresponding to the production control feedback; This is the feedback value for traction speed control; This is the feedback value for controlling the rotational speed of the twisting component; The effective number of twists per revolution of the twisting component, and and All are greater than zero.

[0073] Compare the calculated pitch at the current traction speed with the calculated pitch at the traction speed before the start of the calibration cycle. If the difference between the two does not exceed the allowable error for pitch maintenance, and both control values ​​do not fall outside the target allowable range within the control target maintenance confirmation time, the cable tension is within the product process allowable tension range, and the motor current and frame vibration do not exceed the stable operating range, then the pitch maintenance control is considered stable. If the conditions are still not met after the maximum time limit for reaching the control target, restore the control target value before the start of the calibration cycle, generate a pitch maintenance control failure record, and end the current calibration cycle.

[0074] After the twist pitch control stabilizes, the boundary marking device corresponding to the current production line in the basic record is invoked to form the start boundary mark of the twist pitch holding segment at a position where the rope has been joined and the marking material will not be erased by the guide component. According to the marking type in the basic record, the calibration cycle number, segment type, and boundary type are indicated by color changes, reflection intensity changes, or coded patterns. Boundary types include effective segment start, effective segment end, and invalid segment termination. If the marking device does not return a completion signal, the remaining marking material is insufficient, or the formation position is abnormal, the original control target value is restored, and a boundary mark formation failure record is generated.

[0075] After the initial boundary marker is formed, two control target values ​​are maintained. The calibration segment length is preferentially determined by the cumulative length of the production length encoder, which is independent of the traction speed detection channel to be calibrated and is within its calibration validity period. If the production length encoder is unavailable and the production control channel is independent of the detection channel to be calibrated, the segment length is formed by integrating the speed feedback inside the traction drive over time. If neither type of reference is available, the segment length is calculated based on the traction speed control target value and the holding time, and the source of the length is recorded. The traction speed detection value to be calibrated is only used as synchronously acquired data.

[0076] When the section length reaches the length of the pitch maintenance calibration segment, a valid section end boundary mark is formed, and the actual cable section between the two boundary marks is written into the pitch maintenance calibration segment. During segment formation, if any control value falls outside the target allowable range, the cable tension exceeds the process allowable range, the motor current or frame vibration exceeds the limit, the take-up device performs additional speed compensation, the equipment generates an alarm, or the production task changes, an invalid section stop boundary mark is immediately formed, the original control target value is restored, and the data between the start boundary mark and the stop boundary mark is written into the original data range of the invalid section. When the pitch maintenance calibration segment is invalid, no more pitch change calibration segments are formed.

[0077] After the end boundary marker of the twist pitch holding segment is formed, the two control target values ​​are restored according to the process-allowed rate of change, and data from the first production state recovery section are collected. When the traction speed control value, twisting component speed control value, and cable tension enter the stable range before the start of the calibration cycle and remain within the production state recovery confirmation time, and the auxiliary verification data is normal, the first production state recovery is recorded as complete. If the first production state recovery fails to recover beyond the maximum recovery time, a production state recovery anomaly record is generated, and the current calibration cycle ends.

[0078] After the first production state is restored, maintain the traction speed target value as the process value before the start of the calibration cycle. Increase or decrease the twisting component speed target value according to the adjustment amount in the basic record, and send the twist pitch change control command. Keep the traction speed control value within the original target allowable range. After the twisting component speed control value enters the new target allowable range, calculate the current traction speed and twist pitch. The direction of change of the current calculation result relative to before the start of the calibration cycle is consistent with the basic record, the change amount is not less than the minimum identifiable change amount of the twist pitch in the image, the calculation result is within the product twist pitch allowable range, both control values ​​remain stable within the control target maintenance confirmation time, the cable tension is within the process allowable range, and the motor current and frame vibration do not exceed limits. Write the twist pitch change control as stable. If any condition is not met, restore the original control target value and generate a twist pitch change control failure record.

[0079] After the pitch variation control stabilizes, a starting boundary mark for the pitch variation segment is formed, and the segment length is determined according to the same length source as the pitch maintenance calibration segment. When the segment length reaches the length of the pitch variation calibration segment, a valid segment ending boundary mark is formed, and the actual cable segment between the two boundary marks is written into the pitch variation calibration segment. If any abnormalities occur in the control value, tension, auxiliary verification data, or equipment alarm during segment formation, an invalid segment termination boundary mark is formed and written into the original data range of the invalid segment.

[0080] After the boundary marker for the twist pitch change segment is formed, the target value of the twisting component speed is restored, and data for the second production state recovery segment is collected. When the three production data enter the stable range before the start of the calibration cycle and continue until the production state recovery confirmation time is reached, the second production state recovery is recorded as complete. The original data sequence range of this recovery segment is written into the online calibration segment acquisition record for steel cable, for reading during candidate calibration parameter verification. When the production task ends between two types of calibration segments, or when a reel change or process formula change occurs, the current calibration cycle ends, and an incomplete calibration segment record is generated.

[0081] Each boundary marker passes through the marker detection position sequentially along with the steel cable. The marker detector extracts identification features according to the marker type in the basic record, recording the boundary marker number, detector number, marker arrival time, original data sequence number, and production task number. When the interval between consecutive identification signals from the same detector is less than the marker signal merging time, they are merged, and the midpoint between the earliest entry time and the latest exit time is taken as the marker arrival time. If the duration of the merged signal exceeds the allowable duration calculated from the marker axial width and the current traction speed, a boundary marker tail record is generated, and this marker is not used to determine the section boundary.

[0082] If the detection time is earlier than the marker formation time, the arrival times of multiple detection locations do not conform to the cable operation sequence, the boundary type arrangement is abnormal, or an invalid section termination marker is identified as a normal end marker, a record of abnormal time sequence or boundary type will be generated. If the same boundary marker is missed at a necessary detection location, only the original data of the confirmed station will be retained. Missing boundaries will not be supplemented based on the expected arrival time, nor will cross-station deviation source judgment be performed.

[0083] Calculate the average transmission speed of the boundary markers based on the arrival time of the same boundary marker at adjacent detection locations and the effective path length between the two detection locations:

[0084] ;

[0085] in, Boundary marking is determined by the detection location. Move to the detection location Average transmission speed during the period; For detection location To the detection location The effective path length measured along the centerline of the steel cable; Boundary markers reach the detection location Time; For the same boundary marker to reach the detection position The time, and Greater than .

[0086] to If changes occur during the process, such as changes in control objectives, equipment alarms, wire take-up speed compensation, or changes in production tasks, the average transfer speed will be marked as unusable for traction speed calibration. If the uniform clock synchronization error or the mark detection repetition time difference does not meet the basic recording requirements, the average transfer speed will not be calculated. If the calculated result exceeds the allowable range of traction speed, the mark number, path length version, and detection time will be verified; if inconsistencies exist, an abnormal transfer speed data record will be generated.

[0087] Based on the start and end boundary markers identified at the necessary detection stations, the range of local sensor raw data sequence numbers between the two detections is extracted. The twisting equipment acquisition unit extracts traction speed detection data, twisting component speed detection data, encoder pulses, and mechanical reference position pulses; the image twist measurement station extracts the corresponding image sequence number and image data; the auxiliary verification channel extracts cable tension, motor current, frame vibration, and take-up device operation data within the same formation time interval. When data sequence numbers are discontinuous, production task numbers change, or clock synchronization is abnormal, the corresponding data is marked as cable physical sections that cannot be confirmed.

