A combing cotton web quality abnormality early warning system based on multi-source data fusion

Abstract

The application discloses a combing cotton web quality abnormality early warning system based on multi-source data fusion, and relates to the technical field of intelligent detection and industrial control of textile machinery.The combing cotton web section record is generated by using the production time identifier, the combing head identifier and the carding phase data, and the combing cotton web apparent data, the equipment state data, the environmental disturbance data and the output quality data corresponding to the delay matching relationship are written into the same combing cotton web section record, so that the output end quality change can be corresponded to the combing head source, the carding phase and the combing cotton web section before the gathering into slivers;the current combing cotton web section record is compared through the phase quality portrait, and the quality abnormality early warning record is generated when the apparent abnormality mark and the remaining multi-source data mark meet the same abnormality type, the consistent abnormality direction and the position relationship conditions, so that the early warning result can be bound to the combing head, the carding phase and the combing cotton web section where the abnormality occurs.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection and industrial control technology for textile machinery, and in particular to an early warning system for abnormal quality of combed cotton web based on multi-source data fusion. Background Technology

[0002] A combing machine consists of multiple combing heads operating in parallel. Each combing head forms a corresponding combed cotton web before combing, separating, joining, and converging into a sliver. The combed cotton web is formed by the continuous overlapping of fiber bundles in each nipping cycle. Its width, edge continuity, joining texture, and local looseness affect the uniformity and continuity of the subsequent combed sliver.

[0003] Currently, the monitoring of abnormal quality in combed cotton webs mostly employs methods such as output combed sliver quality monitoring, combed cotton web image detection, web breakage or head breakage detection, and equipment operation status monitoring. Output combed sliver quality monitoring reflects the quality result after multiple combing heads converge; combed cotton web image detection mainly reflects the apparent state at the acquisition location; web breakage or head breakage detection mainly reflects whether there is an interruption; and equipment operation status monitoring mainly reflects operating conditions such as machine speed, load, and suction pressure.

[0004] Under conditions of continuous operation of multiple combing heads, batch and roll changes, speed adjustments, or fluctuating suction pressure, a single combing head may exhibit thin web, missing cotton at the edges, abnormal seam marks, or localized loose combed web at a specific clamping phase. Since the combed webs formed by each combing head subsequently enter a common convergence and conveying path, there is a conveying delay and merging relationship between the change in the output combed sliver quality and the location of the abnormality. Therefore, it is difficult for the output alarm to directly correspond to the combing head, clamping phase, and combed web segment that caused the abnormality.

[0005] Meanwhile, the apparent data, output quality data, equipment status data, and environmental disturbance data of the combed cotton web typically come from different acquisition locations and sampling times. If various types of data lack a unified record relationship associated with the combing head, clamp phase, and combed cotton web segment, abnormal information is likely to manifest as discrete alarm points, making it difficult to distinguish between true abnormalities in the combed cotton web quality, fluctuations in equipment status, and short-term responses caused by environmental disturbances.

[0006] Therefore, how to correlate the quality anomalies of combed cotton web with the corresponding combing head, nipper phase, and combed cotton web segment during continuous production of multiple combing heads, and to form an early warning result based on the consistency of multi-source data, has become a technical problem that needs to be solved. Summary of the Invention

[0007] This application provides an early warning system for quality anomalies in combed cotton webs based on multi-source data fusion, which solves the problem of difficulty in synchronously locating and classifying early warnings of anomalies in combed cotton webs with multiple combing heads.

[0008] This invention provides an early warning system for quality anomalies in combed cotton webs based on multi-source data fusion, comprising:

[0009] Industrial control computing unit, data interface unit, data storage unit, and early warning output unit;

[0010] The data interface unit receives multi-source production data including production time identifier, combing head identifier, clamp phase data, combed cotton web appearance data, output quality data, equipment status data, and environmental disturbance data;

[0011] The industrial control computing unit includes a combed cotton web segment synchronization module, a phase quality profiling module, and a multi-source consistency fusion module;

[0012] The combed cotton web segment synchronization module generates combed cotton web segment identifiers and establishes combed cotton web segment records according to the combing head identifier, clamp phase data and production time identifier. It writes the combed cotton web appearance data, equipment status data, environmental disturbance data and output quality data corresponding to the delay matching relationship into the combed cotton web segment records and stores them in the data storage unit.

[0013] The phase quality profiling module retrieves historical combed cotton web segment records from the data storage unit to generate a phase quality profile.

[0014] The multi-source consistency fusion module compares the combed cotton web segment records with the phase quality profile to generate multi-source data tags;

[0015] The multi-source data tagging includes apparent anomaly tags, output quality response tags, equipment status response tags, and environmental disturbance tags. Each tag contains anomaly type and anomaly direction. When the apparent anomaly tag corresponds to the same anomaly type and anomaly direction as at least one of the output quality response tag, equipment status response tag, and environmental disturbance tag, and satisfies one of the relationships of the same combed cotton web segment or adjacent delayed combed cotton web segments, the multi-source consistency fusion module generates a quality anomaly early warning record.

[0016] The early warning output unit outputs the quality anomaly early warning record.

[0017] In some embodiments, the combed cotton web is a fibrous web formed by the combing heads of a combing machine before they are combed, separated, joined, and aggregated into slivers;

[0018] The combed cotton web segment is a continuous combed cotton web segment formed by the same combing head in one nipper phase segment;

[0019] The combed cotton web segment record includes a combed cotton web segment identifier, a combing head identifier, a clamping phase identifier, a production time identifier, a production batch identifier, a process variety identifier, a data source timestamp, a data validity status, and a profile version identifier. The data validity status includes at least one of the following: synchronized status, missing sampling status, delayed matching status, and verification lock status.

[0020] In some embodiments, the clamp phase data includes at least two of the following: main shaft phase signal, clamp opening and closing stage identifier, separation and engagement stage identifier, top comb action stage identifier, and combed web transfer stage identifier;

[0021] The combed cotton web segment synchronization module divides continuous production time markers into phase segments, binds the phase segments to the combing head identifier that generates the phase segment, and generates combed cotton web segment identifiers according to the start and end time of the phase segment, the phase sequence number, and the combing head identifier.

[0022] When the same combing head continuously outputs multiple phase segments, the combed cotton web segment synchronization module maintains the connection relationship between adjacent combed cotton web segments according to the phase sequence number.

[0023] In some embodiments, the apparent data of the combed cotton web is obtained by feature processing of the image acquisition data of the combed cotton web, and the image acquisition data of the combed cotton web includes single-frame image features and continuous frame change features;

[0024] The apparent data of the combed cotton web includes at least three of the following: width status, edge continuity status, local light and dark uniformity status, joint texture status, web surface looseness status, and visible foreign matter status.

[0025] The industrial control computing unit converts the apparent data of the combed cotton web into status tags, and writes the status tags, acquisition location identifiers, and image timestamps into the corresponding combed cotton web segment records.

