A carbon fiber production whole-process data automatic acquisition system, method and medium
By using a propagation clock and a changing topology recursive acquisition method, the data mismatch problem caused by fluctuations in operating speed and equipment start-up and shutdown in carbon fiber production was solved, achieving the continuity and accuracy of data throughout the entire process and improving the accuracy of data correlation.
Patent Information
- Application Number
- CN202611569011.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-12
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies in carbon fiber production suffer from data mismatch and insufficient correlation accuracy across different production stages due to fluctuations in fiber bundle running speed and equipment start-up and shutdown, making it difficult to achieve continuity and accuracy of production data throughout the entire process.
By adopting the propagation clock and changing topology recursive acquisition method, a data acquisition channel is established, material segments are divided, a full-process propagation clock is generated, process state change markers are identified, and the acquisition strategy is dynamically adjusted to improve the accuracy of data correlation.
It improves the continuity and accuracy of data throughout the carbon fiber production process, reduces the amount of invalid data generated by fixed-period data collection and the risk of missing key change data, and enhances the accuracy of cross-process data correlation.
Smart Images

Figure CN123070503A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of production data acquisition, and in particular to an automatic data acquisition system, method and medium for the entire carbon fiber production process. Background Technology
[0002] Carbon fiber production typically includes continuous production stages such as spinning, drawing, pre-oxidation, and carbonization. Existing technologies mostly collect production data such as fiber bundle tension, process temperature, process pressure, and medium flow rate according to equipment, collection time, or fixed sampling period, and store and associate the data according to the production batch or stage sequence.
[0003] Because fiber bundles experience fluctuations in operating speed, equipment start-up and shutdown, and material stretching or contraction during continuous transport, the arrival time and length position of the same material segment can change in different production sections. Existing fixed-time correspondence methods are prone to data mismatch across sections and make it difficult to dynamically adjust the collection period according to the actual propagation process of changes in process status, resulting in insufficient continuity and correlation accuracy of production data throughout the entire process. Summary of the Invention
[0004] One objective of this invention is to propose an automatic data acquisition system, method, and medium for the entire carbon fiber production process. This invention employs propagation clock and changing topology recursive acquisition to achieve data correlation throughout the entire carbon fiber production process, thereby improving continuity and accuracy.
[0005] An automatic data acquisition method for the entire carbon fiber production process according to an embodiment of the present invention includes the following steps:
[0006] Establish data acquisition channels for each carbon fiber production section and collect production batch data, section path length data, fiber bundle operation data, and process parameter data;
[0007] Based on production batch data and fiber bundle operation data, the continuously conveyed fiber bundle is divided into multiple material segments, and a material segment operation data sequence is generated.
[0008] Based on the process path length data and material segment operation data sequence, the estimated arrival time of each material segment to the downstream production section is determined, and the whole process propagation clock is generated.
[0009] The process parameter data is mapped to the material segment length domain, the process state change markers that maintain the same change topology across segments are identified, the actual propagation time between adjacent production segments is determined, and the process propagation clock of the entire process is corrected.
[0010] Based on the corrected whole-process propagation clock and the node connection order in the phase chain of the changing nodes, the expected arrival time of each changing node of the target material segment in the downstream production section of the target is recursively corrected, and a dynamic acquisition strategy is generated.
[0011] Based on the dynamic acquisition strategy, the data acquisition results are obtained automatically.
[0012] Optionally, the fiber bundle operating data includes fiber bundle operating speed, cumulative conveying length, and equipment start / stop status.
[0013] Optionally, the process parameter data includes fiber bundle tension, process temperature, process pressure, and medium flow rate.
[0014] Optionally, the generation of the material segment operation data sequence specifically includes:
[0015] The current production batch is determined based on the production batch data, and a corresponding production batch identifier is generated;
[0016] Based on the cumulative conveying length in the fiber bundle operation data, the position where the cumulative conveying length reaches the preset material segment length is used as the material segment boundary. The continuously conveyed fiber bundle corresponding to the current production batch is divided into multiple material segments, and material segment identifiers are generated according to the conveying order of the material segments.
[0017] Extract the fiber bundle operation data corresponding to each material segment, associate the fiber bundle operation data with the production batch identifier and material segment identifier, and organize them to generate a material segment operation data sequence.
[0018] Optionally, the generation of the end-to-end process propagation clock specifically includes:
[0019] The length of the material transport path from the current production section to each downstream production section is determined based on the section path length data;
[0020] Extract the time when each material segment enters the current production section, the fiber bundle running speed, and the equipment start-stop status from the material segment operation data sequence. Determine the expected conveying time for each material segment based on the conveying path length, fiber bundle running speed, and equipment start-stop status.
[0021] Based on the time when each material segment enters the current production section and the estimated transport time, the estimated arrival time of each material segment to each downstream production section is determined, and the estimated arrival time is organized according to the material segment identification and the production section sequence to generate the whole process propagation clock.
[0022] Optionally, the correction of the entire process propagation clock specifically includes:
[0023] Based on the material segment identifier and cumulative conveying length, the process parameter data is mapped from the time domain to the material segment length domain to generate a multi-parameter length sequence;
[0024] Extract the parameter type, change direction, change location and adjacent change interval from the multi-parameter length sequence, and use the parameter changes as change nodes to construct the material segment process change topology;
[0025] Using parameter type, change direction, connection order and change interval ratio as constraints, the material segment process change topology of adjacent production sections is scaled and matched to determine the process state change marker that maintains topology consistency across sections.