[0088] The image-based twist measurement device continuously acquires images between the start and end boundary marks. Lens distortion correction and grayscale normalization are performed on each image. Local threshold segmentation and morphological closing operations are used to extract the main cable region, retaining the largest connected region whose area and width conform to the cable's outer diameter range. The midpoints of the upper and lower boundaries of the main region are calculated column by column. After removing outliers that deviate from the midpoint, the cable axis is fitted using the least squares method. A measurement area is established along the cable axis, and candidate trajectories for the outer strand boundaries and interstrand grooves are extracted using grayscale gradients and edge continuity. Adjacent images are correlated based on the trajectory center position, tangent direction, and grayscale profile correlation coefficient. The maximum allowable displacement is determined based on the cable's maximum operating speed, the time interval between adjacent image acquisitions, and the image axial length conversion parameter. When multiple candidate trajectories exist, those exceeding the maximum displacement or direction difference range are first excluded, and then the trajectory with the longest continuous length and the smallest positional residual between adjacent images is selected. Axial position sequences are extracted along the associated trajectory, and normalized correlation coefficients of grayscale profiles at different positions are calculated. Positions where the correlation coefficient reaches the repetition position judgment value and the spacing is not less than the minimum identifiable change in image twist pitch are determined as repetition positions. When the spacing between two consecutive adjacent repetition positions meets the consistency range, a complete arrangement cycle is determined, and the median of the lengths of multiple complete arrangement cycles is used as the image twist pitch measurement value. The main body segmentation parameters, trajectory association direction difference, grayscale profile correlation coefficient judgment value, and arrangement cycle consistency range are determined through calibration sample images and historical qualified images, and written into the basic record of online steel cable calibration.

[0089] When a single image can cover the complete twist pitch, two repeated positions of the same measurement object after one complete arrangement cycle are extracted, and the axial pixel distance is converted into the physical distance. When a single image cannot cover the complete twist pitch, the continuous images are converted to a unified cable axial coordinate system based on the acquisition time of adjacent images, the cable movement length formed by independent length references, and the overlapping image area, and then two repeated positions are extracted. For each image twist pitch measurement, the first frame time, last frame time, window center time, window start and end axial positions, and representative axial position are recorded. The result timestamp is taken from the window center time, and the representative axial position is taken from the midpoint of the axial coordinates of the two repeated positions. The measurement window is formed by subtracting the length inside the boundary markers between the start and end boundary markers, and images that cross the valid boundary are not used. When image results are sparse, results completely within the calibration segment are selected according to the measurement window coverage and representative axial position, and incomplete windows are not used to infer the boundary of the change section. When no valid results exist, no valid result record is generated for the twist pitch measurement. When calculating the acquisition delay of the image measurement channel, the window center time and the representative axial position are used as the corresponding references, and the step-by-step sampling point boundary displacement calculation method of the traction speed and twisting component rotation speed detection channel is not directly adopted.

[0090] Images are marked as invalid if the cable body boundary is incomplete, the continuous trajectory length of the measured object is insufficient, the motion blur width exceeds the motion blur judgment value, the change in the cable axis position exceeds the axis position change judgment value, or the coverage ratio of oil and reflective areas exceeds the coverage ratio judgment value. Each judgment value is determined through calibration sample testing. Keeping the industrial camera, lens, light source, exposure time, and installation position constant, the cable running speed, cable axis position, and oil and reflective coverage area are changed, and images are repeatedly acquired and the image twist length is calculated. The maximum motion blur width, maximum axis position change, and maximum oil and reflective area coverage ratio corresponding to the conditions where the cable body boundary can be continuously extracted, the continuous trajectory length of the measured object meets the twist length calculation requirements, and the repeated measurement error of the image twist length does not exceed the image twist length judgment error are determined as the corresponding judgment values. The continuous trajectory length judgment value is determined based on the product twist length upper limit and the complete trajectory length required to complete one twist length measurement. Each judgment value, test data, and camera configuration are written into the cable online calibration basic record.

[0091] Images are selected based on the angle between the cable axis and the image's axial centerline. The preset observation angle range is determined by repeatedly acquiring data from calibrated samples at different tilt angles. The maximum positive or negative angle corresponding to the condition where the trajectory can be continuously extracted and the repeated measurement error of the image twist does not exceed the image twist judgment error is taken, with a value range of ±5 to ±20 degrees. If the difference between the current measurement value and both the previous and next valid measurement values ​​exceeds the image twist judgment error, but the difference between the previous and next valid measurement values ​​does not exceed this error, the current measurement value is marked as an isolated value. This rule is not applied to the first and last measurement values. The median of the remaining valid measurement values ​​is taken as the representative image twist value of the calibration segment. The minimum number of valid images is determined based on the number of historical qualified images where the median tends to stabilize with the increase of the number of images, ranging from five to twenty images. The minimum length of the valid calibration segment is determined based on the product's twist upper limit, the measurement window width, and the exclusion length inside the boundary marker, ensuring that after deducting the exclusion lengths at both ends, it can cover three to six complete arrangement cycles. If the number of valid images is less than the minimum number of valid images in the basic record of online calibration of steel cable, or if the length of the calibration segment is less than the minimum valid calibration segment length, the corresponding segment will be marked as insufficient image data.

[0092] Calculate the representative cable tension values ​​for the two types of calibration segments. The representative tension value is the median of the effective tension samples within the segment. If the difference between the two types of representative tension values ​​exceeds the allowable tension difference for the calibration segment, an inconsistency flag is added to the segment tension condition, and the two types of segments are not used to calculate the image twist ratio correction parameter. If the physical length of any segment at the necessary inspection station is less than the minimum effective calibration segment length, an insufficient segment length flag is added.

[0093] The system summarizes the control commands, production control channel data, boundary marker records, average transmission speed, raw data of the test channel to be calibrated, image twist measurement values, mechanical reference position pulses, auxiliary verification data, raw data sequence range, length source, data integrity status, and segment validity status of the two types of calibration segments and two production status recovery sections to generate a steel cable online calibration segment acquisition record.

[0094] S3. Through the online calibration segment acquisition and recording of the steel cable, the steel cable transmission time and the acquisition delay of the detection channel are calculated respectively. The average transmission speed of the boundary marker and the mechanical reference position pulse are used to verify the traction speed and the rotation speed of the twisting component. Combined with the image twist measurement results of the two types of calibration segments, candidate calibration parameters for each detection channel are generated.

[0095] Read the basic records of online cable calibration and the acquisition records of online cable calibration segments, verify the correspondence between production tasks, equipment, detection channels, section boundary markers, original data serial numbers and parameter versions, convert the data acquired at different detection locations to the corresponding physical sections of the cable, calculate the acquisition delay and proportional correction parameters of the traction speed detection channel, the twisting component speed detection channel and the image twist measurement channel, and generate cable detection channel calibration judgment records and cable candidate calibration parameter records based on the average transmission speed of boundary markers, mechanical reference position pulses and twist measurement results in the two types of calibration segments.

[0096] The processor reads two types of records according to the production task number and calibration cycle number, and compares the cable type, rope-binding mold number, traction device number, twisting component number, image twist measurement device number, marker detector number, acquisition unit number, process formula version, effective twist count version, effective path length version, image axial length conversion parameter version, twist transfer calibration version, and current calibration parameter version. If any number or version is inconsistent, the original data sequence number is discontinuous, the boundary marker number is missing, or the boundary type order is abnormal, a calibration record associated abnormality marker is generated, and candidate parameters are not calculated.