[0026] In some embodiments, the output quality data includes at least one of combing strip quantity fluctuation state, stripe fluctuation state, short segment unevenness state, and quality spectrum event state;

[0027] The combed cotton web segment synchronization module establishes a delay matching relationship based on the conveying path length of the combed cotton web from the corresponding combing head to the output quality inspection position, the speed status in the equipment status data, the production cycle time, and the output quality inspection time window.

[0028] The delay matching relationship includes a start delay and an end delay. The start delay represents the time offset corresponding to the arrival of the start boundary of the combed cotton web segment at the output quality detection position, and the end delay represents the time offset corresponding to the arrival of the end boundary of the combed cotton web segment at the output quality detection position.

[0029] The combed cotton web segment synchronization module reverses the output quality detection time window to the combed cotton web formation time window based on the start delay and the end delay, and matches the output quality data to the combed cotton web segment record that occurred earlier.

[0030] In some embodiments, the delayed matching relationship includes a cross-segment matching ratio, which represents the coverage relationship between the same output quality detection time window after delayed back-calculation and multiple preceding combed cotton web segment records;

[0031] When the combed cotton web formation time window obtained by reverse calculation corresponds to one combed cotton web segment record, the combed cotton web segment synchronization module will output quality data and write it into the combed cotton web segment record; when the combed cotton web formation time window obtained by reverse calculation corresponds to multiple combed cotton web segment records, the combed cotton web segment synchronization module will output quality data and write it into the corresponding combed cotton web segment records respectively, and synchronously write the corresponding cross-segment matching ratio.

[0032] In some embodiments, the equipment status data includes at least two of the following: vehicle speed status, separation roller speed status, drafting roller speed status, suction pressure status, main motor load status, and roll changing status.

[0033] The environmental disturbance data includes at least one of temperature and humidity status, fly ash concentration status, and lighting intensity status;

[0034] The industrial control computing unit writes the equipment status data and environmental disturbance data into the corresponding combed cotton web segment record according to the production time identifier, and writes the data into the combed cotton web segment record corresponding to the phase segment that coincides with the sampling time when the data sampling time falls into the phase segment boundary.

[0035] In some embodiments, the phase quality profiling module groups historical combed cotton web segment records according to at least one of production batch, process type, speed range, operating stage and suction pressure state, and establishes a phase quality profile corresponding to the combing head-nip phase combination in each group;

[0036] The phase quality profile includes an apparent state benchmark, an output quality response benchmark, an equipment state response benchmark, and an environmental disturbance benchmark. Each benchmark includes a benchmark state, an allowable fluctuation range, a frequency of occurrence, a number of consecutive occurrences, and an adjacent phase correlation state.

[0037] In some embodiments, the phase quality profiling module reads the warning result identifiers recorded in the historical combed cotton web segment when updating the phase quality profiling;

[0038] When a historical combed cotton web segment record is already bound to a quality anomaly warning record, the phase quality profiling module writes the historical combed cotton web segment record into the anomaly sample set, or excludes the historical combed cotton web segment record when calculating the apparent state benchmark, output quality response benchmark, equipment state response benchmark, and environmental disturbance benchmark.

[0039] The abnormal sample set is bound and stored with the corresponding abnormal type, comb head identifier, clamp phase identifier, trigger time, and handling result identifier.

[0040] In some embodiments, the quality anomaly early warning record includes a combed cotton web segment identifier, a combing head identifier, a clamp phase identifier, anomaly type, early warning level, trigger time, and multi-source data markers that participate in generating the early warning, wherein the multi-source data markers that participate in generating the early warning include a marker level;

[0041] The anomaly types include at least two of the following: thin web risk, edge cotton shortage risk, seam mark anomaly risk, periodic fluctuation risk, air suction disturbance risk, and separation seam anomaly risk.

[0042] The early warning output unit writes the quality anomaly early warning record into the quality traceability database and sends a process handling instruction to the combing machine control system or the workshop production execution system. The process handling instruction includes at least one of the following: reviewing the combing head, checking the suction pressure status, checking the environmental disturbance data, marking the combed cotton web segment to be traced, and retrieving the corresponding historical combed cotton web segment record.

[0043] The warning level is determined based on the number of multi-source data markers, the marker level, the number of consecutive occurrences, and the response status of adjacent delayed combed cotton web segments within the corresponding combed cotton web segment.

[0044] Through the above technical solution, the present invention can achieve at least the following beneficial effects:

[0045] This invention uses production time identifiers, combing head identifiers, and clamping phase data to generate combed cotton web segment records. It also writes combed cotton web appearance data, equipment status data, environmental disturbance data, and output quality data corresponding to delay matching relationships into the same combed cotton web segment record. This allows changes in output quality to establish a correspondence with the source of the combing head, clamping phase, and combed cotton web segment before sliver formation. The current combed cotton web segment record is compared using a phase quality profile. When the appearance anomaly marker and other multi-source data markers meet the conditions of the same anomaly type, consistent anomaly direction, and positional relationship, a quality anomaly warning record is generated. This allows the warning result to be linked to the combing head, clamping phase, and combed cotton web segment that caused the anomaly.

[0046] By defining the combed cotton web segment as a continuous combed cotton web section formed within a single phase segment of the same combing head, and storing the combed cotton web segment identifier, phase identifier, production time identifier, and data validity status in the combed cotton web segment record, a verifiable synchronization relationship is formed between multi-source production data within the same record.

[0047] By dividing the clamping phase data into phase segments and generating combed cotton web segment identifiers according to the start and end times of the phase segments, phase sequence numbers, and combing head identifiers, the combed cotton web segments continuously output by the same combing head maintain a front-to-back connection, making it easier to identify abnormal states that recur with each clamping cycle.

[0048] By back-calculating the output quality detection time window to the combed cotton web formation time window according to the start and end delays, and writing the cross-segment matching ratio when spanning multiple combed cotton web segment records, the merged output quality data can be allocated to one or more combed cotton web segment records that occurred earlier.

[0049] By grouping historical combed cotton web segment records according to production batch, process type, speed range, operating stage, and suction pressure status, and establishing a phase quality profile corresponding to the combing head-nip phase combination, the current combed cotton web segment records can be compared with historical benchmarks under similar production conditions.

[0050] By writing historical combed cotton web segment records that have been bound to quality anomaly warning records into the abnormal sample set when updating the profile, or excluding such historical combed cotton web segment records when calculating the benchmark, the benchmark data and abnormal samples of the phase quality profile are saved separately, reducing the interference of abnormal samples on the subsequent comparison benchmark.

[0051] By binding and saving quality anomaly warning records with combed cotton web segment identifiers, combing head identifiers, nip phase identifiers, anomaly types, warning levels, and multi-source data tags, the warning output can correspond to specific combing heads, specific nip phases, and specific combed cotton web segments. Attached Figure Description

[0052] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0053] Figure 1 This is a framework diagram of the early warning system for abnormal quality of combed cotton web based on multi-source data fusion in the embodiment. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0055] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0056] To facilitate understanding, the relevant terms and concepts involved in the embodiments of this application will be introduced below:

[0057] Combed cotton web refers to the fibrous web formed by the individual combing heads of a combing machine after the cotton lap is fed in, combed, separated, and joined together, and before it is fused into a sliver. A combed cotton web segment refers to a continuous section of combed cotton web formed within a single nib phase segment by the same combing head.