[0026] Based on the occurrence time of the changing node in the same process state change mark in the adjacent production section, construct the node time displacement vector, perform common displacement-relative offset decomposition on the node time displacement vector, and determine the common displacement component as the actual propagation time.
[0027] The estimated arrival time of the corresponding material segment to the adjacent downstream production segment is corrected based on the actual propagation time, and the estimated arrival time of the corresponding material segment to the subsequent production segment is updated according to the production segment sequence, generating the corrected whole-process process propagation clock.
[0028] Optionally, the generation of the dynamic acquisition strategy specifically includes:
[0029] The target material segment and the target downstream production segment are determined based on the material segment identifier associated with the process state change marker, and a change node phase chain is generated based on the distance of the change position of each change node relative to the starting position of the material segment.
[0030] The estimated arrival time of the target material segment to the target downstream production section is extracted from the corrected whole-process process propagation clock. The estimated transit time is determined based on the material segment length and fiber bundle running speed. The estimated arrival time of each changing node is determined according to the phase chain of the changing nodes.
[0031] Before the first change node is expected to arrive, the data acquisition channel for the corresponding parameter type is opened. When the parameter type and change direction of the acquired data match the first change node, the actual arrival time of the first change node is locked, and the expected arrival time of the remaining change nodes is corrected.
[0032] For each subsequent change node locked, the local phase propagation rate is determined based on the actual arrival interval and phase interval of the adjacent locked change nodes. The estimated arrival time of the unlocked change nodes is then recursively corrected, and a dynamic acquisition strategy is generated.
[0033] Optionally, the generation of the automatic data collection results specifically includes:
[0034] According to the dynamic data collection strategy, production data of each production section is collected and the production data is associated with production batch identifier, material section identifier and corresponding production section;
[0035] The production data of the same material segment are associated with the production data of each production segment according to the production segment sequence, and the data collection results are automatically generated.
[0036] An automatic data acquisition system for the entire carbon fiber production process according to an embodiment of the present invention includes:
[0037] The data acquisition module is used to establish data acquisition channels for each carbon fiber production section and to collect production batch data, section path length data, fiber bundle operation data, and process parameter data.
[0038] The material segment division module is used to divide the continuously conveyed fiber bundle into multiple material segments based on production batch data and fiber bundle operation data, and generate a material segment operation data sequence.
[0039] The propagation clock generation module is used to determine the estimated arrival time of each material segment to the downstream production segment based on the segment path length data and material segment operation data sequence, and generate the entire process propagation clock.
[0040] The propagation clock correction module is used to map process parameter data to the material segment length domain, identify process state change markers that maintain consistent change topology across segments, determine their actual propagation time between adjacent production segments, and correct the entire process propagation clock.
[0041] The dynamic acquisition strategy generation module is used to recursively correct the expected arrival time of each changing node in the downstream production section of the target material segment based on the corrected whole-process process propagation clock and the node connection order in the phase chain of the changing nodes, and generate a dynamic acquisition strategy.
[0042] The automatic data acquisition result generation module is used to obtain automatic data acquisition results based on dynamic acquisition strategies.
[0043] According to an embodiment of the present invention, a computer-readable medium stores a computer program that, when executed by a processor, enables the processor to perform an automatic data acquisition method for the entire carbon fiber production process.
[0044] The beneficial effects of this invention are:
[0045] This invention establishes data acquisition channels for each stage of carbon fiber production. Based on production batch data, cumulative conveying length, and fiber bundle operation data, it divides continuously conveyed fiber bundles into multiple material segments with material segment identifiers. This allows production data, originally recorded separately according to equipment or acquisition time, to be associated with the same material segment as a unified object. The estimated arrival time of each material segment to the downstream production stage is determined based on the stage path length, fiber bundle running speed, and equipment start / stop status. A full-process process propagation clock is constructed, which can characterize the time relationship of material segments passing through different production stages along the production process, reducing the impact of running speed variations and equipment shutdowns on the cross-stage data correspondence.
[0046] This invention maps fiber bundle tension, process temperature, process pressure, and medium flow rate from the time domain to the material segment length domain. By constructing and scaling the material segment process change topology according to parameter type, change direction, connection order, and change interval ratio, it ensures that the same set of process state changes can still establish a correspondence between adjacent production segments even after the material segment has stretched or contracted. By constructing node time displacement vectors and performing common displacement and relative offset decomposition, it can eliminate time offset interference caused by local parameter response delays or abnormal changes in nodes, obtaining the actual propagation time of the process state change markers. Based on this, it corrects the expected arrival time of the corresponding material segment to the adjacent downstream production segment and subsequent production segments, improving the accuracy of the overall process propagation clock.
[0047] This invention further predicts the arrival time of each changing node to the target downstream production section based on the phase chain of the changing nodes. After the first changing node is locked by the actual collected data, the estimated arrival time of the remaining changing nodes is corrected. Furthermore, the local phase propagation rate is determined based on the actual arrival interval and phase interval of adjacent locked changing nodes, and the estimated arrival time of the unlocked changing nodes is recursively corrected. This allows for the dynamic determination of data acquisition channels and target acquisition periods according to different parameter types and node connection sequences, reducing the amount of invalid data generated by fixed-period acquisition and the risk of missing key changing data. It also accurately associates the collected data with production batch identifiers, material segment identifiers, and corresponding production sections, improving the continuity, relevance, and cross-section correlation accuracy of data acquisition throughout the carbon fiber production process. Attached Figure Description
[0048] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:
[0049] Figure 1 This is a flowchart of an automatic data acquisition method for the entire carbon fiber production process proposed in this invention;
[0050] Figure 2This is a thermal diagram showing the material segment arrival error across processes before and after correction of the automatic data acquisition method for the entire carbon fiber production process proposed in this invention. Detailed Implementation
[0051] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.