[0097] When the records are consistent, the boundary markers, effective status, length source, tension representative value, control channel data, and raw data of the detection channel to be calibrated are read from the pitch maintenance calibration segment, the first production state recovery segment, the pitch change calibration segment, and the second production state recovery segment. If the pitch maintenance calibration segment is invalid, the parameter calculation for the current calibration cycle is stopped. If the pitch maintenance calibration segment is valid but the pitch change calibration segment is invalid, insufficient in length, or insufficient in image data, only the independent verification data of traction speed and twisting component speed are retained, and an incomplete record of the image pitch calibration segment is generated.

[0098] The location where the section boundary markers are formed is set as the starting point of the cable's axial position. For sections where the production control objective remains unchanged, the axial position of the cable corresponding to the sampled value is calculated based on the difference between the average transmission speed of the boundary markers and the sampling time. For sections where the production control objective changes, the cumulative length of the independent production length encoder is read. If the production length encoder is unavailable, time integration is performed on the production control speed feedback from a different source than the traction speed detection channel to be calibrated. If both types of data are unavailable, only the position range corresponding to the starting and ending boundaries is recorded. The cable transmission time is used to establish the relationship between workstation sections, and the detection channel acquisition delay is used to correct for time offsets caused by sensor response, filtering, data processing, and communication. The two types of time are stored separately.

[0099] The candidate acquisition delay range is determined according to the timestamp type of the detection channel. When the timestamp records the original sampling time, the candidate range is from zero to the sum of the sensor response time and the filtering time; when the timestamp records the data output time, the candidate range is from zero to the sum of the sensor response time, the filtering time, the data processing time, and the upper limit of the communication delay. The candidate value interval is taken as the corresponding sampling period. When the candidate acquisition delay is positive, the corresponding time of the detection data is shifted forward; the unified clock synchronization error only includes the position comparison allowable error.

[0100] When the traction speed detection channel and the production control speed feedback are independent of each other and use a unified clock, the axial position of the steel cable corresponding to the entry and exit of the calibration target range of the production control speed feedback is used as the reference boundary; when the two channels originate from the same source, the candidate acquisition delay of the traction speed is not calculated. The twisting component speed detection channel uses the mechanical reference speed calculated by the mechanical reference position pulse as a reference, and uses the start point of the first complete rotation cycle that continuously meets the target range and the end point of the last complete rotation cycle that continuously meets the target range as the reference boundary. The image twist measurement channel uses the end point of the exclusion length inside the start boundary of the twist change calibration segment and the start point of the exclusion length inside the end boundary as the reference boundary.

[0101] For each candidate acquisition delay, the detection data is converted to the axial position of the steel cable. The median of the detection data in the stable section before the calibration segment is formed is used as the baseline value. The change judgment value is the larger of the historical stable fluctuation range of the corresponding detection channel and the allowable error of the sensor calibration. When the difference between the detection data and the baseline value continuously reaches the change judgment value, and the continuous length reaches the minimum change confirmation length in the basic record of the online calibration of the steel cable, the first position that meets the condition is determined as the starting position of the detection change segment. When the detection data enters the stable range after the change and continuously reaches the minimum stable confirmation length, the first position that meets the condition is determined as the ending position. An exit judgment value less than the change judgment value is used to reduce repeated crossings caused by reciprocating fluctuations near the threshold. When multiple crossings exist, the segment with the longest continuous length and the largest overlap with the reference change segment is selected, consistent with the control adjustment direction. When multiple candidate segments still exist, the segment with the smallest boundary position difference is selected. When no significant change is formed, only one boundary is formed, or multiple candidate segments cannot be distinguished, the candidate acquisition delay of that detection channel is not calculated, and a record of "detection change segment cannot be determined" is generated. After the complete formation of the detected change section, calculate the boundary position difference:

[0102] ;

[0103] in, For the first Each detection channel uses a candidate acquisition delay. The difference in boundary position at that time; and These are the starting and ending positions of the detected change segment, respectively; and These are the starting and ending positions of the reference change segment, respectively.

[0104] Candidate acquisition delays are sorted in ascending order of boundary position difference, and the candidate value at the top of the list is selected. When multiple candidate values ​​are the same, the candidate value with the smaller difference from the current acquisition delay is selected. If the boundary position difference corresponding to a candidate value is not less than the result corresponding to the current parameter, the current parameter is retained. If the top-ranked candidate value is at the upper limit of the search, a record of insufficient acquisition delay range is generated, and this value is not written into the verifiable parameters.

[0105] When determining the traction speed proportional correction parameters, the screening boundary markers are determined by the detection position. Move to the detection location No changes in control targets, equipment alarms, wire take-up speed compensation, or production task changes were recorded during this period. Raw traction speed detection data within the same time range were read, and after applying a candidate acquisition delay, the sampled values ​​were arranged in ascending order of timestamp. If multiple sampled values ​​exist at the same timestamp, the median was used. If the timestamps could not strictly increase, the number of samples was no greater than one, or the last sampling time was no greater than the first sampling time, the corresponding interval was marked as invalid. When the difference between adjacent sampling times did not exceed the maximum allowable sampling interval, a time-weighted average was calculated based on the duration from the previous sampled value to the next sampling time.

[0106] ;

[0107] in, The time-weighted average of the detection data within the target time range; For the first The detected values ​​at each sampling time; For the first Each sampling time; This is the number of samples after merging duplicate timestamps. The maximum allowable sampling interval is determined based on the rated sampling period of the detection channel, the upper limit of communication jitter, and the historical continuous sampling interval, and is written into the basic record of online steel cable calibration. When the time difference between any two adjacent samples exceeds the maximum allowable sampling interval, the time-weighted average of that interval is not calculated. When the sampling period is fixed, the timestamps are continuous, and there are no missing samples, the arithmetic mean is used.

[0108] When the difference between the average transmission speed of boundary markers and the average traction speed exceeds the allowable error for traction speed calibration, and the direction of the difference is consistent in both types of calibration segments, the effective intervals where the average value of the original traction speed detection data is greater than the minimum effective denominator of the traction speed are selected. The minimum effective denominator of the traction speed is the larger of the lower limit of the effective range of the traction speed detection channel and three times the detection resolution, and is written into the basic record of online cable calibration. When the average value of the original traction speed detection data is zero, negative, or does not exceed the minimum effective denominator of the traction speed, the corresponding interval is not included in the proportional calculation; when the original detection value remains unchanged continuously within the freeze confirmation time, and the average transmission speed of the boundary markers indicates continuous cable movement, the corresponding interval is marked as the detection value frozen and excluded. The proportional results are calculated for the remaining effective intervals, and the median is taken:

[0109] ;

[0110] in, For candidate proportional correction parameters of traction speed; For the first Average transmission speed of boundary markers for each effective interval; The average value of the raw traction speed detection data when candidate acquisition delay is applied but current proportional correction parameters are not applied; This represents the median of the proportion results for each effective interval. The candidate proportion correction parameters for the twisting component speed adopt the same denominator selection rule. The minimum effective denominator for the twisting component speed is the larger value between the lower limit of the effective range of the speed detection channel and the minimum measurable speed obtained by converting the pulse count resolution of the mechanical reference position. If the average speed is zero, negative, does not exceed the minimum effective denominator, or is determined to be a frozen value, it will not be included in the calculation of the twisting component speed proportion result.