[0058] Combed cotton web segment records refer to data records that store multi-source production data, synchronization status, and early warning results using combed cotton web segment identifiers as indexes. Nip-phase data refers to data characterizing the stage of the nipper, cylinder, top comb, separation and joining, and combed cotton web transfer within one working cycle of the comber.

[0059] Phase quality profile refers to the baseline data set formed by aggregating historical combed cotton web segment records according to combing head identification and clamp phase data. Multi-source data labeling refers to the state-based judgment results generated from combed cotton web appearance data, output quality data, equipment status data, and environmental disturbance data. Output quality data refers to the combed sliver quality status data collected from the output quality detection location, which is the location on the conveying path where the quality detection device is located after the combed cotton web is assembled into a sliver. The output quality detection time window refers to the time interval within which the output quality detection location continuously collects the combed sliver quality status data and forms a set of output quality data; its start and end times are recorded using production time identifiers. Quality spectrum event status refers to the spectral event labels in the output quality data corresponding to periodic fluctuations or short-segment unevenness; these labels include event type, event time window, and event level.

[0060] The production time identifier refers to the unified time record received and allocated to production data by the industrial control computing unit. Combing head identifiers, clamping phase data, combing web appearance data, equipment status data, environmental disturbance data, and output quality data are all aligned according to the production time identifier. The clamping phase identifier refers to the stage number obtained from parsing the clamping phase data, used to distinguish different phase segments within the same clamping cycle. Adjacent delayed combing web segments refer to combing web segments determined by delay matching relationships that have a sequential relationship with the current combing web segment in the output quality detection time window; the adjacent delayed combing web segment identifier set refers to the data set storing the above combing web segment identifiers.

[0061] A clamp phase segment refers to a time interval divided according to the clamp phase data within the same clamp cycle. The same combed cotton web segment relationship refers to the relationship where two multi-source data markers correspond to the same combed cotton web segment identifier. Marker position relationship refers to the relationship between two multi-source data markers that satisfy the same combed cotton web segment relationship or the adjacent delayed combed cotton web segment relationship.

[0062] Example 1:

[0063] like Figure 1 As shown, this embodiment employs a combed cotton web quality anomaly early warning system based on multi-source data fusion, including an industrial control computing unit, a data interface unit, a data storage unit, and an early warning output unit;

[0064] The data interface unit receives multi-source production data, including production time identifier, combing head identifier, clamp phase data, combed cotton web appearance data, output quality data, equipment status data, and environmental disturbance data.

[0065] The industrial control computing unit includes a combed cotton web segment synchronization module, a phase quality profiling module, and a multi-source consistency fusion module;

[0066] The combed cotton web segment synchronization module generates a combed cotton web segment identifier based on the combing head identifier, clamp phase data and production time identifier, establishes a combed cotton web segment record, and writes the combed cotton web appearance data, equipment status data, environmental disturbance data and output quality data corresponding to the delay matching relationship into the combed cotton web segment record and then stores it in the data storage unit.

[0067] The phase quality profiling module retrieves historical combed cotton web segment records from the data storage unit based on the combing head identifier and clamp phase data to generate a phase quality profile.

[0068] The multi-source consistency fusion module compares the current combed cotton web segment record with the corresponding phase quality profile to generate multi-source data tags. These tags include apparent anomaly tags, output quality response tags, equipment status response tags, and environmental disturbance tags. Each tag includes anomaly type, anomaly direction, data source timestamp, and tag level. When an apparent anomaly tag forms a tag position relationship with at least one of the output quality response tag, equipment status response tag, and environmental disturbance tag, and this relationship is either the same combed cotton web segment relationship or an adjacent combed cotton web segment relationship determined by a delayed matching relationship, and both correspond to the same anomaly type and have the same anomaly direction, the multi-source consistency fusion module generates a quality anomaly warning record.

[0069] The early warning output unit outputs a quality anomaly warning record.

[0070] In one implementation, the consistent triggering condition is jointly determined by the anomaly type, anomaly direction, data source timestamp, and tag level. Apparent anomaly tags are generated by comparing combed web appearance data with an apparent state baseline; output quality response tags are generated by comparing output quality data with an output quality response baseline; equipment status response tags are generated by comparing equipment status data with an equipment status response baseline; and environmental disturbance tags are generated by comparing environmental disturbance data with an environmental disturbance baseline. Anomaly direction is used to characterize the direction in which data is higher, lower, stronger, weaker, interrupted, or missing relative to the corresponding baseline. Thin web risk corresponds to narrower web width, lighter local brightness, and lower combed sliver basis weight; edge cotton shortage risk corresponds to interrupted edge continuity and missing web width boundaries; seam mark anomaly risk corresponds to seam texture position offset and abnormal continuous frame texture changes; suction disturbance risk corresponds to higher or lower suction pressure and increased web looseness.

[0071] In one embodiment, the combed cotton web is a fibrous web formed by the combing heads of the combing machine before they comb, separate, join, and converge into slivers; the combed cotton web segment is a continuous combed cotton web segment formed by the same combing head within a single nipping phase segment; the combed cotton web segment record includes a combed cotton web segment identifier, a combing head identifier, a nipping phase identifier, a production time identifier, a production batch identifier, a process variety identifier, a data source timestamp, a data validity status, and a profile version identifier, wherein the data validity status includes at least one of the following: synchronized status, missing sampling status, delayed matching status, and verification lock status.

[0072] The combed cotton web segment record is stored using the combed cotton web segment identifier as the primary index. The combed cotton web segment identifier is generated by combining the combing head identifier, clamp phase identifier, and production time identifier, where the production time identifier uses unified clock data received by the industrial control computing unit. The data source timestamp records the time when various types of data enter the data interface unit, and the data validity status records the available status of the corresponding data during the synchronization process. A synchronized status indicates that the corresponding data has been written to the combed cotton web segment record matching the acquisition time; a missing sampling status indicates that the specified data was not obtained within the corresponding acquisition window; a delay matching status indicates that the output quality data has been bound to the earlier combed cotton web segment record according to the delivery delay; and a verification locked status indicates that the corresponding combed cotton web segment record has entered the manual or system verification process.

[0073] The profile version identifier refers to the version record generated or updated during the creation of the phase quality profile, used to ensure that the combed cotton web segment record, phase quality profile, and early warning result maintain the same baseline source. The equipment attention record refers to the status record saved when equipment status response markers appear consecutively but no quality anomaly early warning record is generated. The early warning result identifier is a field recording whether the combed cotton web segment record has been bound to a quality anomaly early warning record, whether it has been bound to an equipment attention record, or whether it is in a review locked state. The handling result identifier is a field recording the execution status of the handling instruction during the process, and the review result identifier is a field recording the conclusion of manual or system review.

