Non-woven fabric production line quality monitoring method and system

Abstract

The invention relates to the technical field of production monitoring, and discloses a quality monitoring method and system for a non-woven fabric production line, and the method comprises the steps: synchronously capturing the linear speed information, gram weight information and appearance image information in the non-woven fabric production line; performing frame alignment on the gram weight information and the appearance image information to obtain synchronization feature data; fusing the texture anomaly candidate region and the gram weight information fragment in the synchronized feature data to obtain a fused defect description; matching a combination mode of image morphological features and surface density statistical distribution features in the fusion defect description to obtain a defect category; reversely mapping to a process flow diagram, and positioning primary process links causing defect categories to obtain a judgment conclusion; integrating the judgment conclusion, the fragment of the synchronization feature data and the identifier of the primary process link to obtain a quality monitoring report; the quality monitoring efficiency of the non-woven fabric production line can be improved.

Description

Technical Field

[0001] This invention relates to the field of production monitoring technology, and in particular to a method and system for quality monitoring of nonwoven fabric production lines. Background Technology

[0002] Existing technologies fail to achieve efficient and synchronous integration of multi-dimensional monitoring information such as line speed, basis weight, and appearance images when monitoring the quality of nonwoven fabric production lines. The lack of a unified synchronization benchmark for different types of data collection results in weak correlation between the data in the time and space dimensions, making it impossible to form synchronized characteristic data that can comprehensively reflect the production status, thus hindering subsequent defect analysis.

[0003] Existing technologies have significant shortcomings in defect identification and root cause tracing. They rely solely on single types of monitoring data for defect judgment, failing to fully integrate fabric morphology and weight distribution characteristics. This results in low accuracy in defect category determination. Furthermore, there is a lack of in-depth analysis of the correlation between defect feature changes and production process stages, making it difficult to accurately pinpoint the primary process stage that causes defects. Consequently, quality monitoring cannot provide effective guidance for production optimization, resulting in poor overall monitoring efficiency and practicality. Therefore, improving the efficiency of quality monitoring in nonwoven fabric production lines has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a method and system for quality monitoring of nonwoven fabric production lines to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for quality monitoring of a nonwoven fabric production line, comprising: S1. Synchronously capture linear speed information reflecting the material transport rhythm, weight information reflecting the fiber web forming quality, and appearance image information reflecting the microstructure of the fabric in the nonwoven fabric production line. S2. Using the beat pulse in the linear velocity information as a synchronization reference, perform frame alignment between the weight information and the appearance image information to obtain the synchronization feature data of the nonwoven fabric production line. S3. The texture anomaly candidate region and basis weight information fragment in the synchronized feature data are fused to obtain the fusion defect description of the nonwoven fabric production line; S4. Based on the typical feature patterns in the preset process knowledge rule base, match the combination of image morphological features and surface density statistical distribution features in the fused defect description to obtain the defect category of the nonwoven fabric production line. S5. Based on the defect category and the variation gradient and periodicity of the areal density statistical distribution characteristics along the direction of travel of the nonwoven fabric production line, the process flow diagram of the nonwoven fabric production line is mapped in reverse, and the primary process link that causes the defect category is located, so as to obtain the judgment conclusion of the nonwoven fabric production line. S6. Integrate the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line.

[0006] In a preferred embodiment, the synchronous capture of linear velocity information reflecting the material transport rhythm, basis weight information reflecting the fiber web forming quality, and appearance image information reflecting the microstructure of the fabric surface in the nonwoven fabric production line includes: By monitoring the rotational speed pulse signal in the nonwoven fabric production line, the linear speed information of the nonwoven fabric production line can be obtained; The weight information of the nonwoven fabric production line is obtained by measuring the intensity attenuation of rays after penetrating the fiber web in the nonwoven fabric production line. Optical signals of the fabric surface are collected by scanning laterally along the nonwoven fabric production line to obtain the appearance image information of the nonwoven fabric production line.

[0007] In a preferred embodiment, the step of using the beat pulse in the linear velocity information as a synchronization reference to perform frame alignment between the weight information and the appearance image information to obtain the synchronization feature data of the nonwoven fabric production line includes: The equally spaced beat pulses in the linear velocity information are marked as synchronization event markers for the nonwoven fabric production line; Based on the time window defined by the synchronization event marker, the weight signal values ​​of the weight information are merged to obtain the weight signal frame of the nonwoven fabric production line. Based on the time window, multiple rows of appearance image data in the appearance image information are stitched together to obtain the appearance image frame of the nonwoven fabric production line. The weight signal frame and the appearance image frame are correlated and aligned to obtain the synchronization data of the nonwoven fabric production line; Based on the timing of the synchronization event markers, the synchronization data is arranged to obtain the synchronization characteristic data of the nonwoven fabric production line.

[0008] In a preferred embodiment, fusing the texture anomaly candidate regions and basis weight information fragments in the synchronized feature data to obtain a fusion defect description of the nonwoven fabric production line includes: Traverse the appearance image frames in the synchronized feature data to locate areas in the appearance image frames where there are obvious differences in grayscale or texture, and obtain the texture anomaly candidate areas of the nonwoven fabric production line. Based on the position coordinates of the texture anomaly candidate region in the appearance image frame, trace the abnormal weight information in the synchronized feature data to obtain the weight information fragment of the nonwoven fabric production line. Visual morphological recognition is performed on the edge contours and texture arrangement of the candidate regions for texture anomalies to obtain the morphological feature set of the nonwoven fabric production line; Extract the rising edge, falling edge, and plateau trend of the numerical change in the basis weight information segment to obtain the distribution feature set of the nonwoven fabric production line; The morphological feature set and the distribution feature set are combined side by side to obtain the fusion defect description of the nonwoven fabric production line.

[0009] In a preferred embodiment, the method of matching the combination of image morphological features and areal density statistical distribution features in the fused defect description based on typical feature patterns in a preset process knowledge rule base to obtain the defect category of the nonwoven fabric production line includes: The typical visual morphological feature patterns and quantitative distribution feature patterns of the nonwoven fabric production line are retrieved from the preset process knowledge rule base. Separate the image morphological feature subset and the areal density statistical distribution feature subset from the description of the fusion defect; The image morphological feature subset is compared with the typical visual morphological feature pattern, and the areal density statistical distribution feature subset is compared with the quantitative distribution feature pattern to obtain the comparison result of the nonwoven fabric production line. Based on the comparison results, the process knowledge rule base is matched to obtain the defect categories of the nonwoven fabric production line.

[0010] In a preferred embodiment, the step of reverse mapping based on the variation gradient and periodicity of the defect category and the areal density statistical distribution characteristics along the travel direction of the nonwoven fabric production line to the process flow chart of the nonwoven fabric production line, and locating the primary process step that triggers the defect category to obtain the judgment conclusion of the nonwoven fabric production line includes: Based on the travel direction of the nonwoven fabric production line, the numerical changes of the areal density statistical distribution characteristics are analyzed to obtain the gradient information and periodicity information of the nonwoven fabric production line. Based on the defect cause type of the defect category, the process flow diagram of the nonwoven fabric production line is topologically traced to obtain the set of candidate defect process links of the nonwoven fabric production line. Based on the change pattern formed by the gradient information and the periodic information, the candidate defect process step set is causally converged to obtain the primary process step of the nonwoven fabric production line. The defect category, the primary process step, and the change pattern are associated and encapsulated to obtain the judgment conclusion of the nonwoven fabric production line.