[0052] refer to Figure 1 A method for automatic data acquisition throughout the entire carbon fiber production process includes the following steps:
[0053] Establish data acquisition channels for each carbon fiber production section and collect production batch data, section path length data, fiber bundle operation data, and process parameter data;
[0054] Based on production batch data and fiber bundle operation data, the continuously conveyed fiber bundle is divided into multiple material segments, and a material segment operation data sequence is generated.
[0055] Based on the process path length data and material segment operation data sequence, the estimated arrival time of each material segment to the downstream production section is determined, and the whole process propagation clock is generated.
[0056] The process parameter data is mapped to the material segment length domain, the process state change markers that maintain the same change topology across segments are identified, the actual propagation time between adjacent production segments is determined, and the process propagation clock of the entire process is corrected.
[0057] Based on the corrected whole-process propagation clock and the node connection order in the phase chain of the changing nodes, the expected arrival time of each changing node of the target material segment in the downstream production section of the target is recursively corrected, and a dynamic acquisition strategy is generated.
[0058] Based on the dynamic acquisition strategy, the data acquisition results are obtained automatically.
[0059] In this embodiment, the fiber bundle operation data includes fiber bundle operation speed, cumulative conveying length, and equipment start / stop status.
[0060] In this embodiment, the process parameter data includes fiber bundle tension, process temperature, process pressure, and medium flow rate.
[0061] In this embodiment, the generation of the material segment operation data sequence specifically includes:
[0062] The current production batch is determined based on the production batch data, and a corresponding production batch identifier is generated;
[0063] Based on the cumulative conveying length in the fiber bundle operation data, the position where the cumulative conveying length reaches the preset material segment length is used as the material segment boundary. The continuously conveyed fiber bundle corresponding to the current production batch is divided into multiple material segments, and material segment identifiers are generated according to the conveying order of the material segments.
[0064] In practice, the preset material segment length is set to 10m, and each 10m of fiber bundle is divided into one material segment.
[0065] Extract the fiber bundle operation data corresponding to each material segment, associate the fiber bundle operation data with the production batch identifier and material segment identifier, and organize them according to the material segment conveying order to generate a material segment operation data sequence.
[0066] In this embodiment, the generation of the entire process propagation clock specifically includes:
[0067] The transport path length of each material segment from the current production segment to each downstream production segment is determined based on the segment path length data;
[0068] In practice, the section path length data is the distance that carbon fiber travels along the actual conveying path between adjacent production sections.
[0069] Extract the time when each material segment enters the current production section, the fiber bundle running speed, and the equipment start-stop status from the material segment operation data sequence. Determine the expected conveying time for each material segment based on the conveying path length, fiber bundle running speed, and equipment start-stop status.
[0070] In practice, the equipment start-stop status refers to the operating status of the carbon fiber conveying equipment, including the start-up status, the running status, and the stop status.
[0071] Based on the time when each material segment enters the current production section and the estimated transport time, the estimated arrival time of each material segment to each downstream production section is determined, and the estimated arrival time is organized according to the material segment identification and the production section sequence to generate the whole process propagation clock.
[0072] The whole-process propagation clock is a time-series data structure that organizes the time when each material segment enters the current production segment and the expected arrival time to each downstream production segment according to the material segment identifier and production segment sequence. The expected arrival time is used to characterize the time relationship of each material segment arriving at different production segments along the carbon fiber production process.
[0073] This invention uses a full-process process propagation clock to time-correspond data collected from different production sections of the same material segment according to the actual transportation process. This reduces cross-section data mismatch caused by changes in fiber bundle running speed and equipment downtime, and provides a time reference for subsequent marking and correction of expected arrival times based on changes in process status and for determining the target collection period of downstream production sections. This improves the continuity and correlation accuracy of data collection throughout the carbon fiber production process.
[0074] In this embodiment, the correction of the entire process propagation clock specifically includes:
[0075] Based on the material segment identifier and cumulative conveying length, the process parameter data is mapped from the time domain to the material segment length domain to generate a multi-parameter length sequence;
[0076] In practice, fiber bundle tension, process temperature, process pressure, and medium flow rate are arranged according to the acquisition time to form time-domain process parameter data; then, the material segment corresponding to each process parameter data is determined according to the material segment identifier, and the difference between the cumulative conveying length at each acquisition time and the starting cumulative conveying length of the corresponding material segment is determined as the length position within the material segment; the process parameter data are rearranged according to the material segment identifier and length position to generate a multi-parameter length sequence.
[0077] In practice, the time domain is a data representation formed by arranging fiber bundle tension, process temperature, process pressure, and medium flow rate using the acquisition time as an index; the material segment length domain is a data representation formed by determining the length position within a material segment based on the material segment identifier, the difference between the cumulative conveying length corresponding to each acquisition time and the initial cumulative conveying length of the corresponding material segment, and using the length position within the material segment as the main index and retaining the acquisition time as the associated attribute; the multi-parameter length sequence is a sequence formed by correspondingly organizing fiber bundle tension, process temperature, process pressure, and medium flow rate and their acquisition times according to the material segment identifier and the length position within the material segment.