[0111] The system reads data on the clamping status of the traction wheel, the surface condition of the traction wheel, and the tension of the steel cable. If there is an abnormal clamping condition, slippage alarm, or the range of speed ratios across different tension ranges exceeds the allowable range for proportional parameter consistency, a traction transmission status change record is generated, but no candidate proportional correction parameter for traction speed is formed. The allowable range for proportional parameter consistency and the upper limit of the proportional result range are determined based on historical qualified calibration records; if historical records are insufficient, they are determined using the allowable error of the traction speed detection channel verification and the error of repeated measurements, respectively. If there are fewer than three valid intervals, the proportional correction directions of the two types of calibration segments are opposite, the proportional result range exceeds the limit, or the change in the candidate proportional correction parameter relative to the current parameter exceeds the allowable change in a single instance, the traction speed detection channel is written into the continued acquisition record.

[0112] The threshold distance to the calibration point for a single-point corresponding record is determined based on the traction speed, twisting component speed, and offline pitch measurement error. The stable image interval should contain at least three consecutive valid measurements and three complete arrangement cycles, with fluctuations not exceeding the image pitch judgment error. The permissible change in the offset parameter per instance is determined based on the variation range of historical qualified parameter versions and the image axial length conversion error. Both the permissible pitch maintenance error and the permissible pitch variation error are based on the image pitch judgment error, plus the corresponding traction speed and twisting component speed conversion errors. All judgment values ​​are uniformly recorded in the cable online calibration basic record.

[0113] When determining the proportional correction parameters for the twisting component's speed, the time interval between adjacent mechanical reference position pulses is read, and the mechanical reference speed is calculated based on one rotation of the twisting component corresponding to one effective pulse cycle. If the reference position switch is repeatedly triggered, pulses are missing, the timing sequence is abnormal, encoder data is interrupted, or speed fluctuations within the cycle exceed the limit, the corresponding cycle is excluded. If the difference between the mechanical reference speed and the average value of the original speed detection data within the same time range exceeds the allowable error for speed calibration, and the direction of the difference in the two types of calibration segments is consistent, the following calculation is performed:

[0114] ;

[0115] in, Correction parameters for candidate ratios of twisting component rotation speed to replace current parameters; For the first The mechanical reference speed for one effective rotational cycle; The average value of the original rotational speed detection data when candidate acquisition delay is applied but current proportional correction parameters are not applied.

[0116] Insufficient number of effective rotation periods, opposite directions of proportional correction for the two types of segments, or excessive range of proportional results. When the change relative to the existing parameters exceeds the allowable change in a single step, no verifiable candidate proportional correction parameter for the rotational speed is generated.

[0117] Candidate acquisition delay and candidate scaling correction parameters are applied to the raw detection data of traction speed and twisting component rotation speed. Channels without candidate parameters use the current parameters. If the raw data conforms to the independent reference relationship, or if the difference between the raw data and the independent reference data is reduced to within the allowable error after applying the candidate parameters, the channel is marked as usable for twist pitch conversion; if any channel still does not conform to the independent reference relationship, the deviation of the image twist pitch measurement channel is not judged.

[0118] According to traction speed Twisting component speed Calculate the traction speed and twist pitch by calculating the effective number of twists:

[0119] ;

[0120] in, Calculate the pitch for the traction speed; The effective number of twists per revolution of the twisting component, and and All are greater than zero.

[0121] Read the lay transfer calibration record corresponding to the current cable type, rope die specification, and tension range. When the record includes lay transfer ratio parameters and lay transfer offset parameters, follow the... Calculate the twist pitch by converting the image measurement position; record the results in a corresponding table. If the value equals a value in the table, the corresponding image measurement position is directly read and the twist pitch is calculated. If the value is between two adjacent values, linear interpolation is performed. If the table contains breakpoints, or the order of adjacent values ​​is abnormal, or... When the coverage area is exceeded, no image candidate parameters are generated. When the record contains only data corresponding to a single point, only the twist pitch maintenance calibration segment closest to that calibration point is verified. When the current tension representative value spans two calibration tension intervals, calibration data from different calibration tension intervals are not mixed.

[0122] Calculate the converted lay length based on the image measurement position for each valid interval of the two types of calibration segments. Arrange the segments according to the axial position of the cable and exclude isolated values. Take the median of the remaining three or more valid values ​​as the representative value of the converted lay length based on the image measurement position. and When there are no valid values ​​or fewer than three remaining valid values, no image candidate parameters are formed. The image twist representative value is read after pixel distance to entity distance conversion but before applying current image correction parameters. and The difference between the representative tension values ​​of the two calibration segments does not exceed the allowable tension difference, the number of valid images meets the requirements, and | | Not less than the minimum identifiable change in image twist pitch, | When | is greater than the allowable error for twist pitch variation, and the two sets of representative values ​​change in the same direction, the candidate twist pitch ratio correction parameter for the image is expressed as:

[0123] ; ;

[0124] in, Correction parameters for candidate twist ratios in the image; These are candidate offset correction parameters for image twist.

[0125] Less than or equal to zero or If the change exceeds the corresponding single allowable change, the image twist exceeds the measurement range after parameter application, or the image twist within a segment continuously changes along the axial direction without forming a stable interval, no verifiable image candidate parameter will be generated. If only one of the twist maintenance relationship or twist change relationship is not met, the image twist measurement channel will be written to the continued acquisition record.

[0126] When the average transmission speed of the boundary markers is inconsistent with the traction speed detection result, but the mechanical reference position pulse verification speed is consistent with the speed detection result, the traction speed detection channel is marked as the channel to be calibrated; when the former is consistent but the latter is inconsistent, the twisting component speed detection channel is marked as the channel to be calibrated. When two basic detection channels meet the independent reference relationship or meet the independent reference relationship after applying candidate parameters, the image twist measurement channel is judged separately: when the twist maintenance relationship error exceeds the allowable twist maintenance error, a candidate offset correction parameter is formed according to the offset correction condition; when the twist change error exceeds the allowable twist change error, a candidate proportional correction parameter is formed according to the proportional correction condition; when both errors exceed the limit, two types of candidate parameters are formed simultaneously. When only one error exceeds the limit, but the corresponding candidate parameter can bring that error into the allowable range and the other error does not increase beyond the allowable range, the image twist measurement channel is marked as the channel to be calibrated and enters the candidate parameter verification. When multiple channels form candidate parameters simultaneously, they are saved and written to the combined verification mark respectively.

[0127] The detection difference changes synchronously with the cable tension, motor current, frame vibration, or take-up speed compensation. When the correction direction required for the same detection channel is opposite in two types of calibration segments, or when the candidate parameter causes other channels that already meet the independent reference relationship to produce out-of-limit differences, a record corresponding to the change in production status and the detection difference is generated, and no publishable candidate parameter is formed.

[0128] Write the detection channel number, reference data, candidate acquisition delay, ratio difference, comparison results of the two types of twist distance relationship, channel mark to be calibrated, reason for not being able to form parameters, calibration segment number, boundary mark number, and original data sequence range into the steel cable detection channel calibration judgment record; write the candidate acquisition delay, each candidate ratio correction parameter, image twist distance candidate offset correction parameter, applicable steel cable model, rope mold specification, tension range, traction speed range, twisting component speed range, current parameter version, and parameter formation basis into the steel cable candidate calibration parameter record.

[0129] S4. Select calibration segments and production status recovery sections that did not participate in the calculation of candidate calibration parameters, and verify the candidate calibration parameters of each detection channel. When the independent reference error, twist pitch maintenance relationship error and twist pitch change relationship error meet the corresponding allowable range, generate candidate calibration parameter verification records and parameter version generation condition records.