[0074] In one embodiment, the clamp phase data includes at least two of the following: spindle phase signal, clamp opening and closing stage identifier, separation and engagement stage identifier, top combing action stage identifier, and combed web transfer stage identifier; the combed web segment synchronization module divides the continuous production time identifier into phase segments, binds the phase segment to the combing head identifier that generates the phase segment, and generates combed web segment identifiers according to the start and end time of the phase segment, the phase sequence number, and the combing head identifier; when the same combing head continuously outputs multiple phase segments, the combed web segment synchronization module maintains the sequential connection relationship of adjacent combed web segment records according to the phase sequence number.

[0075] The phase segment of each clamping cycle is determined by the spindle phase signal and the separation / engagement stage identifier. The industrial control computing unit determines the start and end times of a clamping cycle based on the spindle phase signal, and determines the phase segments within that cycle based on the clamp opening / closing stage identifier, separation / engagement stage identifier, top combing action stage identifier, and combed web transfer stage identifier. The start and end times, phase sequence numbers, and combing head identifiers of the phase segments are written into the same combed web segment record. The sequential connection relationship between adjacent combed web segment records includes the identifier of the previous combed web segment, the identifier of the next combed web segment, and the adjacent phase sequence number, used to record the order of combed web segments continuously formed by the same combing head.

[0076] In one embodiment, the apparent data of the combed cotton web is obtained by feature processing of the image acquisition data of the combed cotton web, which includes single-frame image features and continuous frame variation features; the apparent data of the combed cotton web includes at least three of the following: width status, edge continuity status, local brightness uniformity status, joint texture status, loose web surface status, and visible foreign matter status; the industrial control computing unit converts the apparent data of the combed cotton web into status labels and writes the status labels, acquisition location identifiers, and image timestamps into the corresponding combed cotton web segment records.

[0077] Image acquisition data of the combed cotton web is obtained by an image acquisition device located in the output area of ​​the combing head or the conveying area before sliver formation. Single-frame image features include width boundaries, discontinuous edge regions, local grayscale distribution, bonding texture regions, and visible foreign object regions. Continuous frame variation features include continuous width variation, continuous edge sway variation, continuous variation of bright and dark areas, and continuous variation of bonding texture position. The industrial control computing unit converts these features into width status, edge continuity status, local brightness uniformity status, bonding texture status, web looseness status, and visible foreign object status, and writes these status tags into the combed cotton web segment record. The status tag includes status category, status direction, status level, image timestamp, and acquisition location identifier.

[0078] In one embodiment, the equipment status data includes at least two of the following: machine speed status, separating roller speed status, drafting roller speed status, suction pressure status, main motor load status, and roll changing status; the environmental disturbance data includes at least one of the following: temperature and humidity status, fly waste concentration status, and lighting intensity status; the industrial control computing unit writes the equipment status data and environmental disturbance data into the corresponding combed cotton web segment record according to the production time identifier, and when the data sampling time falls into the phase segment boundary, it writes the data into the combed cotton web segment record corresponding to the phase segment that coincides with the sampling time.

[0079] In one implementation, the phase quality profiling module groups historical combed cotton web segment records according to at least one of production batch, process type, speed range, operating stage, and suction pressure state, and establishes a phase quality profile corresponding to the combing head-nip phase combination within each group; the phase quality profile includes an apparent state benchmark, an output quality response benchmark, an equipment state response benchmark, and an environmental disturbance benchmark, and each benchmark includes a benchmark state, an allowable fluctuation range, a frequency of occurrence, a number of consecutive occurrences, and an adjacent phase correlation state.

[0080] The data source for generating the phase quality profile is historical combed cotton web segment records. The phase quality profile is grouped according to production batch, process type, speed range, operating stage, and suction pressure status. Within each group, a baseline data set is established based on the combing head identifier and the nip phase identifier. The apparent state baseline records the common states of the combed cotton web appearance data under the corresponding combing head and nip phase; the output quality response baseline records the output quality data state of the corresponding combed cotton web segment after delayed matching; the equipment status response baseline records the equipment operating state under the corresponding production state; and the environmental disturbance baseline records the environmental change state under the corresponding production state. The phase quality profile is updated with newly added historical combed cotton web segment records, retaining the profile version identifier and update time during updates.

[0081] Speed ​​range refers to the operating speed range defined by the speed status in the equipment status data. Operating phase refers to one of the following: batch start-up phase, stable operation phase, roll change phase, or shutdown recovery phase. Baseline status refers to the status label or combination that appears most frequently in historical combed cotton web segment records within the same group. Allowable fluctuation range refers to the normal sample distribution range of historical combed cotton web segment records within the same group. Frequency of occurrence refers to the number of times the same type of status label appears in the same group. Consecutive occurrence count refers to the number of times the same type of status label appears consecutively in adjacent combed cotton web segment records. Adjacent phase correlation status refers to the succession relationship of status labels between preceding and following phase segments under the same combing head identifier.

[0082] In one implementation, the phase quality profiling module reads the warning result identifier of the historical combed cotton web segment record when updating the phase quality profile; when the historical combed cotton web segment record is already bound to the quality anomaly warning record, the phase quality profiling module writes the historical combed cotton web segment record into the anomaly sample set, or excludes the historical combed cotton web segment record when calculating the apparent state benchmark, output quality response benchmark, equipment state response benchmark, and environmental disturbance benchmark; the anomaly sample set is bound and saved with the corresponding anomaly type, combing head identifier, clamp phase identifier, trigger time, and handling result identifier.

[0083] The abnormal sample set is used to store historical combed cotton web segment records for which quality anomaly warnings have been generated. The abnormal sample set includes the combed cotton web segment identifier, combing head identifier, clamping phase identifier, anomaly type, warning level, trigger time, multi-source data marker, handling result identifier, and review result identifier. The phase quality profile module reads the warning result identifier and review result identifier when updating the phase quality profile. When a historical combed cotton web segment record is confirmed as an abnormal sample, that historical combed cotton web segment record is included in the abnormal sample set storage; when a historical combed cotton web segment record is confirmed as a normal sample, that historical combed cotton web segment record participates in the corresponding phase quality profile update.

[0084] In one implementation, the quality anomaly early warning record includes a combed cotton web segment identifier, a combing head identifier, a clamping phase identifier, anomaly type, early warning level, trigger time, and multi-source data markers involved in generating the early warning; the anomaly type includes at least two of the following: thin web risk, edge cotton shortage risk, seam mark anomaly risk, periodic fluctuation risk, suction disturbance risk, and separation / seaming anomaly risk; the early warning output unit writes the quality anomaly early warning record into the quality traceability database and sends process handling instructions to the combing machine control system or workshop production execution system. The process handling instructions include at least one of the following: reviewing the combing head, checking the suction pressure status, checking environmental disturbance data, marking the combed cotton web segment to be traced, and retrieving the corresponding historical combed cotton web segment record; the early warning level is determined based on the number of multi-source data markers, the marker level, the number of consecutive occurrences, and the response status of adjacent delayed combed cotton web segments within the corresponding combed cotton web segment.