[0078] Extract the parameter type, change direction, change location and adjacent change interval from the multi-parameter length sequence, and use the parameter changes as change nodes to construct the material segment process change topology;
[0079] In specific implementation, for fiber bundle tension, process temperature, process pressure, and medium flow rate in each production section, the same length sampling interval as the current multi-parameter length sequence is used. The absolute value of the difference between adjacent length positions of the corresponding process parameters in historical stable production batches is calculated. The 95th percentile value of each absolute difference, sorted from smallest to largest, is set as the change threshold for the corresponding production section and the corresponding parameter type. For the current material segment, the difference between various process parameters between adjacent length positions is calculated sequentially according to the length position within the material segment. When the absolute value of the difference reaches the corresponding change threshold, the next length position is determined as the change position. The parameter type is determined according to the process parameter category, and the change direction is determined according to the positive or negative value of the difference. The parameter change is constructed as a change node. The change nodes are arranged from smallest to largest according to the change position. The difference between adjacent change positions is determined as the adjacent change interval. The change nodes are connected according to the length direction of the material segment to construct the process change topology of the material segment.
[0080] In practice, the material segment process change topology uses the parameter changes identified within a single material segment as change nodes, and connects the change nodes in ascending order of change position to form an ordered structure. Each change node records the parameter type, change direction, and change position, and the connection between adjacent change nodes records the adjacent change interval. This is used to characterize the combination, change order, and spatial interval relationship of fiber bundle tension, process temperature, process pressure, and medium flow rate along the length of the material segment.
[0081] Using parameter type, change direction, connection order and change interval ratio as constraints, the material segment process change topology of adjacent production sections is scaled and matched to determine the process state change marker that maintains topology consistency across sections.
[0082] In practice, for the material segment process change topology formed by the same material segment in adjacent production segments, candidate node chains composed of continuously changing nodes are extracted according to the connection order. Candidate node chains with the same number of nodes, the same corresponding node parameter type, and the same change direction are selected. Each adjacent change interval is divided by the total length of the corresponding candidate node chain to generate a change interval ratio sequence. The difference between the corresponding change interval ratios in two candidate node chains is calculated in turn. When the absolute value of each difference is not greater than the interval ratio matching threshold, it is determined that the two candidate node chains maintain the same connection structure. The corresponding change nodes and their connection relationships are identified as process state change markers that maintain topological consistency across segments.
[0083] In practice, the interval ratio matching threshold is set to 0.1. This means that if the absolute value of the difference between the corresponding change interval ratios of adjacent production sections is no greater than 0.1, the corresponding change intervals are considered matched. The candidate node chain is a set of changing nodes and their connection relationships, continuously extracted from the material segment process change topology according to the node connection order, used to match the corresponding node chains of adjacent production sections. The process state change marker is a corresponding identifier established between two candidate node chains in adjacent production sections that have the same parameter type arrangement, change direction arrangement, connection order, and approximate change interval ratio after stretching and shrinking matching. This marker is not a single parameter change, but is jointly composed of the upstream candidate node chain, the downstream candidate node chain, and the corresponding relationships of each changing node in the two candidate node chains. It is associated with the change position and original acquisition time of each corresponding changing node, used to indicate the correspondence between the same set of process state changes in adjacent production sections. Stretching and shrinking matching, while maintaining consistency in parameter type, change direction, and connection order, identifies the same candidate node chain whose positional scale changes due to the stretching or shrinking of the material segment length by comparing the change interval ratio.
[0084] Based on the occurrence time of the changing node in the same process state change mark in the adjacent production section, construct the node time displacement vector, perform common displacement-relative offset decomposition on the node time displacement vector, and determine the common displacement component as the actual propagation time.
[0085] In specific implementation, when constructing change nodes, the acquisition time of the process parameter data corresponding to the parameter change location is associated with the occurrence time of the change node; according to the correspondence of each change node in the same process state change mark, the occurrence time of each change node in the upstream production section and the adjacent downstream production section is extracted respectively, and the difference between the downstream occurrence time and the corresponding upstream occurrence time is determined as the node time displacement. The node time displacements are arranged according to the node connection order to construct a node time displacement vector; the median value of each node time displacement in the node time displacement vector after sorting from smallest to largest is determined as the initial common displacement component, and the difference between each node time displacement and the initial common displacement component is determined as the relative offset component of the corresponding change node; the node time displacements with the absolute value of the relative offset component greater than the corresponding relative offset threshold are determined as abnormal node time displacements and removed, and the median value of the retained node time displacements after sorting from smallest to largest is determined as the common displacement component, and the common displacement component is determined as the actual propagation time of the process state change mark between adjacent production sections;
[0086] In practice, for each adjacent production section and each parameter type, the 95th percentile value of the relative offset component in the historical stable production batches, after sorting the absolute values from smallest to largest, is set as the relative offset threshold.
[0087] The estimated arrival time of the corresponding material segment to the adjacent downstream production segment is corrected based on the actual propagation time, and the estimated arrival time of the corresponding material segment to the subsequent production segment is updated according to the production segment sequence to generate the corrected whole-process process propagation clock.