[0130] Read the basic record of online calibration of steel cable, the acquisition record of online calibration segment of steel cable, the calibration judgment record of steel cable detection channel, and the candidate calibration parameter record of steel cable. Select the complete calibration segment and the production status recovery section that did not participate in the candidate parameter calculation, and verify the candidate acquisition delay, candidate ratio correction parameters of traction speed detection channel and twisting component speed detection channel, as well as the candidate parameters of image twist measurement channel in sequence. Then, apply the candidate parameters that have been verified by each detection channel to the same set of raw data to generate the steel cable candidate calibration parameter verification record and parameter version generation condition record.

[0131] The processor reads four types of records according to the production task number, calibration cycle number, and candidate parameter number. It compares the cable type, rope-binding mold number, traction device number, twisting component number, image twist measurement device number, detection channel number, acquisition unit number, effective twist count version, effective path length version, image axial length conversion parameter version, twist transfer calibration version, current calibration parameter version, and the range of original data serial numbers referenced by the candidate parameters. If the current parameter version changes after the candidate parameters are formed, or if related equipment, mechanical reference position switches, image measurement devices, or effective path structures are replaced, or if the applicable range of the candidate parameters is inconsistent with the verification data, a record indicating that the verification conditions are not met is generated, and the corresponding verification is not performed.

[0132] Verification data is selected from complete calibration cycles that follow the candidate parameter formation cycle and have consistent applicability. Candidate acquisition delay and proportional correction parameters are verified using two types of calibration segments and boundary markers that were not involved in parameter calculation, through intervals and complete mechanical rotation cycles. The production state recovery section is only used to compare whether the traction speed, twisting component speed, cable tension, and image lay length have recovered to their stable correspondence before the start of the calibration cycle after parameter application; it is not used to verify acquisition delay alone. Data where the original sequence number overlaps with the parameter calculation data, or where different acquisition units generate different sequence numbers but correspond to the same section boundary marker range, is excluded as duplicate data.

[0133] Read the boundary markers, original data sequence number range, data integrity status, and segment validity status of the verification section. Stop the corresponding verification item if there is a missing boundary marker, abnormal boundary order, interrupted original data sequence number, abnormal unified clock synchronization, equipment alarm, speed compensation of the take-up device, cable tension exceeding the applicable range of candidate parameters, or changes in production tasks. Stop the verification of candidate parameters for twisting component speed if the mechanical reference position pulse is missing, repeatedly triggered, or the installation position changes; stop the verification of candidate parameters for image twist pitch if the version of the image axial length conversion parameter, camera or lens number, image resolution, exposure time, light source parameters, or measurement object type changes.

[0134] The processor retains the original detection data of the verification section and generates two sets of calculation results using both current calibration parameters and candidate calibration parameters. These two sets of results are bound to the same steel cable physical section and original data sequence number. Candidate parameters are not directly written to the sensor, acquisition unit, or production line controller. Each detection channel first verifies the candidate acquisition delay, then verifies the proportional correction parameter. If no candidate acquisition delay is generated and the boundary position difference corresponding to the current acquisition delay is within the allowable range, the current acquisition delay is used. If neither type of acquisition delay meets the positional tolerance, the corresponding proportional correction parameter is not verified.

[0135] When verifying the candidate acquisition delay of the traction speed detection channel, the production control speed feedback from a different source than the detection channel to be calibrated is used as a reference, and the axial position of the steel cable corresponding to its entry and exit from the target range is used as the reference boundary; when the two channels are from the same source, the candidate acquisition delay is not verified. When verifying the candidate acquisition delay of the twisting component speed detection channel, the mechanical reference speed calculated by the mechanical reference position pulse is used, and the starting point of the first complete rotation cycle that continuously meets the target range and the ending point of the last complete rotation cycle that continuously meets the target range are used as the reference boundary. When verifying the candidate acquisition delay of the image twist pitch, the boundary of the effective segment corresponding to the twist pitch change calibration segment at the image measurement position is used as a reference.

[0136] The current acquisition delay and the candidate acquisition delay are applied to the raw data of the same detection channel, respectively, to calculate the boundary position difference between the detected change segment and the reference change segment. A verification pass record is generated when the boundary position difference corresponding to the candidate acquisition delay is less than the result corresponding to the current parameters, and both the starting and ending position differences do not exceed the allowable error of the cable entity segment position. A verification fail record is generated when the candidate acquisition delay only decreases one boundary position difference while increasing the other, causing the detected change segment to exceed the boundary marker limit, or when opposite time movement directions are formed in two independent verification segments.

[0137] When verifying the candidate proportional correction parameters for traction speed, boundary markers are selected based on the following conditions: the control target remains unchanged, the traction wheel clamping state is normal, no slippage alarm is generated, and the cable tension is within the applicable range of the candidate parameters. For each effective range, using the original traction speed detection data corresponding to the determined acquisition delay, the absolute difference between the average detection value after processing with the current parameters and candidate parameters and the average transmission speed of the boundary markers is calculated. The median error for all effective ranges is also calculated.

[0138] ; ;

[0139] in, This is the verification error when using the current traction speed proportional correction parameters; This is the verification error when using candidate traction speed proportional correction parameters; The parameters are adjusted proportionally to the current traction speed. For candidate traction speed ratio correction parameters; For the first Average transmission speed of boundary markers for each effective interval; The average value of the raw traction speed detection data is obtained using a predetermined acquisition delay but without proportional correction parameters.

[0140] When the number of valid intervals reaches the minimum number of verification intervals in the basic record of online cable calibration, and the verification error of the candidate parameters is less than the verification error of the current parameters, the signed error of each valid interval is calculated. The signed error is the traction speed detection value after applying the corresponding parameters minus the average transmission speed of the boundary markers. When the absolute error of each valid interval after applying the candidate parameters does not exceed the allowable error of traction speed calibration, and the signed error does not increase in the opposite direction, a verification pass record for the traction speed candidate proportional correction parameter is generated. Positive deviations are corrected to negative deviations with smaller amplitudes and not exceeding the allowable error, or negative deviations are corrected to positive deviations with smaller amplitudes and not exceeding the allowable error, and are allowed to pass; when the error amplitude does not decrease or exceeds the allowable error after the sign is changed, a verification fail record is generated. The same signed error judgment rule is used for the candidate parameters of the twisting component speed, where the signed error is the speed detection value minus the speed verified by the mechanical reference position pulse.

[0141] When verifying candidate proportional correction parameters for the twisting component's rotational speed, rotational cycles with complete mechanical reference position pulses, no repeated triggering within one rotational cycle, continuous encoder data, and rotational speed fluctuations within the cycle not exceeding a stable range are selected. Using the original rotational speed detection data corresponding to the determined acquisition delay, the absolute difference between the average rotational speed detection value after processing with the current parameters and candidate parameters and the mechanical reference rotational speed are calculated, and the median error for all effective rotational cycles is also calculated.

[0142] ; ;

[0143] in, This is to verify the error when using the current twisting component speed ratio correction parameter; This is the verification error when using the rotational speed ratio correction parameter of the candidate twisting component; Correction parameters for the rotational speed ratio of the current twisting components; Correct parameters for the rotational speed ratio of candidate twisting components; For the first The mechanical reference speed for one effective rotational cycle; The value is the average of the original rotational speed detection data that was processed using a predetermined acquisition delay but without proportional correction parameters.

[0144] The number of effective rotation cycles reaches the minimum number of verification rotation cycles in the base record. Less than If the error after applying the candidate parameters in each effective rotation cycle does not exceed the allowable error for the twisting component speed calibration, and the error change direction is consistent in the speed increase, speed hold, and speed recovery intervals, a verification pass record for the candidate proportional correction parameters for the twisting component speed is generated. If the candidate parameters cause the error to exceed the limit in any cycle, or if the processed speed fluctuation exceeds the actual fluctuation range of the mechanical reference speed, a verification fail record is generated.