[0011] In a preferred embodiment, the step of performing causal convergence on the candidate defect process step set based on the change pattern constituted by the gradient information and the periodic information to obtain the primary process step of the nonwoven fabric production line includes: Based on the fluid dynamics and mechanical transmission principles involved in the nonwoven fabric production line, the rate of change of the gradient information and the repetition interval of the periodic information are relabeled to obtain the process state characteristics of the nonwoven fabric production line. The process state characteristics are compared with the process knowledge rule base to obtain the link matching degree of the nonwoven fabric production line. Based on the matching degree of the links, links that are strongly correlated with the process state characteristics are selected from the candidate defect process links to obtain the root cause links to be determined in the nonwoven fabric production line. Based on the failure mechanism of the defect category, the root cause of the problem is determined by matching the failure mechanism to obtain the primary process of the nonwoven fabric production line.

[0012] In a preferred embodiment, the formula for calculating the link matching degree is as follows: ; In the formula, The matching degree of the aforementioned link. The current value of the gradient information. The current value of the periodic information. The gradient typical value in the aforementioned process knowledge rule base. The typical periodic value is found in the process knowledge rule base. The preset gradient weight factor, The preset periodic weighting factor, This refers to a process step in the nonwoven fabric production line.

[0013] In a preferred embodiment, the integration of the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line includes: Based on the judgment conclusion, the synchronized feature data is trimmed and extracted to obtain evidence data fragments of the nonwoven fabric production line. Based on a preset report template, the defect attribute description, evidence data fragments, and identifiers of the primary process steps in the judgment conclusion are structurally filled in to obtain a draft report for the nonwoven fabric production line. The temporal and spatial positions of the evidence data fragments in the synchronized feature data are converted and encoded to obtain the data index information of the nonwoven fabric production line. The draft report is assembled with the data index information to obtain the quality monitoring report of the nonwoven fabric production line.

[0014] To address the above problems, the present invention also provides a quality monitoring system for a nonwoven fabric production line, the system comprising: The information synchronization and capture module is used to synchronously capture linear speed information reflecting the material conveying rhythm, weight information reflecting the fiber web forming quality, and appearance image information reflecting the micro-morphology of the fabric surface in the nonwoven fabric production line. The synchronization feature generation module is used to perform frame alignment between the weight information and the appearance image information using the beat pulse in the linear velocity information as a synchronization reference, so as to obtain the synchronization feature data of the nonwoven fabric production line. The defect description fusion module is used to fuse the texture anomaly candidate regions and basis weight information fragments in the synchronized feature data to obtain the fused defect description of the nonwoven fabric production line. The defect category matching module is used to match the combination of image morphological features and surface density statistical distribution features in the fused defect description based on typical feature patterns in the preset process knowledge rule base, so as to obtain the defect category of the nonwoven fabric production line. The primary process location module is used to reverse map the change gradient and periodicity of the defect category and the areal density statistical distribution characteristics along the travel direction of the nonwoven fabric production line to the process flow diagram of the nonwoven fabric production line, and locate the primary process link that causes the defect category, so as to obtain the judgment conclusion of the nonwoven fabric production line. The monitoring report generation module is used to integrate the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves precise frame alignment between weight information and appearance image information by synchronously capturing three key monitoring information types: linear velocity, weight, and appearance image. Using the beat pulse in the linear velocity as a unified synchronization benchmark, it forms complete and coherent synchronized feature data. Then, by fusing candidate regions of texture anomalies with corresponding weight information fragments, it constructs a fused defect description, comprehensively covering the image morphology and surface density statistical distribution features, and significantly improving the accuracy of defect category determination.

[0016] 2. This invention combines the variation gradient and periodicity of defect categories and areal density statistical distribution characteristics along the production line direction, reverse-maps them to the process flow chart, and accurately locates the primary process links. It integrates judgment conclusions, synchronized feature data fragments, and process link identifiers to generate a quality monitoring report, providing clear guidance for production line process optimization and significantly improving the efficiency and practical application value of nonwoven fabric production line quality monitoring. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for quality monitoring in a nonwoven fabric production line according to an embodiment of the present invention. Figure 2 A functional module diagram of a nonwoven fabric production line quality monitoring system provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for quality monitoring of a nonwoven fabric production line. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for quality monitoring of a nonwoven fabric production line can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a quality monitoring method for a nonwoven fabric production line according to an embodiment of the present invention. In this embodiment, the quality monitoring method for a nonwoven fabric production line includes: S1. Synchronously capture linear speed information reflecting the material transport rhythm, weight information reflecting the fiber web forming quality, and appearance image information reflecting the microstructure of the fabric in the nonwoven fabric production line. In this embodiment of the invention, the synchronous capture of linear velocity information reflecting the material transport rhythm, basis weight information reflecting the fiber web forming quality, and appearance image information reflecting the microstructure of the fabric surface in the nonwoven fabric production line includes: By monitoring the rotational speed pulse signal in the nonwoven fabric production line, the linear speed information of the nonwoven fabric production line can be obtained; The weight information of the nonwoven fabric production line is obtained by measuring the intensity attenuation of rays after penetrating the fiber web in the nonwoven fabric production line. Optical signals of the fabric surface are collected by scanning laterally along the nonwoven fabric production line to obtain the appearance image information of the nonwoven fabric production line.

[0021] A speed sensor is installed at the drive roller of the nonwoven fabric production line. The device maintains a fixed distance from the roller and is in real time attached to the roller surface. During the rotation of the roller, the sensor will generate regular speed pulse signals. The sensor continuously receives and records these pulse signals. By correlating the generation frequency of the pulse signals with the fixed circumference of the roller, linear speed information that can reflect the rhythm of material transport can be directly obtained.

[0022] In the nonwoven fabric production line, X-ray emitting devices and X-ray receiving devices are respectively installed at corresponding positions above and below the moving fiber web. The X-ray emitting devices continuously emit X-rays of stable intensity into the fiber web. When the X-rays penetrate the moving fiber web, the material density of the fiber web will attenuate the intensity of the X-rays. The X-ray receiving devices capture the actual intensity of the X-rays after they penetrate the fiber web in real time. By comparing the initial intensity of the X-rays when they are emitted with the attenuated intensity after penetration, the basis weight information that reflects the quality of the fiber web forming can be accurately obtained.

[0023] A linear optical acquisition device is installed above the fabric surface in the nonwoven fabric production line. The device performs a uniform scanning motion along the transverse direction of the production line. During the scanning process, the optical acquisition device continuously captures the optical signals reflected by the fabric surface. Differences in the microstructure of the fabric surface will cause corresponding changes in the intensity, wavelength, and other characteristics of the reflected optical signals. The optical signals continuously acquired during the scanning process are sequentially integrated according to the transverse scanning order to completely restore the microstructure of the fabric surface and obtain appearance image information that can reflect the microstructure of the fabric surface.

[0024] The beneficial effects are as follows: by monitoring the rotation speed pulse signal, the linear velocity information reflecting the material transport rhythm is accurately obtained; by measuring the intensity attenuation of rays after penetrating the traveling fiber web, the basis weight information reflecting the fiber web forming quality is accurately obtained; and by scanning the fabric surface optical signal along the production line, the appearance image information reflecting the micro-morphology of the fabric surface is comprehensively captured. The simultaneous capture of these three types of key monitoring information can completely cover the core quality-related dimensions in the nonwoven fabric production process, providing a reliable data foundation for subsequent frame alignment of various information and construction of synchronized feature data, effectively ensuring the accuracy of subsequent defect identification and process location, and thus improving the overall effect of quality monitoring in the nonwoven fabric production line.