[0088] In practice, the original estimated propagation time of the corresponding material segment between adjacent production segments is extracted from the whole process propagation clock. The difference between the actual propagation time and the original estimated propagation time is calculated to generate the propagation time correction. The propagation time correction is then added to the original estimated arrival time of the corresponding material segment to the adjacent downstream production segment and each subsequent production segment to generate the corrected estimated arrival time for each production segment. Based on this, the corrected whole process propagation clock is generated.
[0089] Compared to existing technologies that typically record production parameters according to equipment, acquisition time, or production process, and associate continuous material data with longitudinal position coordinates, the difference in this invention lies not in simply acquiring data or performing length mapping, but in: treating material segments as a unified object, mapping multiple types of process parameters to the length domain of the material segment and constructing a process change topology, and then using parameter type, change direction, connection order, and change interval ratio to perform scaling matching on the change topology of adjacent production segments. Subsequently, abnormal time displacements are eliminated and the actual propagation time is determined by decomposing the common displacement-relative offset of corresponding change nodes to correct the entire process propagation clock. The reason for adopting this method is that during the continuous transport of carbon fiber, there are speed fluctuations, equipment start-ups and shutdowns, and material stretching or contraction. Simply relying on timestamps or fixed transport speeds can easily cause data mismatches between different production segments of the same material segment. However, the change topology can preserve the combination order and relative spatial structure of process changes, so that the same change pattern can still be identified after the position scale changes. This can reduce the local parameter response delay and the interference of abnormal nodes on the propagation time calculation, and improve the accuracy of cross-segment data association of the same material segment, expected arrival time correction, and target acquisition period determination.
[0090] In this embodiment, the generation of the dynamic acquisition strategy specifically includes:
[0091] The target material segment and the target downstream production segment are determined based on the material segment identifier associated with the process state change marker, and a change node phase chain is generated based on the distance of the change position of each change node relative to the starting position of the material segment.
[0092] In practice, based on the material segment identifier associated with the process state change marker, the target material segment is located from the material segment operation data sequence, and the actual length of the target material segment is determined. The candidate node chain located in the downstream production section in the process state change marker is used as the current change node template, and the production section immediately following the downstream production section and not yet reached by the target material segment is determined as the target downstream production section. According to the node connection order in the current change node template, the change position of each change node is extracted, the starting position of the material segment is set as the zero point, the length distance between the change position of each change node and the zero point is calculated, and the length distance is divided by the actual length of the target material segment to obtain the node phase value. The node phase values are arranged according to the node connection order, and the node phase values are associated with the parameter type, change direction, and node identifier of the corresponding change node to generate a change node phase chain.
[0093] The estimated arrival time of the target material segment to the target downstream production section is extracted from the corrected whole-process process propagation clock. The estimated transit time is determined based on the material segment length and fiber bundle running speed. The estimated arrival time of each changing node is determined according to the phase chain of the changing nodes.
[0094] In specific implementation, the estimated arrival time of the target material segment from its starting position to the target downstream production section is extracted from the corrected full-process propagation clock. The fiber bundle running speed corresponding to the formation of the target material segment is extracted from the material segment running data sequence, and the average fiber bundle running speed when the equipment is in operation is calculated to obtain the material segment formation speed. The actual length of the target material segment is divided by the material segment formation speed to obtain the estimated passage time of the target material segment through the target downstream production section. According to the node connection order in the phase chain of the changing nodes, the phase value of each node is multiplied by the estimated passage time to obtain the node time offset of each changing node relative to the starting position of the target material segment. The node time offset is then added to the estimated arrival time of the target material segment to obtain the estimated arrival time of each changing node to the target downstream production section.
[0095] Before the first change node is expected to arrive, the data acquisition channel for the corresponding parameter type is opened. When the parameter type and change direction of the acquired data match the first change node, the actual arrival time of the first change node is locked, and the expected arrival time of the remaining change nodes is corrected.
[0096] In practice, based on the parameter type corresponding to the first change node, the corresponding fiber bundle tension, process temperature, process pressure, or medium flow rate acquisition channel is selected from the data acquisition channels. The estimated arrival time of the first change node is subtracted by 5 seconds to obtain the opening time of the data acquisition channel. The estimated arrival time of the first change node is added by 3 seconds to obtain the predetermined closing time. The time period between the opening time and the predetermined closing time is determined as the acquisition period for the first change node. The difference between adjacent sampled values is calculated according to the acquisition time sequence. When the absolute value of the fiber bundle tension difference reaches 3% of the stable average, the absolute value of the process temperature difference reaches 2℃, and the absolute value of the process pressure difference reaches 3%, the acquisition period is determined. When the value reaches 2% of the stable average or the absolute value of the difference in medium flow reaches 3% of the stable average, it is determined that the corresponding parameter has changed, and the direction of change is determined according to the sign of the difference. When the parameter type and the direction of change match the first change node, the corresponding sampling time is locked as the actual arrival time of the first change node. The actual arrival time of the first change node is subtracted from the expected arrival time of the first change node to obtain the time correction amount. The time correction amount is added to the expected arrival time of the remaining change nodes to obtain the corrected expected arrival time of the remaining change nodes, and a phase-time correspondence relationship is established with the actual arrival time of the first change node as the time reference.
[0097] For each subsequent change node locked, the local phase propagation rate is determined based on the actual arrival interval and phase interval of the adjacent locked change nodes. The estimated arrival time of the unlocked change nodes is recursively corrected, and a dynamic acquisition strategy is generated.