[0085] The quality traceability database stores the following identifiers: combed cotton web segment identifier, combing head identifier, nip phase identifier, production batch identifier, process variety identifier, trigger time, anomaly type, warning level, multi-source data marker, process handling instruction, handling result identifier, and verification result identifier. The warning output unit sends process handling instructions to the combing machine control system or workshop production execution system based on the warning level. These instructions are linked to the combed cotton web segment identifier, combing head identifier, and nip phase identifier for subsequent quality traceability, allowing for the location of the corresponding combing head, nip phase, and combed cotton web segment.

[0086] The label level refers to the discrete intensity field of the multi-source data labels, with values ​​of 1, 2, and 3. A higher value indicates a greater deviation of the corresponding data from the phase quality profile. The warning level refers to the output level of the quality anomaly warning record, determined by the number of multi-source data labels involved in generating the warning, the label level, the number of consecutive occurrences, and the response status of adjacent delayed combed cotton web segments. When the number of multi-source data labels involved in generating the warning increases, the label level rises, or the number of consecutive occurrences increases within the same combed cotton web segment, the warning level is updated according to the level rules under the corresponding profile version identifier.

[0087] In an optional implementation of Example 1, when determining that the anomaly direction is consistent, the multi-source consistency fusion module sets an anomaly direction dictionary for each anomaly type. The anomaly direction dictionary is used to convert the original anomaly directions generated by different data sources into normalized direction codes under the same anomaly type.

[0088] The anomaly direction dictionary includes anomaly type, original anomaly direction set, standard direction code, process type identifier, vehicle speed range, data source location identifier, and profile version identifier. Original anomaly direction refers to the state of multi-source data labels relative to the corresponding baseline, such as being too high, too low, enhanced, weakened, interrupted, missing, loosely enhanced, or having discontinuous texture. Normalized direction code refers to the standard direction result output by the anomaly direction dictionary; when the original anomaly direction is missing, falls into more than two direction categories simultaneously, or the corresponding anomaly direction dictionary cannot be obtained, the normalized direction code is empty.

[0089] The anomaly direction dictionary is invoked according to the process type identifier, speed range, data source location identifier, and profile version identifier. The data source location identifier includes the collection location identifier for combed cotton web appearance data, the detection location identifier for output quality data, the sampling point identifier for equipment status data, or the sampling point identifier for environmental disturbance data. Within the same production batch, the dictionary content remains unchanged when the process type identifier, speed range, data source location identifier, and profile version identifier remain unchanged. When the production batch, process type, speed range, data source location identifier, or profile version identifier changes, the corresponding anomaly direction dictionary is invoked again.

[0090] When a data source timestamp of a multi-source data tag corresponds to a time window and the time window spans multiple speed intervals, the speed interval with the highest coverage duration is taken as the speed interval of the multi-source data tag. When there are two or more speed intervals with the same coverage duration and both have the highest coverage duration, the multi-source consistency fusion module sets the normalized direction code of the multi-source data tag to an empty result and writes the data validity status recorded in the corresponding combed cotton web segment into the verification lock status.

[0091] When generating normalized direction codes, the multi-source consistency fusion module writes the corresponding process type identifier, speed range, data source location identifier, and profile version identifier as direction dictionary call keys into the multi-source data markers. These direction dictionary call keys are only used for version binding and consistency determination of the normalized direction codes. For two multi-source data markers participating in the same abnormal direction consistency determination, the process type identifier, speed range, and profile version identifier in their direction dictionary call keys should be consistent. If inconsistencies exist, the multi-source consistency fusion module re-executes direction normalization processing according to the profile version identifier corresponding to the trigger time. If re-normalization is not possible, the corresponding normalized direction code is set to an empty result.

[0092] The multi-source consistency fusion module performs directional normalization on each multi-source data tag:

[0093] ,

[0094] ,

[0095] in, The serial number for multi-source data is... And the exception type number is The set of direction categories matched at that time; Number the data from multiple sources. This is the exception type number; For direction category number; The exception type number is The set of direction categories at time; The serial number for multi-source data is... The exception type number carried at the time; The serial number for multi-source data is... The original direction of the anomaly carried at that time; The exception type number is And the direction category number is At that time, it is permissible to classify the original abnormal directions into the same standard direction code set; The serial number for multi-source data is... And the exception type number is Normalized direction code at time; for The number of elements in; for The unique direction category number in the sequence; The exception type number is And the direction category number is Standard direction code for time; An empty result indicating that no valid normalized direction code was matched.

[0096] The original anomaly directions in the anomaly direction dictionary are derived from the baseline state and allowable fluctuation range of the phase quality profile. For data with numerical change directions, the original anomaly directions are determined according to the deviation direction relative to the upper or lower limit of the allowable fluctuation range; those above the upper limit are marked as high or enhancing directions, and those below the lower limit are marked as low or weakening directions. For state-type data, the original anomaly directions are determined according to the change result of the state label relative to the baseline state, including interruptions, missing values, loose enhancements, or texture discontinuities. Taking thin web risk as an example, narrower width, lighter local brightness, lower combed sliver basis weight, and lower basis weight responses in short-segment unevenness are categorized as quantity shortage direction; taking edge cotton shortage risk as an example, interrupted edge continuity, missing corresponding combing head side areas, and decreased basis weight responses in output quality data are categorized as side shortage direction; taking suction disturbance risk as an example, suction pressure below the lower limit of the allowable fluctuation range, increased web looseness, and fly concentration above the upper limit of the allowable fluctuation range are categorized as increased disturbance direction; taking separation and bonding abnormality risk as an example, discontinuous bonding texture, abnormal response during the separation and bonding stage, and periodic evenness fluctuations are categorized as bonding instability direction. The above classifications are only used within the same abnormality type, and the standard direction codes between different abnormality types are not interchangeable.

[0097] Periodic fluctuation risk corresponds to consecutive occurrences of the same type of status label, quality spectrum event status, or evenness fluctuation status under the same combing head identifier and the same clamp phase identifier; bonding mark anomaly risk corresponds to bonding texture position shift, bonding texture discontinuity, and abnormal response during the separation bonding stage. All of the above anomaly types are categorized into standard direction codes within their respective anomaly types using an anomaly direction dictionary, and are bound and stored with the corresponding combed cotton web segment identifier, data source timestamp, and image version identifier.

[0098] When two multi-source data markers belong to the same anomaly type, have the same normalized direction code after anomaly direction dictionary transformation, and are located in the same combed cotton web segment or adjacent delayed combed cotton web segments, the two multi-source data markers are determined to have consistent anomaly directions.