[0145] After the candidate parameters for the traction speed detection channel and the twisting component speed detection channel are independently verified, the original detection data of the same cable segment are processed using the corresponding parameters. The traction speed is calculated based on the effective twisting count, and the twist pitch is calculated by converting the image measurement position according to the twist pitch transfer calibration record corresponding to the current cable type, rope mold specification, and cable tension range. If the current parameters and candidate parameters of any basic detection channel cannot satisfy the independent reference relationship, the image twist pitch candidate parameters are not verified.

[0146] After the image twist candidate acquisition delay passes verification, or the boundary position difference corresponding to the current image acquisition delay does not exceed the allowable error, verify the image twist candidate ratio correction parameters and candidate offset correction parameters. Read the twist holding calibration segment and twist change calibration segment that were not involved in the candidate parameter calculation, and check the number of effective images, the difference in tension representative values, the image measurement object, and the image acquisition configuration of the two segments. If the number of effective images is insufficient, the difference in tension representative values ​​between the two segments exceeds the allowable tension difference of the calibration segment, or the image acquisition configuration is inconsistent, stop the image candidate parameter verification.

[0147] The original twist measurements of the two types of calibration segments are processed using both current and candidate image parameters. The twist retention change, the difference between the twist change and the twist change converted from the image measurement position, and the range of measurements within the segment are calculated. A verification pass record is generated when the twist retention change corresponding to the candidate parameters does not exceed the allowable twist retention error, the twist change direction is consistent, the difference in change does not exceed the allowable twist change error, and the range of measurements is not greater than the upper limit of the repeated measurement error of the image twist in the current parameter processing result and the basic record.

[0148] The permissible error for the location of the steel cable section is determined by multiplying the permissible time error corresponding to the cross-station data by the average transmission speed of the boundary markers, and then adding it to the repetition error of the boundary marker positions. After adding a verification interval, the median error is considered to have stabilized when the change in the median error does not exceed 10% of the corresponding calibration permissible error. The permissible deviation for production status recovery, the stable range before start-up, and the stable correspondence are all based on the control quantity stability permissible deviation in the basic records.

[0149] Read the first and second production state recovery sections, and calculate the absolute difference between the current image parameters and the candidate image parameters relative to the converted lay length at the image measurement position. If the difference corresponding to the candidate parameter is less than the result corresponding to the current parameter, and the image lay length measurement value within the recovery section does not continuously increase or decrease along the cable axis, the recovery section verification is recorded as passed. If the candidate parameter only passes in the recovery section but not in the complete two types of calibration segments, no image candidate parameter verification pass record is generated.

[0150] When multiple detection channels generate candidate parameters, they are applied together in a fixed order. First, the verified candidate acquisition delay and candidate ratio correction parameters for traction speed are applied to the raw traction speed data. Then, the verified candidate acquisition delay and candidate ratio correction parameters for twisting component speed are applied to the raw speed data. The image measurement position is converted into twist pitch using the results of the two types of processing. Finally, the verified candidate acquisition delay, candidate ratio correction parameters, and candidate offset correction parameters for image twist pitch are applied to the raw measured value of image twist pitch.

[0151] After joint application, if the independent reference error of traction speed, the independent reference error of twisting component speed, the error of twist pitch maintenance relationship, and the error of twist pitch change relationship all do not exceed the corresponding allowable errors, and the error of each channel does not increase relative to the individual verification results, a record of successful joint application of multi-channel candidate parameters is generated. If each channel candidate parameter passes verification individually, but any error exceeds the limit after joint application, a record of failed joint application is generated, and no record of parameter version generation conditions is formed.

[0152] Candidate parameters are only verified a cumulative number of times within the pre-defined range of cable type, rope die specifications, traction speed range, twisting component speed range, cable tension range, and image acquisition configuration based on the online calibration record. The applicable range is not redefined based on a single verification result. The minimum number of verifications is determined based on the error dispersion of historical qualified calibration records and the repeatability of candidate parameters, taking three to ten independent valid segments corresponding to the median verification error stabilizing with the addition of new independent valid segments. The maximum verification cycle is determined based on the frequency of production tasks forming complete calibration segments and the allowable retention time of calibration parameters, taking three to eight production cycles. When historical records are insufficient, the maximum number of verifications is determined based on the number of repeated measurements in the detection channel and the duration of the production task, and the source is recorded. When verification results within the same preset range are inconsistent, the candidate parameter is written to the inconsistent verification result status. If the minimum number of verifications is not reached even after the maximum verification cycle, or if two consecutive verification cycles produce pass and fail results respectively, it indicates that the candidate parameter lacks cross-cycle repeatability; cumulative verification is stopped, and the current calibration parameter is retained.

[0153] The candidate parameter number, detection channel number, production task number for verification, calibration cycle number, cable physical section range, acquisition delay verification result, current parameter verification error, candidate parameter verification error, recovery section verification result, verification results of the two types of calibration segments, results of the joint application of multi-channel candidate parameters, reasons for verification failure, and the range of original data serial numbers referenced are written into the cable candidate calibration parameter verification record. The number of independent verification segments is three to ten, and the number of consecutively passed calibration cycles is two to five. The specific values ​​are determined based on the error dispersion of historical qualified calibration records, repeated measurement results of detection channels, and the frequency of production tasks forming complete calibration segments, and are written into the cable online calibration basic record. When the candidate parameter reaches the specified number of independent verification segments and consecutively passed calibration cycles, and the results of the joint application of multi-channel candidate parameters are passed, the candidate parameter number, number of verification passes, number of consecutively passed cycles, applicable scope, and current parameter version are written into the parameter version generation condition record. A new calibration parameter version is generated based on this record, the applicable scope and effective time are set, the new calibration parameter version is activated after the current production task ends, and the original calibration parameter version is marked as deactivated and retained. If a new calibration parameter version fails verification or the original data continuity is abnormal after it is enabled, stop using the new calibration parameter version, restore the original calibration parameter version, and generate a parameter version rollback record.

[0154] To verify the feasibility of establishing cross-inspection station data correspondence using segment boundary markers, and to verify the impact of calculating cable transfer time and inspection channel acquisition delay separately on the corresponding results of cable segments, a multi-source sensor data co-simulation environment for a cable production line was established. Simulation objects included the traction device, twisting component, segment boundary marker device, marker detector, traction speed detection channel, twisting component speed detection channel, image twist measurement channel, and auxiliary verification channel. Based on the cable online calibration baseline records, the effective twist count, effective path length between inspection stations, image axial length conversion parameters, twist transfer parameters, inspection channel sampling period, filtering time, and communication delay were set.

[0155] In the simulation, the effective path length from detection position a to detection position b was set to 6.40 meters, and the effective number of twists was set to 1. The sampling periods for traction speed, twisting component rotation speed, cable tension, and image channel were set to 10 milliseconds, 10 milliseconds, 20 milliseconds, and 40 milliseconds, respectively. The traction speed and rotation speed channels adopted a first-order inertial response model with time constants of 40 milliseconds and 60 milliseconds, respectively, and were superimposed with communication delays of 20 milliseconds, 40 milliseconds, or 60 milliseconds and moving average filters of 30 milliseconds, 60 milliseconds, or 90 milliseconds. The control quantity first held the reference value for 2 seconds, then linearly changed to the target value within 0.5 seconds, held for 4 seconds, and then recovered within 0.5 seconds. The twist pitch holding calibration segment synchronously changed the traction speed and rotation speed, while the twist pitch changing calibration segment only changed the rotation speed.

