Monitoring system and method applied to carbon fiber production line

Abstract

The invention provides a monitoring system and method applied to a carbon fiber production line, and the system comprises a data collection module which collects process related data and product image data; the data processing module is used for preprocessing the data and matching and marking the data according to production batches; the big data module is used for analyzing the processed process related data based on a big data technology; the data analysis module performs defect analysis based on a machine vision technology and a fuzzy clustering algorithm, and performs association rule mining based on the big data module through an Apriori algorithm; and the early warning module is used for receiving the abnormal data and respectively carrying out early warning prompt. Defect analysis is carried out based on the machine vision technology and the fuzzy clustering algorithm, association rule mining is carried out through the Apriori algorithm, the hidden association relationship between the product defects and the production process is accurately found, the carbon fiber production line can be integrally monitored, and the production quality of the carbon fiber production line is improved.

Description

Technical Field

[0001] The present application relates to the technical field of production line monitoring, and in particular to a monitoring system and method applied to a carbon fiber production line. Background Art

[0002] Carbon fiber materials have the advantages of high strength and light weight in industry and are widely used in aerospace, automobile manufacturing and other fields. However, during the production process, carbon fiber tows are prone to defects such as pilling, hair, long and short filaments, and broken filaments. If the defects are not detected, the roll of carbon fiber tow products will be scrapped, resulting in huge economic losses, and the quality of products made with scrapped carbon fiber tows will also be affected to a certain extent. The existing carbon fiber tow defect detection technology is mainly based on manual inspection. Workers stand under the carbon fiber tow production line for visual inspection, and the production line works 24 hours a day. There are many carbon fiber tows on a production line, and several workers are required to perform visual inspections in different areas. It is difficult to ensure the quality of the carbon fiber tows. At the same time, the labor cost is high, time-consuming and labor-intensive, and it is difficult to meet the needs of large-scale production. In addition, the carbon fiber filaments on the production line are easy to float in the air, and human contact or inhalation will cause certain discomfort symptoms.

[0003] In the prior art, a more novel method is to use deep learning for detection, but deep learning requires the collection of a large number of defect sample data sets. It is difficult to establish a large sample defect data set, which requires a lot of time, energy, and money, and the development cycle is long. If new defects appear, the deep learning method needs to collect a large number of defect data sets and retrain, resulting in an increase in labor costs. Deep networks require high-end GPUs to train in a reasonable time with a large amount of data. These GPUs are very expensive, but without them, it is not practical to train deep networks to achieve high performance. To effectively use such high-end GPUs, fast CPUs, SSD storage, and fast and large-capacity RAM are also required.

[0004] However, the above methods are all aimed at product defect detection, which is to detect problems in the final products. They all have universality problems and cannot be generalized. In addition, there is a lack of overall monitoring of the carbon fiber production line. When product defects occur, it is impossible to accurately locate the production process that may cause the defects.

[0005] Application Contents

[0006] The present application is made in view of the above-mentioned issues, and its purpose is to provide a monitoring system and method for a carbon fiber production line, to detect abnormal data in the production process through big data technology, to perform defect analysis on the image data of the carbon fiber based on machine vision technology and fuzzy clustering algorithm, and finally to mine the association rules of product defects and abnormal data based on the Apriori algorithm based on the big data module, to accurately find the hidden association between product defects and the production process, to promptly check the production process with the association when defects occur, to be able to monitor the carbon fiber production line as a whole, to prevent accidents in the production process, to promptly deal with problems when they occur, and to improve the production quality of the carbon fiber production line.

[0007] Specifically, the first aspect of the present application provides a monitoring system for a carbon fiber production line, comprising:

[0008] A data acquisition module is used to collect process-related data of each carbon fiber production process and image data of the produced carbon fiber;

[0009] The data processing module is used to pre-process all the data collected by the data collection module and match and mark the data according to production batches;

[0010] The big data module is used to analyze the processed process-related data based on big data technology and filter the abnormal data to obtain the normal range of the corresponding data;

[0011] The data analysis module is used to perform defect analysis on the image data of carbon fiber based on machine vision technology and fuzzy clustering algorithm, and then filter the abnormal data in the process-related data with the same mark through the big data module. Finally, the Apriori algorithm is used to mine the association rules between product defects and abnormal data based on the big data module to find the abnormal data of the production process that matches the product defects and send them to the early warning module;

[0012] The early warning module is used to receive product defect data and corresponding production process abnormal data and issue early warning prompts respectively.

[0013] Furthermore, the process-related data specifically include the type of raw material, oxidation time, oxygen concentration, temperature at the oxidation process collection time point, mass before carbonization, mass after carbonization, electrolyte pH value, and processing current value.