[0099] ,

[0100] in, The serial number for multi-source data is... and multi-source data labeling sequence number The value for determining the consistency of abnormal directions is 1, which indicates that the consistency of abnormal directions is satisfied, and 0 indicates that the consistency of abnormal directions is not satisfied. A sequence number for a multi-source data item used in consistency determination; This serves as the sequence number for another multi-source data point used in the consistency determination. This is an indicator function that takes the value 1 when the condition is true and 0 when the condition is false. The serial number for multi-source data is... The exception type number carried at the time; The serial number for multi-source data is... The exception type number carried at the time; The serial number for multi-source data is... And the exception type number is Normalized direction code at time; The serial number for multi-source data is... And the exception type number is Normalized direction code at time; The serial number for multi-source data is... The corresponding combed cotton web segment marking; The serial number for multi-source data is... The corresponding combed cotton web segment marking; The serial number for multi-source data is... The set of identifiers for adjacent delayed combed cotton web segments corresponding to the time; The serial number for multi-source data is... The set of identifiers for adjacent delayed combed cotton web segments corresponding to the time; Refer to the aforementioned definition.

[0101] The multi-source consistency fusion module determines the consistency of the apparent anomaly marker with the output quality response marker, the equipment status response marker, and the environmental disturbance marker. If the consistency value of at least one apparent anomaly marker with at least one output quality response marker, equipment status response marker, or environmental disturbance marker is 1, and both correspond to the same anomaly type, then the consistency triggering condition is met, and a quality anomaly warning record is generated.

[0102] If only the equipment status response markers meet the requirement of consistent abnormal direction, and the preset lower limit of the number of consecutive times of equipment attention is reached in the continuous combed cotton web segment records of the same combing head, while the apparent abnormality marker and the output quality response marker do not appear, then an equipment attention record is generated. The preset lower limit of the number of consecutive times of equipment attention is calibrated by the equipment status response benchmark in the equipment debugging stage or in the historical combed cotton web segment records, and is bound and saved with the profile version identifier.

[0103] For situations where the original anomaly direction falls into multiple direction categories, the original anomaly direction is missing, the data source location identifier is missing, or two multi-source data markers belong to different combed cotton web segments and the identifier set of adjacent delayed combed cotton web segments cannot be obtained, the multi-source consistency fusion module adopts a conservative strategy. It sets the normalized direction code of the multi-source data markers involved in this situation to an empty result and writes the data validity status of the corresponding combed cotton web segment record into the review lock status. This conservative strategy ensures that single data source anomalies that have not completed direction normalization or cannot complete the confirmation of adjacent delay correspondences do not participate in the consistency triggering condition determination of quality anomaly warning records, but are still retained in the quality traceability database for manual review and subsequent profile version updates.

[0104] Example 2:

[0105] Based on Example 1, this example provides a method for determining delayed matching relationships in a combed cotton web quality anomaly early warning system based on multi-source data fusion.

[0106] The output quality data includes at least one of the following: combed sliver weight fluctuation status, evenness fluctuation status, short segment unevenness status, and quality spectrum event status. The combed cotton web segment synchronization module establishes a delay matching relationship based on the conveying path length of the combed cotton web from the corresponding combing head to the output quality detection position, the speed status in the equipment status data, the production cycle, and the output quality detection time window. The delay matching relationship includes a start delay and an end delay. The start delay represents the time offset corresponding to the arrival of the starting boundary of the combed cotton web segment at the output quality detection position, and the end delay represents the time offset corresponding to the arrival of the ending boundary of the combed cotton web segment at the output quality detection position. Based on the start delay and the end delay, the combed cotton web segment synchronization module reverses the output quality detection time window to the combed cotton web formation time window and matches the output quality data to the combed cotton web segment record that occurred earlier.

[0107] In one implementation, the delay matching relationship is determined by the conveying path length from the combing head to the output quality inspection position, the machine speed, the production cycle time, and the output quality inspection time window. The initial delay represents the time offset corresponding to the arrival of the starting boundary of the combed cotton web segment at the output quality inspection position; the termination delay represents the time offset corresponding to the arrival of the ending boundary of the combed cotton web segment at the output quality inspection position; and the cross-segment matching ratio represents the coverage relationship between the same output quality inspection time window, after delay-based back-calculation, and multiple preceding combed cotton web segment records. When the back-calculated combed cotton web formation time window corresponds to only one combed cotton web segment record, the output quality data is written to that combed cotton web segment record; when the back-calculated combed cotton web formation time window corresponds to multiple combed cotton web segment records, the output quality data is written to the corresponding combed cotton web segment records respectively, and simultaneously written to the cross-segment matching ratio.

[0108] In an optional implementation of Example 2, when establishing a delay matching relationship, the combed cotton web segment synchronization module reads the calibrated value of the conveying path length from each combing head to the output quality detection position during the equipment debugging stage or the production batch start-up stage, and performs time conversion on the calibrated value of the conveying path length based on the current vehicle speed and production cycle to obtain the start delay and end delay of the corresponding combing head.

[0109] The current vehicle speed status is calculated using the detection time window of the output quality data and the time-weighted average vehicle speed within a production cycle range that is traced back. The time-weighted average vehicle speed is obtained from the vehicle speed status in the equipment status data, and the sampling time interval between adjacent vehicle speed sampling values ​​is used as the weight. Vehicle speed sampling values ​​corresponding to unstable operating states before and after stopping, strip breakage, and roll change are not included in the calculation of time-weighted average vehicle speed.

[0110] The effective vehicle speed used for delay calculation is the time-weighted average speed greater than 0 and whose operating time window has not reached the preset speed fluctuation limit. Speed ​​abrupt change states are determined based on the ratio of the difference between the maximum and minimum effective vehicle speeds within the same operating time window to the time-weighted average speed of that operating time window. A speed abrupt change state is defined when this ratio reaches the preset speed fluctuation limit. The preset speed fluctuation limit is an implementation parameter, ranging from 5% to 30%, and is calibrated using stable operating samples from the equipment debugging phase or the production batch start-up phase. When stable operating samples have been obtained, the upper quantile of the speed fluctuation ratio in the stable operating samples is taken as the preset speed fluctuation limit and bound to the profile version identifier for storage.

[0111] The initial delay is obtained according to the following formula:

[0112] ,

[0113] in, The serial number for the combed hair is And the output quality data detection window number is The start delay of time; This refers to the serial number of the comb head; To output the quality data detection window number; The serial number for the combed hair is The maximum allowed delay; The serial number for the combed hair is The lower limit of the allowed delay; The serial number for the combed hair is The calibration value of the transport path length used to calculate the initial delay; The serial number for the combed hair is And the output quality data detection window number is Time-weighted average vehicle speed; To calculate the smaller value; This is used to calculate the larger value.

[0114] The termination delay is obtained according to the following formula:

[0115] ,

[0116] in, The serial number for the combed hair is And the output quality data detection window number is Termination delay; The serial number for the combed hair is The calibration value of the transport path length used to calculate the termination delay; The serial number for the combed hair is Production cycle time; , , , , , and Refer to the aforementioned definition.