[0156] Traction speed proportional deviation, rotational speed proportional deviation, and image twist proportional deviation are generated with uniform distributions within the ranges of ±1.5%, ±1.2%, and ±2%, respectively. Offsets are generated within the ranges of ±0.008 m / s, ±1.5 rpm, and ±1 mm, respectively. The standard deviations of random noise are set to 0.003 m / s, 0.4 rpm, and 0.6 mm, respectively. The effective axial field of view for the image is set to 0.36 m, and the cross-frame measurement window does not exceed 0.96 m. The result timestamp is taken from the center time of the window. Frame drops are generated with a 2% independent probability; if three consecutive frames are lost or there are fewer than five effective images within the measurement window, the image twist measurement value is not output. Sixty sets of original records are generated for each scene using a random seed 20250721. Figure 5 and Figure 6 The arithmetic mean and standard deviation are calculated using all valid records without deleting any data. The boundary position error, the error before and after the application of candidate parameters, and the reason for invalidity are saved for each group.

[0157] For each set of simulation data, three processing methods were employed: direct correspondence based on timestamps, compensation only for cable transmission time, and correspondence using segment boundary markers to separate the acquisition delay of the detection channel. The axial positions of the cable corresponding to the starting and ending boundaries at each detection station were calculated. The calculated positions were compared with the simulated cable segment boundary positions, and the larger of the starting and ending boundary position errors was taken as the cable segment boundary position error for that simulation. The average and standard deviation were calculated based on the repeated simulation results to form... Figure 5 .

[0158] Figure 5 In the diagram, the curve "Direct Correspondence by Timestamp" represents the boundary position error caused by not distinguishing between cable delivery time and detection channel acquisition delay; the curve "Compensating Only for Cable Delivery Time" represents the boundary position error after compensating for cable delivery time based on the effective path length between detection stations and the arrival time of boundary markers; the curve "Section Boundary Markers Correspond to and Separate Detection Channel Acquisition Delay" represents the boundary position error formed according to the processing method of this invention, and the error bars represent the standard deviation of repeated simulation results. Figure 5 It is evident that, under scenarios involving changes in traction speed, twisting component rotation speed, increased communication delay, increased filtering time, and fluctuations in cable tension, after establishing the relationship between actual cable segments using segment boundary markings and separating the detection channel acquisition delay, the positional error of the actual cable segment boundary remains within a low range. This indicates that multi-source sensor data from different detection stations can correspond to the same actual cable segment.

[0159] To verify the formation and verification process of candidate calibration parameters, acquisition delay bias, proportional bias, and image twist offset were added to the traction speed detection channel, twisting component speed detection channel, and image twist measurement channel, respectively. The traction speed detection data was verified using the boundary marker average transmission speed, and the twisting component speed detection data was verified using the mechanical reference position pulse. Finally, candidate calibration parameters were formed by converting the image measurement positions in the twist holding calibration segment and the twist changing calibration segment, and then comparing the twist with the image twist measurement results.

[0160] Complete calibration segments and production recovery sections not involved in candidate calibration parameter calculations were selected. The same set of raw data was processed using both current calibration parameters and validated candidate calibration parameters. Independent reference errors for traction speed, twisting component speed, twist pitch maintenance, and twist pitch variation were recorded. Each error was divided by its corresponding allowable error to form a normalized error; a normalized error not exceeding one indicates that the corresponding error is within the allowable range. The normalized error was then calculated based on the average and standard deviation of the repeated simulation results. Figure 6 .

[0161] Figure 6 In the diagram, the curve "Applying Current Calibration Parameters" represents the normalized errors formed after processing the raw test data using the current parameters; the curve "Applying Verified Candidate Calibration Parameters" represents the normalized errors formed after applying the candidate calibration parameters in the order of traction speed, twisting component speed, and image twist pitch; and the curve "Corresponding Upper Limit of Allowable Range" represents the allowable error boundary for each verification item. Figure 6 As can be seen, after the verified candidate calibration parameters are applied together, the independent reference error of traction speed, the independent reference error of twisting component speed, the error of twist pitch maintenance relationship, and the error of twist pitch change relationship are all within the corresponding allowable range. This indicates that the candidate calibration parameters can correspond to the acquisition delay and measurement deviation of each detection channel respectively, and can maintain the calculation relationship between traction speed, twisting component speed and image twist pitch.

[0162] This embodiment also provides an online calibration system for multi-source sensor data of a steel cable production line, including: a basic parameter module, used to acquire data on steel cable production tasks, processes, equipment, and testing channels, determine the effective twisting times, effective path length between testing stations, image axial length conversion parameters, and twist transfer parameters, and generate a basic record for online calibration of the steel cable; a segment acquisition module, used to form twist maintenance calibration segments, production state recovery segments, and twist change calibration segments based on the basic record for online calibration of the steel cable, form segment boundary markers, and map the traction speed, twisting component speed, image twist, and auxiliary verification data acquired from each testing station to the same physical segment of the steel cable, generating a record for online calibration segment acquisition of the steel cable; and a calibration calculation module, used to calculate the steel cable... The online calibration segment acquisition records calculate the cable transmission time and the acquisition delay of the detection channel, verify the traction speed detection data using the average transmission speed of the boundary markers, verify the twisting component speed detection data using the mechanical reference position pulse, and generate candidate calibration parameters for each detection channel by combining the image twist measurement results of the two types of calibration segments. The parameter verification module is used to select the complete calibration segment and the production state recovery section that did not participate in the candidate calibration parameter calculation, and sequentially verify the candidate acquisition delay, candidate ratio correction parameters, and image twist candidate offset correction parameters for each detection channel. When the independent reference error, twist maintenance relationship error, and twist change relationship error meet the corresponding allowable ranges, the module generates candidate calibration parameter verification records and parameter version generation condition records.

[0163] In summary, this invention achieves cross-station data position matching by forming a pitch-maintaining calibration segment and a pitch-variation calibration segment, and by using segment boundary markers to map multi-source sensor data from each workstation to the same physical cable segment, thus improving the accuracy of multi-source sensor data correspondence. By calculating the cable transmission time and detection channel acquisition delay separately, and verifying each detection channel using the average transmission speed of boundary markers, mechanical reference position pulses, and image pitch measurement results, and by using independent segments to verify candidate calibration parameters, the invention achieves detection channel deviation confirmation and improves the reliability of online calibration results.

[0164] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An online calibration method for multi-source sensor data in a steel cable production line, characterized in that, include: Acquire data on steel cable production tasks, processes, equipment, and testing channels; determine the effective number of twists, effective path length between testing stations, image axial length conversion parameters, and twist transfer parameters; and generate basic records for online steel cable calibration. Based on the basic records of online calibration of steel cables, calibration segments for maintaining the twist pitch with proportional changes in traction speed and twisting component speed and calibration segments for twist pitch changes with non-proportional changes are formed in sequence. Multi-source sensor data collected at each testing station are mapped to the same physical section of the steel cable using section boundary markings, thereby generating online calibration segment collection records of steel cables. By collecting and recording data from online calibration segments of the steel cable, the cable transmission time and the acquisition delay of the detection channel are calculated respectively. The average transmission speed of the boundary marker and the pulse of the mechanical reference position are used to verify the traction speed and the rotation speed of the twisting component. The candidate calibration parameters of each detection channel are generated by combining the image twist measurement results of the two types of calibration segments. The calibration segments and production recovery sections that were not involved in the calculation of candidate calibration parameters were selected to verify the candidate calibration parameters of each detection channel. When the independent reference error, twist pitch maintenance relationship error and twist pitch change relationship error meet the corresponding allowable range, a candidate calibration parameter verification record and a parameter version generation condition record are generated.