[0098] In specific implementation, starting from the second locked change node, according to the node connection order in the change node phase chain, the actual arrival time of the next locked change node is subtracted from the actual arrival time of the previous locked change node to obtain the actual arrival interval; the node phase value of the next locked change node is subtracted from the node phase value of the previous locked change node to obtain the phase interval; the phase interval is divided by the actual arrival interval to obtain the local phase propagation rate between the corresponding adjacent change nodes; for the unlocked change node, the node phase value of the change node is subtracted from the node phase value of the most recent locked change node to obtain the remaining phase difference; the remaining phase difference is divided by the local phase propagation rate corresponding to the most recent set of adjacent locked change nodes to obtain the remaining propagation time of the change node relative to the most recent locked change node; the remaining propagation time is added to the actual arrival time of the most recent locked change node to obtain the corrected estimated arrival time of the change node; the corrected estimated arrival time is then calculated. The estimated arrival time minus 5 seconds is used as the opening time of the corresponding data acquisition channel. The revised estimated arrival time plus 3 seconds is used as the scheduled closing time. The period between the opening time and the scheduled closing time is defined as the target acquisition period for the corresponding change node. When the change node is locked within the corresponding acquisition period, the actual arrival time is defined as the actual closing time of the data acquisition channel. When a downstream production section of the target experiences a shutdown, the shutdown duration is added to the revised estimated arrival time of the unlocked change node. For data acquisition channels that have not yet been opened, the shutdown duration is added to the opening time and the scheduled closing time of the corresponding channel. For data acquisition channels that have been opened but whose corresponding change nodes have not yet been locked, the lock on the change node is paused, and the shutdown duration is added to the scheduled closing time. The data acquisition channels, target acquisition periods, node connection order, material segment identifiers, and target downstream production sections corresponding to each change node are associated to generate a dynamic acquisition strategy.
[0099] In this embodiment, the generation of automatic data collection results specifically includes:
[0100] According to the dynamic data collection strategy, production data of each production section is collected and the production data is associated with production batch identifier, material section identifier and corresponding production section;
[0101] The production data of the same material segment are associated with the production data of each production segment according to the production segment sequence, and the data collection results are automatically generated.
[0102] An automated data acquisition system for the entire carbon fiber production process includes:
[0103] The data acquisition module is used to establish data acquisition channels for each carbon fiber production section and to collect production batch data, section path length data, fiber bundle operation data, and process parameter data.
[0104] The material segment division module is used to divide the continuously conveyed fiber bundle into multiple material segments based on production batch data and fiber bundle operation data, and generate a material segment operation data sequence.
[0105] The propagation clock generation module is used to determine the estimated arrival time of each material segment to the downstream production segment based on the segment path length data and material segment operation data sequence, and generate the entire process propagation clock.
[0106] The propagation clock correction module is used to map process parameter data to the material segment length domain, identify process state change markers that maintain consistent change topology across segments, determine their actual propagation time between adjacent production segments, and correct the entire process propagation clock.
[0107] The dynamic acquisition strategy generation module is used to recursively correct the expected arrival time of each changing node in the downstream production section of the target material segment based on the corrected whole-process process propagation clock and the node connection order in the phase chain of the changing nodes, and generate a dynamic acquisition strategy.
[0108] The automatic data acquisition result generation module is used to obtain automatic data acquisition results based on dynamic acquisition strategies.
[0109] A computer-readable medium storing a computer program that, when executed by a processor, enables the processor to perform an automatic data acquisition method for the entire carbon fiber production process.
[0110] refer to Figure 2 , Figure 2 The horizontal axis represents the material segment numbers involved in the simulation verification, corresponding to a total of 200 consecutive material segments; the vertical axis represents the production stages that the material segments sequentially pass through: raw fiber drawing, pre-oxidation, low-temperature carbonization, and high-temperature carbonization. Colors indicate the absolute error between the actual arrival time and the expected arrival time of each material segment in its corresponding production stage; the closer the color is to yellow, the larger the arrival error, and the closer the color is to dark purple, the smaller the arrival error. The upper heatmap shows the error distribution before the entire process propagation clock correction, while the lower heatmap shows the error distribution after recursive correction based on the actual propagation time, the actual arrival time of the first changing node, and the local phase propagation rate.
[0111] Before the correction, significant high-error zones were observed around material segments 55, 112, and 168 in the heatmap. As the material segments continued to be transported from the raw yarn drawing section to the pre-oxidation, low-temperature carbonization, and high-temperature carbonization sections, some errors gradually increased. This indicates that the propagation clock established solely based on the segment path length and initial operating speed is prone to error accumulation during continuous production. After the correction, the high-error areas in the heatmap were significantly reduced, and the colors corresponding to each production segment became generally darker. This demonstrates that determining the actual propagation time through cross-segment process change topology matching, combined with dynamic correction of the expected arrival time using the phase chain of changing nodes, can effectively suppress time shifts caused by equipment downtime, changes in operating speed, and differences in drawing between segments.