[0025] S2. Using the beat pulse in the linear velocity information as a synchronization reference, perform frame alignment between the weight information and the appearance image information to obtain the synchronization feature data of the nonwoven fabric production line. In this embodiment of the invention, the step of using the beat pulse in the linear velocity information as a synchronization reference to perform frame alignment between the weight information and the appearance image information to obtain the synchronization feature data of the nonwoven fabric production line includes: The equally spaced beat pulses in the linear velocity information are marked as synchronization event markers for the nonwoven fabric production line; Based on the time window defined by the synchronization event marker, the weight signal values ​​of the weight information are merged to obtain the weight signal frame of the nonwoven fabric production line. Based on the time window, multiple rows of appearance image data in the appearance image information are stitched together to obtain the appearance image frame of the nonwoven fabric production line. The weight signal frame and the appearance image frame are correlated and aligned to obtain the synchronization data of the nonwoven fabric production line; Based on the timing of the synchronization event markers, the synchronization data is arranged to obtain the synchronization characteristic data of the nonwoven fabric production line.

[0026] By continuously tracking the cycle pulses generated in the linear velocity information, and by identifying the characteristic that the time interval between pulses is always consistent, cycle pulses that meet the medium interval condition are selected. The occurrence time of these pulses is accurately recorded and assigned a unique identifier, and they are directly marked as the synchronization event markers of the nonwoven fabric production line.

[0027] Using the time points corresponding to two adjacent synchronous events as boundaries, a fixed-duration time window is clearly defined. Within each time window, all real-time collected weight signal values ​​are collected, sorted and arranged according to the order of signal generation time, and then these scattered weight signal values ​​are merged into a unified signal set through summarization and integration to obtain the weight signal frame of the nonwoven fabric production line.

[0028] Using the time window defined by the synchronous event marker, extract all multi-line appearance image data obtained by lateral scanning along the production line within each same time window. Based on the lateral scanning position corresponding to each line of data, stitch these multi-line image data sequentially in a fixed direction from left to right or from right to left to ensure that the edges of adjacent lines of images are completely aligned, thus fully restoring the lateral appearance of the fabric within the time window and obtaining the appearance image frame of the nonwoven fabric production line.

[0029] By using the correspondence of time windows, the weight signal frames and appearance image frames generated within the same time window are bound together. The corresponding synchronization event markers of the two are checked to confirm their complete match in the time dimension. At the same time, the horizontal position information of the production line corresponding to the two is recorded to ensure that each set of weight signal frames forms a unique and accurate association with the corresponding appearance image frame, thereby obtaining the synchronization data of the nonwoven fabric production line.

[0030] Based on the chronological order of the synchronous events generated during the production process, all the associated and aligned synchronous data are arranged in an orderly manner, so that the synchronous data are presented sequentially according to the operation process of the production line. Each synchronous data corresponds to a specific stage in the production process, ultimately forming a coherent, complete and time-sequential synchronized characteristic data of the nonwoven fabric production line.

[0031] The beneficial effect is that using the beat pulse in the linear velocity information as a synchronization benchmark ensures that the frame alignment process between weight information and appearance image information has a clear and unified reference standard, enabling precise matching of different types of monitoring data in the time dimension. By defining time windows to merge weight signal values ​​and stitching together multiple rows of appearance image data, the scattered monitoring data are transformed into structured weight signal frames and appearance image frames. After correlation and alignment, synchronized data is obtained. Finally, synchronized feature data is formed by arranging the data according to the time sequence of synchronization event markers. This fully preserves the correlation and coherence of multi-dimensional data in the production process, effectively eliminating information deviations caused by data asynchrony. It provides accurate, complete, and logically clear data support for subsequent quality monitoring stages such as defect description construction and defect category determination, ensuring the reliability and effectiveness of the overall monitoring process.

[0032] S3. The texture anomaly candidate region and basis weight information fragment in the synchronized feature data are fused to obtain the fusion defect description of the nonwoven fabric production line; In this embodiment of the invention, fusing the texture anomaly candidate region and basis weight information fragment in the synchronized feature data to obtain the fusion defect description of the nonwoven fabric production line includes: Traverse the appearance image frames in the synchronized feature data to locate areas in the appearance image frames where there are obvious differences in grayscale or texture, and obtain the texture anomaly candidate areas of the nonwoven fabric production line. Based on the position coordinates of the texture anomaly candidate region in the appearance image frame, trace the abnormal weight information in the synchronized feature data to obtain the weight information fragment of the nonwoven fabric production line. Visual morphological recognition is performed on the edge contours and texture arrangement of the candidate regions for texture anomalies to obtain the morphological feature set of the nonwoven fabric production line; Extract the rising edge, falling edge, and plateau trend of the numerical change in the basis weight information segment to obtain the distribution feature set of the nonwoven fabric production line; The morphological feature set and the distribution feature set are combined side by side to obtain the fusion defect description of the nonwoven fabric production line.

[0033] The system examines all appearance image frames in the synchronized feature data frame by frame, comparing the grayscale values ​​and texture structure of each region with the surrounding regions pixel by pixel. When the difference between the grayscale value of a certain region and the surrounding regions reaches a significant level, or when the arrangement and density of the texture are significantly different from the surrounding normal regions, the region is directly identified as a candidate region for texture abnormality in the nonwoven fabric production line.

[0034] Accurately record the horizontal and vertical position coordinates of candidate regions with texture abnormalities in the corresponding appearance image frames. Based on these position coordinates, search for the basis weight signal data of the corresponding spatial position in the synchronized feature data within the same time window, filter out abnormal basis weight signals that deviate from the normal basis weight range, and integrate and collect these continuous abnormal basis weight signals to obtain basis weight information fragments of the nonwoven fabric production line.

[0035] By meticulously observing the edge of candidate areas for texture anomalies using visual recognition methods, the curvature, smoothness, and closure of the edges are recorded. At the same time, the arrangement features of the textures within the area, such as direction, interweaving method, and distribution density, are analyzed. These extracted edge contour features and texture arrangement features are systematically organized and classified to obtain the morphological feature set of the nonwoven fabric production line.

[0036] By sorting through all the weight values ​​in the weight information segment in chronological order and tracking the dynamic changes of the values, when the value shows a continuous increasing trend, it is identified as an rising edge feature; when the value shows a continuous decreasing trend, it is identified as a falling edge feature; when the value remains constant for a period of time without obvious fluctuations, it is identified as a plateau trend feature. These three types of features are summarized and integrated to obtain the distribution feature set of the nonwoven fabric production line.

[0037] All edge contour features and texture arrangement features included in the morphological feature set are arranged in a one-to-one correspondence with the rising edge, falling edge and plateau trend features in the distribution feature set according to the feature attribute classification. This ensures that the detailed information of the two types of features is completely preserved and interconnected, forming a fusion defect description of the nonwoven fabric production line that simultaneously covers image morphological features and basis weight distribution features.

[0038] The beneficial effects are as follows: by traversing the appearance image frames, the candidate regions of texture anomalies with obvious differences in grayscale or texture are accurately located, providing an intuitive visual basis for defect identification. Then, based on their position coordinates, the abnormal weight information in the synchronized feature data is traced to obtain weight information fragments, establishing a spatial correlation between visual anomalies and weight anomalies. Subsequently, the morphological feature set related to edge contours and texture arrangement is extracted through visual morphological recognition, and the distribution feature set is extracted by combining the weight value change trend. The two types of feature sets are combined side by side to form a fused defect description, which comprehensively covers the image morphological features and surface density statistical distribution features of defects, making the defect description more complete and comprehensive. It effectively avoids the identification bias caused by a single feature dimension, provides high-quality feature support for subsequent accurate matching of defect categories, and ensures the reliability and comprehensiveness of defect identification in the quality monitoring of nonwoven fabric production lines.