2. The online calibration method for multi-source sensor data in a steel cable production line as described in claim 1, characterized in that, The process of determining the effective number of twists, the effective path length between detection stations, the image axial length conversion parameter, and the twist transfer parameter to generate the basic record for online cable calibration includes: Read the structural data and transmission relationship data of the twisting component to determine the effective number of twists per revolution of the twisting component; Determine the effective path length between testing stations along the actual operating centerline of the steel cable; The image axial length conversion parameters are determined based on the physical length and pixel length of the image calibration component; The pitch transmission parameters are determined based on the traction speed, twisting component speed, effective twisting times, and offline pitch measurement values ​​of the corresponding cable section.

3. The online calibration method for multi-source sensor data of a steel cable production line as described in claim 1 or 2, characterized in that, The process of determining the effective number of twists, the effective path length between detection stations, the image axial length conversion parameter, and the twist transfer parameter to generate the basic record for online cable calibration includes: Save the lay pitch transmission calibration data according to the cable type, rope mold specifications, and cable tension range; When there are at least three traction speeds for calculating the pitch and the corresponding offline pitch measurement values, determine the pitch transmission ratio parameter and the pitch transmission offset parameter. If the fitting result does not meet the error requirements for pitch measurement, save the correspondence between the pitch calculated by traction speed and the offline pitch measurement value.

4. The online calibration method for multi-source sensor data of a steel cable production line as described in claim 3, characterized in that, The pitch maintenance calibration segment, which sequentially forms a pitch change calibration segment with the traction speed and the twisting component rotation speed changing proportionally, and a pitch change calibration segment with a non-proportional change, includes: When the production task, process formula and equipment operating status meet the conditions for forming a calibration segment, the traction speed and twisting component speed are adjusted synchronously so that the difference between the calculated twist pitch before and after the adjustment does not exceed the allowable error for twist pitch maintenance, thus forming a twist pitch maintenance calibration segment. After restoring the production status before calibration, maintain the traction speed and adjust the twisting component speed so that the twist pitch change reaches the minimum identifiable twist pitch change in the image and is within the allowable twist pitch range of the product, thus forming a twist pitch change calibration segment.

5. The online calibration method for multi-source sensor data of a steel cable production line as described in claim 4, characterized in that, The step of mapping the multi-source sensor data collected by each testing station to the same steel cable physical section using section boundary markings includes: After the traction speed control value, twisting component speed control value, steel cable tension, motor current and frame vibration meet the corresponding stability conditions, the start boundary mark and end boundary mark of the calibration segment are formed respectively. Based on the boundary marker number, boundary marker arrival time, and original data sequence number, extract the traction speed, twisting component speed, image twist pitch, and auxiliary verification data for each inspection station located between the same starting and ending boundary markers.

6. The online calibration method for multi-source sensor data of a steel cable production line as described in claim 5, characterized in that, The calculation of the cable transmission time and the acquisition delay of the detection channel includes: The steel cable transfer time is determined based on the arrival time of the same section boundary mark at different testing stations and the effective path length between testing stations. The steel cable transfer time is then used to correspond the steel cable entity sections in the data of different testing stations. The acquisition delay of the detection channel is determined based on the difference between the starting position and the ending position of the detection data change segment and the reference change segment, and the acquisition delay of the detection channel is used to correct the time offset caused by sensor response, data filtering, data processing and communication.

7. The online calibration method for multi-source sensor data of a steel cable production line as described in claim 6, characterized in that, The data used to verify the traction speed and twisting component rotation speed using boundary marker average transmission speed and mechanical reference position pulses include: The average transmission speed of the boundary markers is calculated based on the time difference and effective path length of the boundary markers passing through adjacent detection positions in the same section. The average transmission speed of the boundary markers is then compared with the original detection data of the traction speed within the same time range to determine the candidate ratio correction parameters for the traction speed. The mechanical reference speed is calculated based on the time interval between adjacent mechanical reference position pulses. The mechanical reference speed is then compared with the original detection data of the twisting component speed within the corresponding rotation cycle to determine the candidate proportional correction parameter for the twisting component speed.

8. The online calibration method for multi-source sensor data in a steel cable production line as described in claim 7, characterized in that, The generation of candidate calibration parameters for each detection channel by combining the image twist measurement results of the two types of calibration segments includes: The traction speed verified by the average transmission speed of the boundary markers and the rotation speed of the twisting component verified by the mechanical reference position pulse are used to calculate the representative value of the twist pitch by converting the image measurement position of the twist pitch maintenance calibration segment and the twist pitch change calibration segment respectively. Read the valid images of the two types of calibration segments, complete the conversion between pixel distance and entity distance, and take the median of the twist measurement values ​​of the valid images as the original representative value of the image twist. Based on the image measurement positions of the two types of calibration segments, the representative values ​​of the twist pitch and the original representative values ​​of the image twist pitch are converted, and the candidate ratio correction parameters and candidate offset correction parameters of the image twist pitch are determined.

9. The online calibration method for multi-source sensor data of a steel cable production line as described in claim 8, characterized in that, The verification of candidate calibration parameters for each detection channel, when the independent reference error, twist maintenance relationship error, and twist variation relationship error meet the corresponding allowable ranges, generates candidate calibration parameter verification records and parameter version generation condition records, including: Select complete calibration segments whose formation time is later than that of candidate calibration parameters and which have not participated in the calculation of candidate calibration parameters. First, verify the candidate acquisition delay of each detection channel, and then use the verified candidate acquisition delay to verify the corresponding candidate ratio correction parameter. The stable correspondence after the application of verification parameters was carried out using the production status recovery section; The verified traction speed, twisting component speed, and image twist candidate parameters are applied sequentially to the same set of raw data. When the independent reference error of traction speed, the independent reference error of twisting component speed, the error of twist maintenance relationship, and the error of twist change relationship are all within the corresponding allowable range, a parameter version generation condition record is generated.

10. An online calibration system for multi-source sensor data of a steel cable production line, based on the online calibration method for multi-source sensor data of a steel cable production line according to any one of claims 1 to 9, characterized in that, include: The basic parameter module is used to acquire data on steel cable production tasks, processes, equipment, and testing channels, determine the effective number of twists, the effective path length between testing stations, the image axial length conversion parameters, and the twist transfer parameters, and generate basic records for online calibration of steel cables. The segment acquisition module is used to form a pitch maintenance calibration segment, a production status recovery segment, and a pitch change calibration segment based on the basic records of online cable calibration. It also forms segment boundary marks and maps the traction speed, twisting component speed, image pitch, and auxiliary verification data collected by each testing station to the same physical segment of the cable, generating online cable calibration segment acquisition records. The calibration calculation module is used to calculate the cable transmission time and detection channel acquisition delay based on the online calibration segment acquisition records of the cable, verify the traction speed detection data using the average transmission speed of the boundary marker, verify the twisting component speed detection data using the mechanical reference position pulse, and generate candidate calibration parameters for each detection channel by combining the image twist measurement results of the two types of calibration segments. The parameter verification module is used to select the complete calibration segment and the production state recovery segment that did not participate in the calculation of candidate calibration parameters, and sequentially verify the candidate acquisition delay, candidate ratio correction parameters and image twist candidate offset correction parameters of each detection channel. When the independent reference error, twist maintenance relationship error and twist change relationship error meet the corresponding allowable range, the module generates candidate calibration parameter verification records and parameter version generation condition records.

Images