[0014] According to the different types of raw materials, the carbon fibers produced include polyacrylonitrile-based carbon fibers, asphalt-based carbon fibers, and viscose-based carbon fibers; the oxidation time, oxygen concentration, and temperature at the time point of the oxidation process collection are relevant data in the pre-oxidation process of the carbon fibers; the mass before carbonization and the mass after carbonization are relevant data in the carbonization process; the electrolyte pH value and the treatment current value are relevant data in the surface treatment process.

[0015] Pre-oxidation process: The carbon fiber precursor is pre-oxidized in the air at a temperature between 200-300°C to rearrange the fiber atomic structure and pick up oxygen molecules in the air.

[0016] Carbonization process: The pre-oxidized fiber enters the carbonization furnace and undergoes high-temperature carbonization treatment under the protection of inert gas at a temperature range of 700-1800°C, ultimately obtaining a carbon material with a carbon content greater than 90%.

[0017] Surface treatment process: The carbonized fiber needs to be surface treated and sizing to improve its performance and application adaptability.

[0018] Furthermore, the preprocessing includes performing missing value processing and data normalization on process-related data, and removing noise from image data.

[0019] Missing value processing is to fill the small amount of missing data in a type of process data with the median value, and directly delete the data with missing values ​​exceeding 50%; data normalization is to scale the data to between 0 and 1 to eliminate the impact of different dimensions.

[0020] Furthermore, the big data technology is used to analyze the processed process-related data, and the abnormal data therein is screened to obtain the normal range of the corresponding data, specifically including:

[0021] Step 1: Collect production process-related data from the company's internal database, the Internet, and carbon fiber production standards;

[0022] Step 2: Preprocess the collected data and perform hashing to remove duplicates;

[0023] Step 3: Classify and organize the processed data and store it in the cloud database as a data source;

[0024] Step 4: Screen out abnormal data based on the interquartile range method for similar data of the data source and determine the normal range of the corresponding data.

[0025] Furthermore, the defect analysis of the carbon fiber image data based on machine vision technology and fuzzy clustering algorithm specifically includes:

[0026] Step 1: Perform grayscale enhancement, image enhancement and edge detection on the preprocessed image data;

[0027] Step 2: Apply photometric stereo to the image data to calculate the gradient matrix of the image;

[0028] Step 3: Input the gradient matrix of the image into the fuzzy C-means clustering algorithm and perform clustering analysis using pixels as data points to distinguish normal areas from areas with possible defects.

[0029] Step 4: Segment the image according to the division between normal areas and areas with possible defects;

[0030] Step 5: Extract features from images of areas where defects may exist, and identify specific defect types based on feature differences.

[0031] Furthermore, the defect types include pilling defects, hairy wire defects, long and short wire defects, and broken wire defects.

[0032] Pilling defect: Pilling refers to the loose aggregation of rovings on carbon fiber yarns, which are usually identified by manual observation or weighing to detect whether there are broken filaments.

[0033] Hair defects: Hair is a thin and long fiber attached to the surface of carbon fiber. Machine vision technology can be used to monitor the hair on each carbon fiber yarn in real time, and save defect images and related information.

[0034] Long and short filament defects: Long and short filaments refer to the inconsistency of the length of the carbon fiber yarn. Although the specific detection method is not described in detail in the information I searched, it can be inferred that through image processing and deep learning algorithms, areas of different lengths can be identified and classified.

[0035] Broken wire defect: Broken wire refers to the phenomenon of local damage or breakage in carbon fiber yarn. After identifying each carbon fiber area, the length of their minimum circumscribed rectangle can be calculated. If the length of a certain area is shorter than expected, it is judged as a broken wire defect.

[0036] Furthermore, the association rule mining of product defects and abnormal data based on the big data module by using the Apriori algorithm specifically includes:

[0037] Step 1: Each image is regarded as a "transaction", and the image features, defect types and corresponding abnormal data types are regarded as an "item";

[0038] Step 2: Set the minimum support threshold and the minimum confidence threshold, and use the Apriori algorithm to search for frequent itemsets layer by layer;

[0039] Step 3: Generate frequent item sets whose support is greater than the minimum support threshold after iteration;

[0040] Step 4: Mining strong association rules with confidence greater than the minimum confidence threshold based on frequent item sets.

[0041] A second aspect of the present application provides a monitoring method applied to a carbon fiber production line, comprising the following steps:

[0042] Step 1: Collect process-related data of each carbon fiber production process and image data of the produced carbon fiber;

[0043] Step 2: Pre-process all data collected by the data acquisition module and match and mark the data by production batch;

[0044] Step 3: Perform abnormal analysis on the processed process-related data based on big data technology, and perform defect analysis on the carbon fiber image data based on machine vision technology and fuzzy clustering algorithm;

[0045] Step 4: Set the minimum support threshold and the minimum confidence threshold, and use the Apriori algorithm to mine association rules of product defects and abnormal data;

[0046] Step 5: Issue early warnings for product defects in the production process and the production process where the matching abnormal data is located.