[0039] S4. Based on the typical feature patterns in the preset process knowledge rule base, match the combination of image morphological features and surface density statistical distribution features in the fused defect description to obtain the defect category of the nonwoven fabric production line. In this embodiment of the invention, the method of matching the combination of image morphological features and areal density statistical distribution features in the fused defect description based on typical feature patterns in a preset process knowledge rule base to obtain the defect category of the nonwoven fabric production line includes: The typical visual morphological feature patterns and quantitative distribution feature patterns of the nonwoven fabric production line are retrieved from the preset process knowledge rule base. Separate the image morphological feature subset and the areal density statistical distribution feature subset from the description of the fusion defect; The image morphological feature subset is compared with the typical visual morphological feature pattern, and the areal density statistical distribution feature subset is compared with the quantitative distribution feature pattern to obtain the comparison result of the nonwoven fabric production line. Based on the comparison results, the process knowledge rule base is matched to obtain the defect categories of the nonwoven fabric production line.

[0040] The pre-set process knowledge rule base stores complete feature patterns corresponding to various known defects in the nonwoven fabric production process. These patterns are constructed not only based on a large number of real-world production defect cases accumulated over a long period, covering fabric defect manifestations under different production scenarios and equipment conditions, but also through verification and optimization using professional theories such as materials science and production technology. Atypical defect features caused by accidental factors are eliminated to ensure the scientific validity and universality of the patterns. The knowledge base is activated through a dedicated data retrieval interface. Based on the core requirements of current defect identification, fabric-related defects are accurately located. The selection criteria are clearly defined as retaining only feature patterns directly related to fabric quality, excluding patterns corresponding to non-fabric defects such as equipment mechanical failures and circuit anomalies. Typical visual morphological feature patterns and quantitative distribution feature patterns are accurately selected. The typical visual morphological feature patterns include edge contour standards and texture arrangement specifications corresponding to various defects. The quantitative distribution feature patterns cover core content such as the standard of the rising and falling edges of the areal density value corresponding to the defect, and the numerical stability range of the plateau trend, comprehensively covering the core feature manifestations of common fabric defects.

[0041] The fusion defect description is a collection of defect-related features, containing two core types of information: image morphological features and areal density statistical distribution features. These are systematically classified and split according to the essential differences in feature attributes. Each feature in the fusion defect description is identified individually, and features representing the visual presentation of the fabric surface are extracted separately. These features specifically involve the shape of edge contours, the arrangement of textures, etc. These features are organized and grouped by category to form a clearly structured subset of image morphological features. Simultaneously, features reflecting the dynamic changes in areal density values ​​are accurately identified, completely separating features showing a continuously increasing value (rising edge), a continuously decreasing value (falling edge), and a plateau trend that remains constant without significant fluctuations. Each feature clearly corresponds to a specific numerical change process, forming an independent subset of areal density statistical distribution features. During the splitting process, each feature is individually verified to ensure no feature omissions or misjudgments, guaranteeing the independence and integrity of the two feature subsets.

[0042] Each feature in the image morphological feature subset is compared one by one with typical visual morphological feature patterns. For edge contour features, details such as curvature, smoothness, closure, and contour size ratio are checked to see if they are completely consistent with typical patterns. For example, if the edge of a defect in the typical pattern is a smooth closed contour, the edge of the current feature is checked to see if it is free of jagged edges and breaks and forms a complete closed loop. For texture arrangement features, the direction, interweaving method, distribution density, and presence of local abnormal clusters are compared to see if they are consistent with typical patterns. Simultaneously, the numerical change trends in the subset of areal density statistical distribution features are compared with the quantitative distribution feature patterns in a refined manner. For rising edge features, it is confirmed whether the rate of increase and the overall increase or decrease are within the range specified by the typical pattern; for falling edge features, it is checked whether the rate of decrease and the total change are consistent with the standard of the typical pattern; for plateau trend features, it is verified whether the duration of numerical stability and the fluctuation range meet the requirements of the typical pattern. During the comparison process, the compliance status of each feature is recorded in detail, including the specific manifestations of complete matching, partial matching, and non-match. The comparison results of the two types of features are comprehensively summarized to form a detailed and logically clear comparison result of the nonwoven fabric production line.

[0043] Based on the detailed matching results of the two types of features with corresponding typical patterns, a comprehensive matching screening is conducted on the feature pattern combinations corresponding to various defects in the process knowledge rule base. First, the degree of matching between the subset of image morphological features and typical visual morphological feature patterns is statistically analyzed to clarify the number of matched features and the matching status of core features. Then, the matching data between the subset of areal density statistical distribution features and quantitative distribution feature patterns is statistically analyzed, including details of numerical trend matching and the matching status of key parameters. The overall matching degree of the two types of features is comprehensively calculated. Each feature pattern combination corresponding to each defect type in the knowledge base is examined one by one. Defect types that do not meet the basic standard of matching degree or whose core features do not match are eliminated. The candidate defect type with the highest matching degree is then identified, and it is further verified whether this candidate type meets the complete matching condition, i.e., all core features match, the matching degree of secondary features meets the preset requirements, and there are no key feature conflicts. Finally, the defect type corresponding to the feature pattern that fully meets the conditions is clearly identified as the defect category of the nonwoven fabric production line.

[0044] The beneficial effects include the accurate retrieval of typical visual morphological feature patterns and quantitative distribution feature patterns from the pre-set process knowledge rule base, providing a reliable and comprehensive reference for defect category determination. By separating and fusing the two feature subsets in the defect description, defect features of different dimensions are clearly presented, ensuring the targeting and accuracy of the comparison process. At the same time, the two feature subsets are compared with the corresponding typical patterns, realizing a comprehensive verification of defect features and ensuring that the comparison results can fully reflect the feature matching situation. Based on the comparison results, the process knowledge rule base is matched for consistency, which can accurately lock the defect category that best matches the actual defect features, effectively improving the accuracy and reliability of defect category determination. This provides accurate basic defect information for subsequent location of primary process links, ensuring the continuity and effectiveness of the quality monitoring process.

[0045] S5. Based on the defect category and the variation gradient and periodicity of the areal density statistical distribution characteristics along the direction of travel of the nonwoven fabric production line, the process flow diagram of the nonwoven fabric production line is mapped in reverse, and the primary process link that causes the defect category is located, so as to obtain the judgment conclusion of the nonwoven fabric production line. In this embodiment of the invention, the step of reverse mapping based on the variation gradient and periodicity of the defect category and the areal density statistical distribution characteristics along the travel direction of the nonwoven fabric production line to the process flow diagram of the nonwoven fabric production line, and locating the primary process step that triggers the defect category to obtain the judgment conclusion of the nonwoven fabric production line includes: Based on the travel direction of the nonwoven fabric production line, the numerical changes of the areal density statistical distribution characteristics are analyzed to obtain the gradient information and periodicity information of the nonwoven fabric production line. Based on the defect cause type of the defect category, the process flow diagram of the nonwoven fabric production line is topologically traced to obtain the set of candidate defect process links of the nonwoven fabric production line. Based on the change pattern formed by the gradient information and the periodic information, the candidate defect process step set is causally converged to obtain the primary process step of the nonwoven fabric production line. The defect category, the primary process step, and the change pattern are associated and encapsulated to obtain the judgment conclusion of the nonwoven fabric production line.