[0112] Example 1: To verify the feasibility of this invention in practice, it was applied to a continuous production line for precursor fiber drawing, pre-oxidation, low-temperature carbonization, and high-temperature carbonization at a carbon fiber production enterprise in Jiangsu Province. During continuous operation, this production line experienced fluctuations in fiber bundle speed, differences in drawing speed between sections, and short equipment shutdowns. The existing method of directly associating production data with equipment timestamps easily mixed data from adjacent material sections, resulting in significant cross-section time offsets after shutdown recovery. In this example, 72 hours of continuous production data were collected. The fiber bundle running speed ranged from 2.82 m / min to 3.18 m / min, with a cumulative conveying length of 12960 m. Dividing the data into 10 m sections, a total of 1296 material sections were generated. During operation, there were three equipment shutdowns, with a total downtime of 17.6 min.
[0113] Data acquisition channels are established in each production section for fiber bundle running speed, cumulative conveying length, equipment start / stop status, fiber bundle tension, process temperature, process pressure, and medium flow rate, with a sampling frequency set to 10Hz. Production batch identifiers and material segment identifiers are generated based on the production batch and cumulative conveying length. Based on the actual conveying path lengths of 42.5m, 53.4m, and 36.4m between adjacent production sections, combined with the fiber bundle running speed and equipment start / stop status, the estimated arrival time of each material segment to the downstream production section is calculated, generating a full-process process propagation clock. For material segments located in the corresponding conveying path during equipment downtime, the downtime is included in the estimated conveying time.
[0114] The fiber bundle tension, process temperature, process pressure, and medium flow rate collected from each production section were mapped to the material section length domain. Four variable nodes were identified in material section M0864, with node phase values of 0.18, 0.42, 0.67, and 0.86, corresponding to fiber bundle tension, process temperature, process pressure, and medium flow rate, respectively. The time displacements of the four nodes matched in adjacent production sections were 18.4 min, 18.7 min, 20.9 min, and 18.6 min, with an initial common displacement component of 18.65 min. The relative offset components for each node were -0.25 min, 0.05 min, 2.25 min, and -0.05 min, respectively. The relative offset threshold determined from historical stable production batches was 0.6 min; therefore, the abnormal node time displacement corresponding to 20.9 min was removed, and the median value of 18.6 min, after sorting the remaining node time displacements from smallest to largest, was determined as the actual propagation time. The original estimated propagation time in the whole process propagation clock was 17.8 min. The difference between the actual propagation time and the original estimated propagation time was 0.8 min. The 0.8 min was added to the estimated arrival time of material segment M0864 to the adjacent downstream production segment and subsequent production segment.
[0115] The first change node of material segment M0864 was expected to arrive 36.2s after entering the target downstream production section from the starting position of the material segment. The corresponding data acquisition channel opened at 31.2s and was scheduled to close at 39.2s. Actual acquired data showed a change in parameter type and direction at 38.9s. This 38.9s was locked as the actual arrival time of the first change node, yielding a time correction of 2.7s. This time correction was then added to the expected arrival times of the remaining change nodes. The actual arrival time of the second change node was 85.7s. The actual arrival interval between the first two change nodes was 46.8s, with a phase interval of 0.24, resulting in a local phase propagation rate of 0.00513 phase values / s. The remaining phase difference between the third and second change nodes is 0.25, resulting in a remaining propagation time of 48.7s. This 48.7s is then added to the actual arrival time of the second change node to obtain the corrected estimated arrival time of the third change node, which is 134.4s. Based on this, the period from 129.4s to 137.4s is determined as the target acquisition period for the third change node.
[0116] 200 material segments were randomly selected for manual location verification. The original timestamp association method correctly associated 181 material segments, with an accuracy rate of 90.5%. After adopting this invention, 197 material segments were correctly associated, improving the accuracy rate to 98.5%. Verification was performed on 500 manually confirmed process change nodes. The original acquisition method effectively captured 445 nodes, with an effective capture rate of 89.0%. After adopting this invention, 491 nodes were effectively captured, with an effective capture rate of 98.2%. The absolute error of the average arrival time of the first batch of material segments after equipment shutdown decreased from 72.4s to 11.6s. Fiber bundle tension, process temperature, process pressure, and medium flow rate in the four production sections were continuously collected at 10Hz for 72 hours, with a theoretical data record volume of 41,472,000 records. According to system backend statistics, after adopting the dynamic acquisition strategy, 13,686,000 valid records were actually written during each target acquisition period, reducing the data volume by approximately 67.0%. Based on the target acquisition period in the dynamic acquisition strategy, corresponding production data is filtered. The filtered production data is then associated with production batch identifiers, material segment identifiers, and corresponding production segments. Production data corresponding to the same material segment is then aggregated according to the production segment sequence to generate automatic data acquisition results. This demonstrates that the present invention can reduce the writing of invalid production data, improve the capture rate of process changes, and enhance the accuracy of cross-segment data association for the same material segment.
[0117] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A method for automatic data acquisition throughout the entire carbon fiber production process, characterized in that, Includes the following steps: Establish data acquisition channels for each carbon fiber production section and collect production batch data, section path length data, fiber bundle operation data, and process parameter data; Based on production batch data and fiber bundle operation data, the continuously conveyed fiber bundle is divided into multiple material segments, and a material segment operation data sequence is generated. Based on the process path length data and material segment operation data sequence, the estimated arrival time of each material segment to the downstream production section is determined, and the whole process propagation clock is generated. The process parameter data is mapped to the material segment length domain, the process state change markers that maintain the same change topology across segments are identified, the actual propagation time between adjacent production segments is determined, and the process propagation clock of the entire process is corrected. Based on the corrected whole-process propagation clock and the node connection order in the phase chain of the changing nodes, the expected arrival time of each changing node of the target material segment in the downstream production section of the target is recursively corrected, and a dynamic acquisition strategy is generated. Based on the dynamic acquisition strategy, the data acquisition results are obtained automatically.
2. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The fiber bundle operation data includes fiber bundle operating speed, cumulative conveying length, and equipment start / stop status.
3. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The process parameter data includes fiber bundle tension, process temperature, process pressure, and medium flow rate.
4. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The generation of the material segment operation data sequence specifically includes: The current production batch is determined based on the production batch data, and a corresponding production batch identifier is generated; Based on the cumulative conveying length in the fiber bundle operation data, the position where the cumulative conveying length reaches the preset material segment length is used as the material segment boundary. The continuously conveyed fiber bundle corresponding to the current production batch is divided into multiple material segments, and material segment identifiers are generated according to the conveying order of the material segments. Extract the fiber bundle operation data corresponding to each material segment, associate the fiber bundle operation data with the production batch identifier and material segment identifier, and organize them to generate a material segment operation data sequence.
5. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The generation of the entire process propagation clock specifically includes: Based on the section path length data, determine the conveying path length of each material segment from the current production section to each downstream production section; Extract the time when each material segment enters the current production section, the fiber bundle running speed, and the equipment start-stop status from the material segment operation data sequence. Determine the expected conveying time for each material segment based on the conveying path length, fiber bundle running speed, and equipment start-stop status. Based on the time when each material segment enters the current production section and the estimated transport time, the estimated arrival time of each material segment to each downstream production section is determined, and the estimated arrival time is organized according to the material segment identification and the production section sequence to generate the whole process propagation clock.
6. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The revised full-process propagation clock specifically includes: Based on the material segment identifier and cumulative conveying length, the process parameter data is mapped from the time domain to the material segment length domain to generate a multi-parameter length sequence; Extract the parameter type, change direction, change location and adjacent change interval from the multi-parameter length sequence, and use the parameter changes as change nodes to construct the material segment process change topology; Using parameter type, change direction, connection order and change interval ratio as constraints, the material segment process change topology of adjacent production sections is scaled and matched to determine the process state change marker that maintains topology consistency across sections. Based on the occurrence time of the changing node in the same process state change mark in the adjacent production section, construct the node time displacement vector, perform common displacement-relative offset decomposition on the node time displacement vector, and determine the common displacement component as the actual propagation time. The estimated arrival time of the corresponding material segment to the adjacent downstream production segment is corrected based on the actual propagation time, and the estimated arrival time of the corresponding material segment to the subsequent production segment is updated according to the production segment sequence, generating the corrected whole-process process propagation clock.
7. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The generation of the dynamic acquisition strategy specifically includes: The target material segment and the target downstream production segment are determined based on the material segment identifier associated with the process state change marker, and a change node phase chain is generated based on the distance of the change position of each change node relative to the starting position of the material segment. The estimated arrival time of the target material segment to the target downstream production section is extracted from the corrected whole-process process propagation clock. The estimated transit time is determined based on the material segment length and fiber bundle running speed. The estimated arrival time of each changing node is determined according to the phase chain of the changing nodes. Before the first change node is expected to arrive, the data acquisition channel for the corresponding parameter type is opened. When the parameter type and change direction of the acquired data match the first change node, the actual arrival time of the first change node is locked, and the expected arrival time of the remaining change nodes is corrected. For each subsequent change node locked, the local phase propagation rate is determined based on the actual arrival interval and phase interval of the adjacent locked change nodes. The estimated arrival time of the unlocked change nodes is then recursively corrected, and a dynamic acquisition strategy is generated.
8. The automatic data acquisition method for the entire carbon fiber production process according to claim 1, characterized in that, The generation of the automatic data collection results specifically includes: According to the dynamic data collection strategy, production data of each production section is collected and the production data is associated with production batch identifier, material section identifier and corresponding production section; The production data of the same material segment are associated with the production data of each production segment according to the production segment sequence, and the data collection results are automatically generated.
9. An automatic data acquisition system for the entire carbon fiber production process, comprising the automatic data acquisition method for the entire carbon fiber production process as described in any one of claims 1 to 8, characterized in that, include: The data acquisition module is used to establish data acquisition channels for each carbon fiber production section and to collect production batch data, section path length data, fiber bundle operation data, and process parameter data. The material segment division module is used to divide the continuously conveyed fiber bundle into multiple material segments based on production batch data and fiber bundle operation data, and generate a material segment operation data sequence. The propagation clock generation module is used to determine the estimated arrival time of each material segment to the downstream production segment based on the segment path length data and material segment operation data sequence, and generate the entire process propagation clock. The propagation clock correction module is used to map process parameter data to the material segment length domain, identify process state change markers that maintain consistent change topology across segments, determine their actual propagation time between adjacent production segments, and correct the entire process propagation clock. The dynamic acquisition strategy generation module is used to recursively correct the expected arrival time of each changing node in the downstream production section of the target material segment based on the corrected whole-process process propagation clock and the node connection order in the phase chain of the changing nodes, and generate a dynamic acquisition strategy. The automatic data acquisition result generation module is used to obtain automatic data acquisition results based on dynamic acquisition strategies.
10. A computer-readable medium, characterized in that, The computer-readable medium stores a computer program that, when executed by a processor, enables the processor to perform the automatic data acquisition method for the entire carbon fiber production process as described in any one of claims 1 to 8.