[0047] Furthermore, the minimum support threshold is set to 30% to 70%.

[0048] Furthermore, the minimum confidence threshold is set to 50% to 80%.

[0049] In association rule mining, the setting of minimum support and minimum confidence are key parameters, which directly affect the quality and relevance of mining results.

[0050] The minimum support refers to the frequency threshold of the item set in all transactions. It is used to measure the minimum importance that the rule needs to meet, that is, the frequency of an item set in the database must exceed this threshold to be considered a frequent item set.

[0051] The minimum confidence level refers to the probability threshold of another item occurring when a certain item occurs. It indicates the minimum reliability that the association rule needs to meet, that is, in all transactions containing the item, the ratio of the item and other items occurring at the same time must exceed this threshold.

[0052] In the second aspect, the present application also provides a computing device, which has the function of implementing the method described in the first aspect above. The beneficial effects can be found in the description of the first aspect, which will not be repeated here. The function can be implemented by hardware, or by executing the corresponding software implementation through hardware. The hardware or software includes one or more modules corresponding to the above functions. In one possible design, the structure of the device includes an acquisition module, a training module, and optionally, a construction module. These modules can implement the function of the training node in the method example of the first aspect above. Please refer to the detailed description in the method example for details, which will not be repeated here.

[0053] In the third aspect, the present application also provides a computing device, which is used to implement the functions of the method described in the first aspect above. The beneficial effects can be found in the description of the first aspect and will not be repeated here. The structure of the computing device includes a processor and a memory, and the memory is used to store instructions and / or data. The memory is coupled to the processor, and when the processor executes the program instructions stored in the memory, the function of the training node in the example of the first aspect above can be implemented. The structure of the computing device also includes a communication interface for communicating with other devices.

[0054] In a fourth aspect, the present application also provides a computer-readable storage medium, in which instructions are stored, and when the computer-readable storage medium is run on a computer, the computer executes the method in the above-mentioned first aspect and various possible designs of the first aspect.

[0055] In a fifth aspect, the present application also provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the method in the above-mentioned first aspect and various possible designs of the first aspect.

[0056] In a sixth aspect, the present application also provides a computing chip, which is connected to a memory, and the chip is used to read and execute a software program stored in the memory, and to execute the methods in the above-mentioned first aspect and various possible implementation methods of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present drawings or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present drawings. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0058] Figure 1 It is a structural block diagram of a monitoring system applied to a carbon fiber production line;

[0059] Figure 2 The present invention is a flowchart of the steps of a monitoring method applied to a carbon fiber production line.

[0060] The purpose, features and advantages of this figure will be further described in conjunction with the embodiments with reference to the accompanying drawings. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0062] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0063] Unless otherwise specified, all embodiments and optional embodiments of the present application can be combined with each other to form a new technical solution.

[0064] Unless otherwise specified, all technical features and optional technical features of this application can be combined with each other to form a new technical solution.

[0065] If there is no special explanation, all steps of the present application can be performed sequentially or randomly, preferably sequentially. For example, the method includes steps (a) and (b), which means that the method may include steps (a) and (b) performed sequentially, or may include steps (b) and (a) performed sequentially. For example, the method may further include step (c), which means that step (c) may be added to the method in any order. For example, the method may include steps (a), (b) and (c), or may include steps (a), (c) and (b), or may include steps (c), (a) and (b), etc.

[0066] If there is no special explanation, the "include" and "comprising" mentioned in this application are open-ended or closed-ended. For example, the "include" and "comprising" may mean that other components not listed may also be included or only the listed components may be included or only the listed components may be included.

[0067] If not specifically stated, in this application, the term "or" is inclusive. For example, the phrase "A or B" means "A, B, or both A and B". More specifically, any of the following conditions satisfies the condition "A or B": A is true (or exists) and B is false (or does not exist); A is false (or does not exist) and B is true (or exists); or both A and B are true (or exist).

[0068] In order to better understand the solutions of the embodiments of the present application, some relevant terms and concepts that may be involved in the embodiments of the present application are first introduced below.

[0069] (1) Machine vision is an interdisciplinary field that combines computer vision and image processing technology. It aims to enable computers to simulate human visual functions and achieve deep perception, understanding and analysis of image and video content. It obtains image information through cameras or other image acquisition devices, and then uses computer vision algorithms to analyze and process the images. The key steps of machine vision technology include image acquisition, preprocessing, feature extraction, object detection and recognition, etc. These steps work together to enable machines to automatically extract information from images and make corresponding decisions and operations.