[0046] The process of causally converging the candidate defective process steps based on the change patterns formed by the gradient information and the periodic information to obtain the primary process steps of the nonwoven fabric production line includes: Based on the fluid dynamics and mechanical transmission principles involved in the nonwoven fabric production line, the rate of change of the gradient information and the repetition interval of the periodic information are relabeled to obtain the process state characteristics of the nonwoven fabric production line. The process state characteristics are compared with the process knowledge rule base to obtain the link matching degree of the nonwoven fabric production line. Based on the matching degree of the links, links that are strongly correlated with the process state characteristics are selected from the candidate defect process links to obtain the root cause links to be determined in the nonwoven fabric production line. Based on the failure mechanism of the defect category, the root cause of the problem is determined by matching the failure mechanism to obtain the primary process of the nonwoven fabric production line.

[0047] The formula for calculating the matching degree of the links is as follows: ; In the formula, The matching degree of the aforementioned link. The current value of the gradient information. The current value of the periodic information. The gradient typical value in the aforementioned process knowledge rule base. The typical periodic value is found in the process knowledge rule base. The preset gradient weight factor, The preset periodic weighting factor, This refers to a process step in the nonwoven fabric production line.

[0048] Along the direction of the nonwoven fabric production line, the statistical distribution characteristics of fabric density are continuously monitored, and the dynamic fluctuations of the values ​​during the production process are recorded in real time. When the values ​​transition from one stable state to another, the specific range of increase or decrease is carefully observed and recorded. At the same time, the rate of change of the values ​​is accurately captured, and both rapid abrupt changes and slow gradual changes are clearly distinguished. Based on these intuitive and specific numerical changes, the system sorts out the gradient attributes of the changes of this feature at different production stages, forming gradient information of the nonwoven fabric production line that can accurately reflect the trend of numerical changes. At the same time, during the entire monitoring process, close attention is paid to whether the numerical changes show consistent repetitive characteristics. The specific manifestation of this repetitive pattern is recorded in detail, such as whether it is a repetition of the magnitude of numerical rise and fall or a repetition of the rhythm of change. The correlation between this repetitive pattern and the production line operating status, equipment operation procedures, and other related conditions is clarified. After comprehensively collecting this information, the periodic information of the nonwoven fabric production line is obtained.

[0049] The specific cause type of defect corresponding to the defect category is clearly identified, and the inherent logic of the impact of this cause type on the nonwoven fabric production process is deeply analyzed, including the specific path of its action on materials, equipment, or process parameters. Based on this logic, a reverse tracing operation is carried out on the process flow chart of the nonwoven fabric production line. The process flow chart clearly presents the complete production links from raw material input to finished product output, as well as the sequence and relationship of each link. The tracing process starts from the terminal production link where the defect is most likely to appear, and gradually deduces towards each process link in the early stages, such as raw material preparation, web forming, and processing. For each process link, its process principle, operating specifications, equipment functions, and other mechanisms related to the cause of the defect are checked one by one to determine whether the link has the potential to cause this type of defect. All process links that have been checked and confirmed to have the potential to cause this type of defect are screened and collected, and integrated to form a clear and comprehensive set of candidate defect process links for the nonwoven fabric production line.

[0050] By deeply integrating the core characteristics of the numerical change magnitude and rate of change reflected by gradient information with the key content of the repetitive patterns and correlation conditions reflected by periodic information, a complete change pattern is constructed that can comprehensively and accurately reflect the overall changes in the statistical distribution characteristics of fabric density. Subsequently, for each process link in the candidate defect process link set, the process characteristics of the link are analyzed in depth, including its core functions, operating parameter range, and the dimensions of its impact on product quality, as well as the scope of impact that the adjustment of the operating parameters of the link can cover. Then, a causal relationship analysis between the change pattern and each candidate link is established. It is carefully judged whether the fluctuation of the operating state of each candidate link will directly lead to the occurrence of the change pattern. Through comparative analysis, links with only indirect correlation and weak causal relationship are eliminated. The focus is on links that can directly trigger the change pattern and whose process characteristics, operating logic and defect causes are highly consistent. After layers of screening and confirmation, the primary process links of the nonwoven fabric production line are obtained.

[0051] This paper deeply connects and binds the clearly identified defect categories, the primary process links precisely located through multiple rounds of screening, and the previously constructed complete change patterns. Following a logically progressive order, it systematically sorts out the inherent relationships between the three, analyzing in detail how operational anomalies in the primary process links affect the product through specific change patterns, ultimately leading to the generation of this defect category. It clarifies that the defect category originates from the primary process link, passes through the transmission and influence of the corresponding change pattern, and ultimately forms a complete causal chain. All relevant information, including defect category characteristics, the mechanism of action of the primary process link, the manifestation of the change pattern, and related conditions, is systematically integrated. Using standardized, rigorous, and easily understandable expressions, the paper summarizes and organizes this information to form a comprehensive, logically clear, and accurate judgment conclusion for the nonwoven fabric production line.

[0052] Based on the principles of fluid dynamics and mechanical transmission that govern the operation of nonwoven fabric production lines, this study delves into the rate of change of values ​​in gradient information. Fluid dynamics principles govern the flow velocity, pressure distribution, and diffusion state of materials during conveying and forming processes, while mechanical transmission principles determine the rotation speed, transmission ratio, and power transmission efficiency of each component of the equipment. By comparing the rate of change of gradient information with the smoothness of material flow and the presence of stagnation or gushing phenomena, the positive or negative correlation between the two is clarified. At the same time, considering the stability of mechanical transmission efficiency, it is determined whether the rate of change is affected by factors such as equipment speed fluctuations and wear of transmission components. This study meticulously analyzes the interval characteristics of repetitive patterns in periodic information, examining their compatibility with the rotation and reciprocating cycles of core equipment components. It also correlates these intervals with the frequency of material supply intervals and the batch cycle of raw material delivery, clarifying the intrinsic relationship between these interval characteristics and periodic parameters. These in-depth analyses serve as clear labeling criteria. The rate of change of gradient information is characterized and categorized as rapidly increasing, slowly increasing, rapidly decreasing, and slowly decreasing. Similarly, the repetition intervals of periodic information are characterized and categorized as short-cycle repetition, medium-cycle repetition, and long-cycle repetition. Ultimately, this results in a comprehensive and accurate representation of the process status characteristics of a nonwoven fabric production line, accurately reflecting the material flow and equipment operating status.

[0053] The pre-built process knowledge rule base stores a massive number of causal relationship entries between process state characteristics and process steps. These entries are based on long-term accumulated nonwoven fabric production practice data, analysis of numerous defect cases, and theoretical derivations. Each entry clearly records the causal logic between a specific process state characteristic and the corresponding process step, covering all relevant process steps that may be involved in different process states at all production stages, such as raw material preparation, web forming, hot pressing reinforcement, and winding. Using the obtained process state characteristics as core search keywords, entries are searched one by one in the process knowledge rule base according to the category labeled by the feature classification. Each matched causal relationship entry is deeply analyzed to comprehensively analyze the degree of correlation between the corresponding process step and the current process state characteristic in terms of influence path and mechanism. The degree of correlation is quantified based on dimensions such as the directness of the correlation logic, the number of matching features, and the inevitability of the impact result. For example, direct causal relationships are assigned a higher value than indirect causal relationships, and core feature matching is assigned a higher value than secondary feature matching. This quantification method yields the process matching degree of the nonwoven fabric production line, which accurately reflects the degree of correlation between the process steps and the current process state characteristics.