[0117] The corresponding comb head serial number is The calibrated value of the conveying path length from the leading edge of the combed cotton web segment to the output quality inspection position. The corresponding comb head serial number is The calibrated value of the conveying path length from the rear edge of the combed cotton web segment to the output quality detection position, and and All are calibration values ​​greater than 0. The conveying path length calibration value satisfies... When the tracking results of the combed cotton web during the equipment commissioning phase or the production batch start-up phase do not meet the relationship, the combed cotton web segment synchronization module uses the effective conveying path length calibration value under the previous image version identifier to backtrack. If the backtracking value cannot be obtained, the corresponding output quality data is saved in a verification and locking state.

[0118] The combed cotton web segment synchronization module uses the detection time window, start delay, and end delay of the output quality data to inversely calculate the combed cotton web formation time window corresponding to that output quality data.

[0119] , ,

[0120] in, The serial number for the combed hair is And the output quality data detection window number is The starting point of the time window for the formation of the combed cotton web is obtained by reverse calculation. The serial number for the combed hair is And the output quality data detection window number is The end point of the time window for the formation of the combed cotton web obtained by reverse calculation; The output quality data detection window number is The start time of the detection; The output quality data detection window number is The detection end time; and Refer to the aforementioned definition.

[0121] Before performing the cross-segment matching ratio calculation, the combed web segment synchronization module verifies the validity of the start delay, end delay, and the reverse-calculated combed web formation time window. Validity verification includes ensuring the end delay is greater than or equal to the start delay, the start date of the combed web formation time window is earlier than or equal to the end date, and the reverse-calculated combed web formation time window falls within the effective production time range of the current production batch. If the validity verification is not met, the combed web segment synchronization module does not perform the cross-segment matching ratio calculation, saves the corresponding output quality data in a verification locked state, and prohibits this output quality data from participating in the consistency trigger condition determination.

[0122] The effective production time range for the current production batch refers to the time from the start of the production batch and its entry into a stable feeding state until the end of the production batch. A stable feeding state is defined as the state where the machine speed and output quality data fall within the allowable fluctuation range of the corresponding phase quality profile. The time periods corresponding to unstable operating states such as stopping, strip breakage, and roll change are not included in the effective production time range for the current production batch. The time period after roll change is completed re-enters the effective production time range for the current production batch once the machine speed and output quality data have recovered to the allowable fluctuation range of the corresponding phase quality profile.

[0123] When the combed web formation time window obtained through reverse engineering has an effective overlap duration greater than 0 with only one phase segment of a preceding combed web segment record, the combed web segment synchronization module binds the output quality data to that combed web segment record. When the combed web formation time window has an effective overlap with multiple preceding combed web segment records, the combed web segment synchronization module calculates the overlap duration between each preceding combed web segment record and the combed web formation time window, and writes the same output quality data into multiple combed web segment records according to the cross-segment matching ratio.

[0124] ,

[0125] in, The serial number for the combed hair is The output quality data detection window number is And the serial number of the combed cotton web segment is The cross-segment matching ratio at that time; Record the sequence number of the combed cotton web segment participating in the matching; The serial number for the combed cotton web segment is: The start time of the phase segment; The serial number for the combed cotton web segment is: The end time of the phase segment; Record the serial number of the combed cotton web segment participating in the summation; For the serial number of the combed head And the output quality data detection window number is Correspondingly, and with an overlap of more than 0 in the time window of the combed cotton web formation obtained by reverse calculation, a set of candidate combed cotton web segments is recorded; The serial number for the combed cotton web segment is: The start time of the phase segment; The serial number for the combed cotton web segment is: The end time of the phase segment; , , , , and Refer to the aforementioned definition.

[0126] when When empty, the combed cotton web segment synchronization module does not calculate the cross-segment matching ratio, saves the corresponding output quality data in a verification and locking state, and records the matching status of the output quality data as a verification and locking state; the matching status record is a record field of the output quality data in the delayed matching process, used to identify whether the output quality data has completed the combed cotton web segment binding.

[0127] The higher the cross-segment matching ratio, the higher the time correspondence between the combed cotton web segment record and the output quality data. The combed cotton web segment synchronization module only writes the output quality data whose cross-segment matching ratio reaches the preset binding lower limit into the corresponding combed cotton web segment record; the portion that does not reach the preset binding lower limit is only retained in the quality traceability database and does not participate in the consistency trigger condition determination of the same combed cotton web segment.

[0128] The preset binding lower limit is an implementation parameter used to determine whether the time coverage relationship between the output quality data and the combed cotton web segment record meets the binding condition. Its value ranges from 0.1 to 1.0. The preset binding lower limit is calibrated by the marked combed cotton web tracking results during the equipment debugging phase or by stable operating samples during the production batch startup phase, and is bound and saved with the portrait version identifier. When the stable operating samples are insufficient to determine the lower quantile value, the preset binding lower limit is set to 0.5. The upper and lower quantile values ​​are the sample quantile results under the preset quantile caliber, which is bound and saved with the portrait version identifier. Under the same portrait version identifier, the preset binding lower limit remains unchanged. If the matching ratio of all cross-segments corresponding to the same output quality data is lower than the preset binding lower limit, the combed cotton web segment synchronization module saves the output quality data in a verification locked state, not writing it to any combed cotton web segment record, nor allowing it to participate in the consistency trigger condition determination.

[0129] The marked combed cotton web tracking result refers to the tracking data obtained during the equipment commissioning phase or production batch start-up phase after recording the start time, arrival time at the output quality detection position, and corresponding combed cotton web segment identifier for an identifiable combed cotton web segment formed by a specified combing head. The marked combed cotton web tracking result includes combing head identifier, combed cotton web segment identifier, start time, arrival time, output quality detection time window, and image version identifier, which are used to determine the conveyor path length calibration value, delay lower limit, delay upper limit, and preset binding lower limit.

[0130] The start delay and end delay decrease as the vehicle speed increases and increase as the calibrated length of the conveyor path increases; the lower limit and upper limit of the delay are determined by the results of the no-load belt running test, the cotton running test, or the marked combed cotton web tracking results during the equipment commissioning phase, and remain unchanged within the same production batch.

[0131] If, after removing the speed sampling values ​​corresponding to unstable operating states before and after stopping, breaking, and changing rolls, a time-weighted average speed greater than 0 cannot be obtained and the operating time window does not reach the preset speed fluctuation limit, the combed cotton web segment synchronization module will suspend updating the delay matching relationship corresponding to the detection time window, save the corresponding output quality data in a verification and locking state, and prohibit the output quality data from triggering quality anomaly warning records independently.