[0054] The degree of correlation corresponding to the matching degree of all links is clearly defined, and a unified and unique strong correlation judgment standard is set. This standard clearly stipulates that the degree of correlation must meet the requirements of "direct causal relationship + matching of all core features + inevitable occurrence of the impact result". Based on this standard, each process link in the candidate defect process link set is checked one by one in detail. The matching degree of each link is checked to see if it meets the requirements of the strong correlation judgment standard. Process links that fully meet the strong correlation standard are selected. These links have a direct and significant causal relationship with the current process state characteristics and can directly affect the formation of process state characteristics. These selected process links are systematically integrated and collected according to the sequence of the production line process flow to obtain the root cause links to be determined in the nonwoven fabric production line with a clear structure and strong targeting.

[0055] This study delves into the failure mechanisms of defect categories, dissecting the entire defect formation process to clarify the specific process conditions necessary for the defect to occur. This includes the specific ranges of parameters such as temperature, pressure, and humidity. Key manifestations of abnormal equipment operation are identified, such as component speed deviations, excessive vibration amplitude, and decreased sealing performance. Simultaneously, core requirements for material changes are clarified, such as fiber length deviations, excessive raw material humidity, and imbalanced fiber mixing ratios. Based on these clearly defined failure mechanism elements, the process functions of each potential root cause are examined to determine if they have the ability to influence the aforementioned process conditions, equipment status, and material properties. A detailed analysis is conducted to determine if the range of operational parameters for each step covers the abnormal value range required by the failure mechanism. Furthermore, the interaction between this step and other related process steps is investigated to determine if its abnormal operation will transmit through a linkage effect to the key nodes of defect formation. Finally, a precise judgment is made as to whether each potential root cause step can directly trigger the corresponding defect category through the failure mechanism. Steps where all failure mechanism elements can be met through abnormal operation and clearly lead to the defect are identified as the primary process steps in the nonwoven fabric production line.

[0056] The current value of the gradient information comes from the direction of travel along the nonwoven fabric production line. It continuously tracks the dynamic changes of the statistical distribution characteristics of the fabric density, records the numerical fluctuations at different stages of the production process in real time, analyzes the magnitude and speed of increase or decrease when the value transitions from one stable state to another, verifies the rationality of the numerical changes by combining fluid dynamics and mechanical transmission principles, eliminates accidental fluctuations caused by external interference, and accurately extracts specific values ​​that reflect the real trend of change from the results of these system analyses. This value is the current value of the gradient information.

[0057] The current value of the periodic information comes from the continuous monitoring of the numerical changes in the statistical distribution characteristics of fabric density. Various patterns appearing during the numerical changes are carefully recorded, including the repetitive patterns of numerical rises and falls, and the stability of the change intervals. Through multiple verifications, it is confirmed that these patterns are not accidental, but are necessarily related to the operating status of the production line. After clarifying the stability and correlation of the patterns, specific values ​​that can accurately represent the repetitive patterns are extracted from the monitoring data. These values ​​are the current values ​​of the periodic information.

[0058] The gradient typical values ​​in the process knowledge rule base are standard values ​​preset within the process knowledge rule base. The formation process involves collecting gradient-related data corresponding to various process links in long-term nonwoven fabric production practice, covering different gradient performances under normal and defective production conditions. Combining professional theories such as fluid dynamics and mechanical transmission, the collected data is analyzed in depth to eliminate abnormal data caused by accidental factors such as equipment failure and raw material abnormalities. The universal gradient standards corresponding to each process link under different operating conditions are summarized, and these standard values ​​are stored according to process links to form the gradient typical values ​​in the process knowledge rule base.

[0059] The typical values ​​of the cycle in the process knowledge rule base are also preset within the process knowledge rule base. Their generation is based on a comprehensive survey of the actual situation of nonwoven fabric production, collecting relevant data such as equipment operation cycle, material supply cycle, and processing cycle corresponding to different process links, combining periodic defect cases caused by various process links in production practice, verifying the rationality and correlation of the data through theoretical derivation, eliminating abnormal periodic data caused by non-process factors, summarizing the standard cycle values ​​corresponding to each process link under normal and abnormal operating conditions, and storing them in the process knowledge rule base according to process links to form typical values ​​of the cycle.

[0060] The preset gradient weight factor is a fixed value determined based on the actual impact of gradient information in the matching process of the process links. The setting process involves analyzing the directional effect of gradient information on locating the root cause of defects in the process links, such as the criticality of the gradient change rate in judging the material flow state and abnormal equipment transmission efficiency. Multiple rounds of comprehensive evaluation are conducted by technical personnel with rich production experience and professional theoretical knowledge. The accuracy of the evaluation results is verified by combining a large number of defect location cases. Finally, a fixed weight value that can objectively reflect the importance of gradient information is determined, set in advance, and applied to the calculation of the matching degree of the process links.

[0061] The preset periodic weight factor is a fixed value determined based on the degree of influence of periodic information on the matching of process links. The role of periodic information in tracing the causes of defects is set in the time series analysis, such as the importance of the periodic repetition interval to the operating cycle of related equipment and the material supply cycle. Through a professional technical team, production practice data and theoretical research are combined for comprehensive consideration. The reliability of the consideration results is ensured through multiple sets of case verifications. Finally, a fixed weight value that can accurately reflect the influence ratio of periodic information is determined, set in advance and used for the calculation of process matching degree.

[0062] The process links are the actual production steps with specific functions in the nonwoven fabric production line. These steps cover the complete production process from raw material input, fiber web formation, hot pressing reinforcement to winding and unwinding. Each link has clear process requirements and operating specifications. The process links are specific production steps selected from the set of candidate defective process links. The set of candidate defective process links is the set of links that may cause corresponding defects, which are screened by topological tracing of the process flow diagram.

[0063] The core function of this formula is to quantify the degree of correlation between process steps and the characteristics of the current process state, i.e., the degree of step matching, so as to provide an objective basis for subsequent screening of strongly correlated process steps.

[0064] By comparing the currently acquired gradient and periodic information with the typical gradient and periodic values ​​of the corresponding process steps in the process knowledge rule base one by one, the differences between the two sets of values ​​are analyzed in detail. The smaller the difference, the more closely the current process state characteristics match the standard characteristics corresponding to the process step, the more direct the impact of the process step on the current process state, and the greater the corresponding correlation contribution. Then, by using preset gradient weight factors and periodic weight factors, the correlation contribution ratios of the two are adjusted according to their importance in the matching process. Finally, the two sets of correlation contributions are integrated and summarized to form a final result that can comprehensively reflect the correlation between the process step and the current process state characteristics.

[0065] This result can accurately distinguish the strength of the correlation between different process steps and the current process state characteristics. Process steps with higher result values ​​indicate that they are strongly correlated with the current process state characteristics. This provides an objective, accurate and quantifiable basis for efficiently screening out the root causes of the process steps that are strongly correlated with the process state characteristics from the candidate defective process steps, ensuring the scientificity and reliability of the screening process and avoiding bias caused by subjective judgment.

[0066] The beneficial effects are as follows: Gradient and periodic information are obtained by analyzing the numerical changes in the statistical distribution characteristics of areal density based on the direction of travel in a nonwoven fabric production line. Combined with the defect cause type of the defect category, topological tracing of the process flow diagram is performed to accurately screen the set of candidate defective process links. Then, based on the principles of fluid dynamics and mechanical transmission, the rate of change of gradient information and the repetition interval of periodic information are re-labeled to form process state characteristics. By searching causal entries with the process knowledge rule base, the link matching degree is obtained, and strongly correlated undetermined root cause links are screened. Finally, the failure mechanism of the defect category is combined for a matching judgment to accurately locate the primary process link that caused the defect. The entire process is progressive and closely related to cause and effect, ensuring that the judgment conclusions fully cover the defect category, the primary process link, and the change pattern. This provides a clear and reliable basis for the accurate investigation and process optimization of quality problems in nonwoven fabric production lines, significantly improving the pertinence and effectiveness of quality monitoring.

[0067] S6. Integrate the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line.

[0068] In this embodiment of the invention, the integration of the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line includes: Based on the judgment conclusion, the synchronized feature data is trimmed and extracted to obtain evidence data fragments of the nonwoven fabric production line. Based on a preset report template, the defect attribute description, evidence data fragments, and identifiers of the primary process steps in the judgment conclusion are structurally filled in to obtain a draft report for the nonwoven fabric production line. The temporal and spatial positions of the evidence data fragments in the synchronized feature data are converted and encoded to obtain the data index information of the nonwoven fabric production line. The draft report is assembled with the data index information to obtain the quality monitoring report of the nonwoven fabric production line.

[0069] The core information included in the judgment conclusion, such as the defect category, the primary process step, and the change pattern, is clearly defined. Based on this information, the part of the synchronized feature data that is directly related to the judgment conclusion is identified. By defining a specific correlation range, the synchronized feature data part that can corroborate the judgment conclusion is accurately extracted, and irrelevant data content is eliminated to obtain the evidence data fragment of the nonwoven fabric production line.

[0070] The preset report template contains a fixed structured framework, which clearly defines specific areas such as a defect attribute description area, an evidence data display area, and an original process step identification area. According to the area division of the template, the attribute descriptions of the defect type and manifestation in the judgment conclusion are completely filled into the defect attribute description area, the evidence data fragments are imported into the evidence data display area, and the specific identification of the original process step is accurately entered into the original process step identification area. After completing the structured filling of all the specified content, a draft report of the nonwoven fabric production line is obtained.

[0071] The generation order of evidence data fragments in synchronized feature data is traced to determine their corresponding time positions. At the same time, the specific horizontal and vertical areas of the fabric reflected by the evidence data fragments are identified, i.e., spatial positions. Using a unified coding rule, these time and spatial position information are converted into standardized character or symbol combinations to form data index information of the nonwoven fabric production line that can quickly locate the original position of the evidence data fragments.

[0072] The data index information is linked and integrated with the draft report, and the corresponding data index information is attached to the side of the relevant evidence data fragments in the draft report. This ensures that each evidence data fragment can be quickly traced back to its original position in the synchronized feature data through the corresponding index information. This makes the content of the draft report and the data index information form a complete query system, and finally assembles a comprehensive, clearly structured, and easily traceable quality monitoring report for the nonwoven fabric production line.

[0073] The beneficial effects are as follows: Based on the judgment conclusion, synchronized feature data is trimmed and extracted to obtain evidence data fragments, ensuring that the data supporting the judgment conclusion is targeted and relevant. A draft report is formed by structurally filling in a pre-set report template, making the presentation of defect attribute descriptions, evidence data fragments, and primary process identification clear, logical, and coherent. The temporal and spatial locations of the evidence data fragments are converted and encoded to obtain data index information, providing a convenient path for quickly tracing the original source of the data. The draft report and data index information are assembled to form a quality monitoring report, ensuring that the report contains complete judgment results and supporting evidence, and also has data traceability capabilities. This comprehensively improves the completeness, accuracy, and practicality of the quality monitoring report, providing an intuitive and reliable reference for quality control and problem investigation in non-woven fabric production lines.

[0074] like Figure 2 The diagram shown is a functional block diagram of a nonwoven fabric production line quality monitoring system provided in an embodiment of the present invention.

[0075] The nonwoven fabric production line quality monitoring system 100 described in this invention can be installed in an electronic device. Depending on the functions implemented, the nonwoven fabric production line quality monitoring system 100 may include an information synchronization capture module 101, a synchronization feature generation module 102, a defect description fusion module 103, a defect category matching module 104, a primary link location module 105, and a monitoring report generation module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0076] In this embodiment, the functions of each module / unit are as follows: The information synchronization capture module 101 is used to synchronously capture linear speed information reflecting the material transport rhythm, weight information reflecting the fiber web forming quality, and appearance image information reflecting the micro-morphology of the fabric surface in the nonwoven fabric production line. The synchronization feature generation module 102 is used to perform frame alignment between the weight information and the appearance image information using the beat pulse in the linear velocity information as a synchronization reference, so as to obtain the synchronization feature data of the nonwoven fabric production line. The defect description fusion module 103 is used to fuse the texture anomaly candidate region and the basis weight information fragment in the synchronized feature data to obtain the fused defect description of the nonwoven fabric production line. The defect category matching module 104 is used to match the combination of image morphological features and surface density statistical distribution features in the fused defect description based on typical feature patterns in the preset process knowledge rule base, so as to obtain the defect category of the nonwoven fabric production line. The primary process location module 105 is used to reverse map the change gradient and periodicity of the defect category and the areal density statistical distribution characteristics along the travel direction of the nonwoven fabric production line to the process flow diagram of the nonwoven fabric production line, and locate the primary process link that causes the defect category, so as to obtain the judgment conclusion of the nonwoven fabric production line. The monitoring report generation module 106 is used to integrate the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line.

[0077] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0078] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0080] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0081] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for quality monitoring in a nonwoven fabric production line, characterized in that, The method includes: S1. Synchronously capture linear speed information reflecting the material transport rhythm, weight information reflecting the fiber web forming quality, and appearance image information reflecting the microstructure of the fabric in the nonwoven fabric production line. S2. Using the beat pulse in the linear velocity information as a synchronization reference, perform frame alignment between the weight information and the appearance image information to obtain the synchronization feature data of the nonwoven fabric production line. S3. The texture anomaly candidate region and basis weight information fragment in the synchronized feature data are fused to obtain the fusion defect description of the nonwoven fabric production line; S4. Based on the typical feature patterns in the preset process knowledge rule base, match the combination of image morphological features and surface density statistical distribution features in the fused defect description to obtain the defect category of the nonwoven fabric production line. S5. Based on the defect category and the variation gradient and periodicity of the areal density statistical distribution characteristics along the direction of travel of the nonwoven fabric production line, the process flow diagram of the nonwoven fabric production line is mapped in reverse, and the primary process link that causes the defect category is located, so as to obtain the judgment conclusion of the nonwoven fabric production line. S6. Integrate the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line.

2. The method for quality monitoring of a nonwoven fabric production line as described in claim 1, characterized in that, The synchronous capture of linear speed information reflecting the material transport rhythm, basis weight information reflecting the fiber web forming quality, and appearance image information reflecting the microstructure of the fabric surface in the nonwoven fabric production line includes: By monitoring the rotational speed pulse signal in the nonwoven fabric production line, the linear speed information of the nonwoven fabric production line can be obtained; The weight information of the nonwoven fabric production line is obtained by measuring the intensity attenuation of rays after penetrating the fiber web in the nonwoven fabric production line. Optical signals of the fabric surface are collected by scanning laterally along the nonwoven fabric production line to obtain the appearance image information of the nonwoven fabric production line.