[0132] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0133] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A quality anomaly early warning system for combed cotton web based on multi-source data fusion, characterized in that, It includes an industrial control computing unit, a data interface unit, a data storage unit, and an early warning output unit; The data interface unit receives multi-source production data including production time identifier, combing head identifier, clamp phase data, combed cotton web appearance data, output quality data, equipment status data, and environmental disturbance data; The industrial control computing unit includes a combed cotton web segment synchronization module, a phase quality profiling module, and a multi-source consistency fusion module; The combed cotton web segment synchronization module generates combed cotton web segment identifiers and establishes combed cotton web segment records according to the combing head identifier, clamp phase data and production time identifier. It writes the combed cotton web appearance data, equipment status data, environmental disturbance data and output quality data corresponding to the delay matching relationship into the combed cotton web segment records and stores them in the data storage unit. The phase quality profiling module retrieves historical combed cotton web segment records from the data storage unit to generate a phase quality profile. The multi-source consistency fusion module compares the combed cotton web segment records with the phase quality profile to generate multi-source data tags; Multi-source data tags include apparent anomaly tags, output quality response tags, equipment status response tags, and environmental disturbance tags. Each tag includes anomaly type and anomaly direction. When the apparent anomaly marker corresponds to the same anomaly type and direction as at least one of the output quality response marker, equipment status response marker, and environmental disturbance marker, and satisfies one of the relationships of the same combed cotton web segment or adjacent delayed combed cotton web segments, the multi-source consistency fusion module generates a quality anomaly warning record. The early warning output unit outputs the quality anomaly early warning record.

2. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The combed cotton web is a fibrous web formed by the combing heads of the combing machine before they are combed, separated, joined, and aggregated into strips; The combed cotton web segment is a continuous combed cotton web segment formed by the same combing head in one nipper phase segment; The combed cotton web segment record includes a combed cotton web segment identifier, a combing head identifier, a clamping phase identifier, a production time identifier, a production batch identifier, a process variety identifier, a data source timestamp, a data validity status, and a profile version identifier. The data validity status includes at least one of the following: synchronized status, missing sampling status, delayed matching status, and verification lock status.

3. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The clamp phase data includes at least two of the following: main shaft phase signal, clamp opening and closing stage identifier, separation and engagement stage identifier, top comb action stage identifier, and combed web transfer stage identifier; The combed cotton web segment synchronization module divides continuous production time markers into phase segments, binds the phase segments to the combing head identifier that generates the phase segment, and generates combed cotton web segment identifiers according to the start and end time of the phase segment, the phase sequence number, and the combing head identifier. When the same combing head continuously outputs multiple phase segments, the combed cotton web segment synchronization module maintains the connection relationship between adjacent combed cotton web segments according to the phase sequence number.

4. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The apparent data of the combed cotton web is obtained by feature processing of the image acquisition data of the combed cotton web, which includes single-frame image features and continuous frame change features. The apparent data of the combed cotton web includes at least three of the following: width status, edge continuity status, local light and dark uniformity status, joint texture status, web surface looseness status, and visible foreign matter status. The industrial control computing unit converts the apparent data of the combed cotton web into status tags, and writes the status tags, acquisition location identifiers, and image timestamps into the corresponding combed cotton web segment records.

5. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The output quality data includes at least one of the following: combing strip quantity fluctuation state, strip fluctuation state, short segment unevenness state, and quality spectrum event state; The combed cotton web segment synchronization module establishes a delay matching relationship based on the conveying path length of the combed cotton web from the corresponding combing head to the output quality inspection position, the speed status in the equipment status data, the production cycle time, and the output quality inspection time window. The delay matching relationship includes a start delay and an end delay. The start delay represents the time offset corresponding to the arrival of the start boundary of the combed cotton web segment at the output quality detection position, and the end delay represents the time offset corresponding to the arrival of the end boundary of the combed cotton web segment at the output quality detection position. The combed cotton web segment synchronization module reverses the output quality detection time window to the combed cotton web formation time window based on the start delay and the end delay, and matches the output quality data to the combed cotton web segment record that occurred earlier.

6. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 5, characterized in that, The delayed matching relationship includes the cross-segment matching ratio, which represents the coverage relationship between the same output quality detection time window after delayed back-calculation and multiple preceding combed cotton web segment records. When the combed cotton web formation time window obtained by reverse calculation corresponds to a combed cotton web segment record, the combed cotton web segment synchronization module will output quality data and write it into the combed cotton web segment record. When the combed cotton web formation time window obtained by reverse calculation corresponds to multiple combed cotton web segment records, the combed cotton web segment synchronization module will output quality data and write it into the corresponding combed cotton web segment records respectively, and synchronously write the corresponding cross-segment matching ratio.

7. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The equipment status data includes at least two of the following: machine speed status, separating roller speed status, drawing roller speed status, suction pressure status, main motor load status, and roll changing status. The environmental disturbance data includes at least one of temperature and humidity status, fly ash concentration status, and lighting intensity status; The industrial control computing unit writes the equipment status data and environmental disturbance data into the corresponding combed cotton web segment record according to the production time identifier, and writes the data into the combed cotton web segment record corresponding to the phase segment that coincides with the sampling time when the data sampling time falls into the phase segment boundary.

8. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The phase quality profiling module groups historical combed cotton web segment records according to at least one of production batch, process type, speed range, operation stage and suction pressure state, and establishes a phase quality profile corresponding to the combing head-nip phase combination in each group. The phase quality profile includes an apparent state benchmark, an output quality response benchmark, an equipment state response benchmark, and an environmental disturbance benchmark. Each benchmark includes a benchmark state, an allowable fluctuation range, a frequency of occurrence, a number of consecutive occurrences, and an adjacent phase correlation state.

9. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 8, characterized in that, The phase quality profile module reads the warning result identifiers recorded in the historical combed cotton web segment when updating the phase quality profile; When a historical combed cotton web segment record is already bound to a quality anomaly warning record, the phase quality profiling module writes the historical combed cotton web segment record into the anomaly sample set, or excludes the historical combed cotton web segment record when calculating the apparent state benchmark, output quality response benchmark, equipment state response benchmark, and environmental disturbance benchmark. The abnormal sample set is bound and stored with the corresponding abnormal type, comb head identifier, clamp phase identifier, trigger time, and handling result identifier.

10. The early warning system for quality anomalies in combed cotton web based on multi-source data fusion according to claim 1, characterized in that, The quality anomaly early warning record includes combed cotton web segment identifier, combing head identifier, clamp phase identifier, anomaly type, early warning level, trigger time, and multi-source data markers that participate in generating the early warning. The multi-source data markers that participate in generating the early warning include marker levels. The anomaly types include at least two of the following: thin web risk, edge cotton shortage risk, seam mark anomaly risk, periodic fluctuation risk, air suction disturbance risk, and separation seam anomaly risk. The early warning output unit writes the quality anomaly early warning record into the quality traceability database and sends a process handling instruction to the combing machine control system or the workshop production execution system. The process handling instruction includes at least one of the following: reviewing the combing head, checking the suction pressure status, checking the environmental disturbance data, marking the combed cotton web segment to be traced, and retrieving the corresponding historical combed cotton web segment record. The warning level is determined based on the number of multi-source data markers, the marker level, the number of consecutive occurrences, and the response status of adjacent delayed combed cotton web segments within the corresponding combed cotton web segment.

Images