3. The method for quality monitoring of a nonwoven fabric production line as described in claim 1, characterized in that, The synchronization feature data of the nonwoven fabric production line is obtained by using the beat pulse in the linear velocity information as a synchronization reference and performing frame alignment between the weight information and the appearance image information, including: The equally spaced beat pulses in the linear velocity information are marked as synchronization event markers for the nonwoven fabric production line; Based on the time window defined by the synchronization event marker, the weight signal values ​​of the weight information are merged to obtain the weight signal frame of the nonwoven fabric production line. Based on the time window, multiple rows of appearance image data in the appearance image information are stitched together to obtain the appearance image frame of the nonwoven fabric production line. The weight signal frame and the appearance image frame are correlated and aligned to obtain the synchronization data of the nonwoven fabric production line; Based on the timing of the synchronization event markers, the synchronization data is arranged to obtain the synchronization characteristic data of the nonwoven fabric production line.

4. The method for quality monitoring of a nonwoven fabric production line as described in claim 1, characterized in that, The process of fusing texture anomaly candidate regions and basis weight information fragments in the synchronized feature data to obtain a fusion defect description of the nonwoven fabric production line includes: Traverse the appearance image frames in the synchronized feature data to locate areas in the appearance image frames where there are obvious differences in grayscale or texture, and obtain the texture anomaly candidate areas of the nonwoven fabric production line. Based on the position coordinates of the texture anomaly candidate region in the appearance image frame, trace the abnormal weight information in the synchronized feature data to obtain the weight information fragment of the nonwoven fabric production line. Visual morphological recognition is performed on the edge contours and texture arrangement of the candidate regions for texture anomalies to obtain the morphological feature set of the nonwoven fabric production line; Extract the rising edge, falling edge, and plateau trend of the numerical change in the basis weight information segment to obtain the distribution feature set of the nonwoven fabric production line; The morphological feature set and the distribution feature set are combined side by side to obtain the fusion defect description of the nonwoven fabric production line.

5. The method for quality monitoring of a nonwoven fabric production line as described in claim 1, characterized in that, The method of matching the combination of image morphological features and areal density statistical distribution features in the fused defect description based on typical feature patterns in the preset process knowledge rule base obtains the defect categories of the nonwoven fabric production line, including: The typical visual morphological feature patterns and quantitative distribution feature patterns of the nonwoven fabric production line are retrieved from the preset process knowledge rule base. Separate the image morphological feature subset and the areal density statistical distribution feature subset from the description of the fusion defect; The image morphological feature subset is compared with the typical visual morphological feature pattern, and the areal density statistical distribution feature subset is compared with the quantitative distribution feature pattern to obtain the comparison result of the nonwoven fabric production line. Based on the comparison results, the process knowledge rule base is matched to obtain the defect categories of the nonwoven fabric production line.

6. The method for quality monitoring of a nonwoven fabric production line as described in claim 1, characterized in that, The process involves inversely mapping the gradient and periodicity of the defect category and the areal density statistical distribution characteristics along the direction of travel of the nonwoven fabric production line to the process flow diagram of the nonwoven fabric production line, locating the primary process step that triggers the defect category, and obtaining the judgment conclusion of the nonwoven fabric production line, including: Based on the travel direction of the nonwoven fabric production line, the numerical changes of the areal density statistical distribution characteristics are analyzed to obtain the gradient information and periodicity information of the nonwoven fabric production line. Based on the defect cause type of the defect category, the process flow diagram of the nonwoven fabric production line is topologically traced to obtain the set of candidate defect process links of the nonwoven fabric production line. Based on the change pattern formed by the gradient information and the periodic information, the candidate defect process step set is causally converged to obtain the primary process step of the nonwoven fabric production line. The defect category, the primary process step, and the change pattern are associated and encapsulated to obtain the judgment conclusion of the nonwoven fabric production line.

7. The method for quality monitoring of a nonwoven fabric production line as described in claim 6, characterized in that, The process of causally converging the candidate defective process steps based on the change patterns formed by the gradient information and the periodic information to obtain the primary process steps of the nonwoven fabric production line includes: Based on the fluid dynamics and mechanical transmission principles involved in the nonwoven fabric production line, the rate of change of the gradient information and the repetition interval of the periodic information are relabeled to obtain the process state characteristics of the nonwoven fabric production line. The process state characteristics are compared with the process knowledge rule base to obtain the link matching degree of the nonwoven fabric production line. Based on the matching degree of the links, links that are strongly correlated with the process state characteristics are selected from the candidate defect process links to obtain the root cause links to be determined in the nonwoven fabric production line. Based on the failure mechanism of the defect category, the root cause of the problem is determined by matching the failure mechanism to obtain the primary process of the nonwoven fabric production line.

8. The method for quality monitoring of a nonwoven fabric production line as described in claim 7, characterized in that, The formula for calculating the matching degree of the links is as follows: ； In the formula, The matching degree of the aforementioned link. The current value of the gradient information. The current value of the periodic information. The gradient typical value in the aforementioned process knowledge rule base. The typical periodic value is found in the process knowledge rule base. The preset gradient weight factor, The preset periodic weighting factor, This refers to a process step in the nonwoven fabric production line.

9. The method for quality monitoring of a nonwoven fabric production line as described in claim 1, characterized in that, The quality monitoring report of the nonwoven fabric production line is obtained by integrating the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step, including: Based on the judgment conclusion, the synchronized feature data is trimmed and extracted to obtain evidence data fragments of the nonwoven fabric production line. Based on a preset report template, the defect attribute description, evidence data fragments, and identifiers of the primary process steps in the judgment conclusion are structurally filled in to obtain a draft report for the nonwoven fabric production line. The temporal and spatial positions of the evidence data fragments in the synchronized feature data are converted and encoded to obtain the data index information of the nonwoven fabric production line. The draft report is assembled with the data index information to obtain the quality monitoring report of the nonwoven fabric production line.

10. A quality monitoring system for a nonwoven fabric production line, characterized in that, The system for implementing the nonwoven fabric production line quality monitoring method according to claim 1 includes: The information synchronization and capture module is used to synchronously capture linear speed information reflecting the material conveying rhythm, weight information reflecting the fiber web forming quality, and appearance image information reflecting the micro-morphology of the fabric surface in the nonwoven fabric production line. The synchronization feature generation module is used to perform frame alignment between the weight information and the appearance image information using the beat pulse in the linear velocity information as a synchronization reference, so as to obtain the synchronization feature data of the nonwoven fabric production line. The defect description fusion module is used to fuse the texture anomaly candidate regions and basis weight information fragments in the synchronized feature data to obtain the fused defect description of the nonwoven fabric production line. The defect category matching module is used to match the combination of image morphological features and surface density statistical distribution features in the fused defect description based on typical feature patterns in the preset process knowledge rule base, so as to obtain the defect category of the nonwoven fabric production line. The primary process location module is used to reverse map the change gradient and periodicity of the defect category and the areal density statistical distribution characteristics along the travel direction of the nonwoven fabric production line to the process flow diagram of the nonwoven fabric production line, and locate the primary process link that causes the defect category, so as to obtain the judgment conclusion of the nonwoven fabric production line. The monitoring report generation module is used to integrate the judgment conclusion, the fragments of the synchronized feature data, and the identifier of the primary process step to obtain the quality monitoring report of the nonwoven fabric production line.

Images