[0070] (2) Fuzzy clustering algorithm is a partition-based clustering method. Its core idea is to classify data by calculating the degree of membership between data points and cluster centers. This algorithm allows each data point to partially belong to multiple clusters, rather than belonging to only one cluster like traditional hard clustering. It usually uses the concept and operation of fuzzy sets to perform data clustering analysis.

[0071] (3) Fuzzy C-means Clustering (FCM) is a fuzzy clustering algorithm that determines the degree of membership of each data point to each cluster center by minimizing an objective function. This objective function takes into account the weight and distance of each object and requires a value m greater than or equal to 1 to adjust the fuzziness and clarity of the clustering result.

[0072] (4) Apriori algorithm is a classic association rule mining algorithm, which is mainly used to discover frequent item sets and association rules generated based on these frequent item sets from large databases. Its core idea is to use known frequent item sets (i.e., prior knowledge) to more effectively find larger frequent item sets. Specifically, Apriori algorithm mines frequent item sets through an iterative method of layer-by-layer search. First, the algorithm scans all transactions to find the frequencies of all single items, then generates all possible two item sets and calculates their frequencies. Next, the algorithm filters out those item sets that do not meet the conditions according to a certain support threshold, and continues to generate new candidate item sets and verify them. This process is repeated until no new frequent item sets are found. In each step, the algorithm uses a technique called "pruning" to reduce unnecessary calculations. Apriori algorithm mines frequent item sets through two stages: candidate set generation and downward closure detection of the plot. The process includes connection (matrix-like operation) and pruning (removing item sets that do not meet the conditions), and finally forms association rules.

[0073] In this embodiment, Figure 1 As shown, a monitoring system applied to a carbon fiber production line comprises:

[0074] A data acquisition module is used to collect process-related data of each carbon fiber production process and image data of the produced carbon fiber;

[0075] The data processing module is used to pre-process all the data collected by the data collection module and match and mark the data according to production batches;

[0076] The big data module is used to analyze the processed process-related data based on big data technology and filter the abnormal data to obtain the normal range of the corresponding data;

[0077] The data analysis module is used to perform defect analysis on the image data of carbon fiber based on machine vision technology and fuzzy clustering algorithm, and then filter the abnormal data in the process-related data with the same mark through the big data module. Finally, the Apriori algorithm is used to mine the association rules between product defects and abnormal data based on the big data module to find the abnormal data of the production process that matches the product defects and send them to the early warning module;

[0078] The early warning module is used to receive product defect data and corresponding production process abnormal data and issue early warning prompts respectively.

[0079] Furthermore, the process-related data specifically include the type of raw materials, oxidation time, oxygen concentration, temperature at the oxidation process collection time point, mass before carbonization, mass after carbonization, electrolyte pH value, and processing current value.

[0080] According to the different types of raw materials, the carbon fibers produced include polyacrylonitrile-based carbon fibers, asphalt-based carbon fibers, and viscose-based carbon fibers; the oxidation time, oxygen concentration, and temperature at the time point of the oxidation process collection are relevant data in the pre-oxidation process of the carbon fibers; the mass before carbonization and the mass after carbonization are relevant data in the carbonization process; the electrolyte pH value and the treatment current value are relevant data in the surface treatment process.

[0081] Pre-oxidation process: The carbon fiber precursor is pre-oxidized in the air at a temperature between 200-300°C to rearrange the fiber atomic structure and pick up oxygen molecules in the air.

[0082] Carbonization process: The pre-oxidized fiber enters the carbonization furnace and undergoes high-temperature carbonization treatment under the protection of inert gas at a temperature range of 700-1800°C, ultimately obtaining a carbon material with a carbon content greater than 90%.

[0083] Surface treatment process: The carbonized fiber needs to be surface treated and sizing to improve its performance and application adaptability.

[0084] Furthermore, the preprocessing includes missing value processing and data normalization for process-related data, and noise removal for image data.

[0085] Missing value processing is to fill the small amount of missing data in a type of process data with the median value, and directly delete the data with missing values ​​exceeding 50%; data normalization is to scale the data to between 0 and 1 to eliminate the impact of different dimensions.

[0086] Furthermore, the processed process-related data is analyzed based on big data technology, and the abnormal data is screened to obtain the normal range of the corresponding data, including:

[0087] Step 1: Collect production process-related data from the company's internal database, the Internet, and carbon fiber production standards;

[0088] Step 2: Preprocess the collected data and perform hashing to remove duplicates;

[0089] Data preprocessing includes filling missing values, smoothing noisy data, and removing outliers. Hash deduplication is a method that uses a hash algorithm to map data to a unique hash value to achieve deduplication. Specifically, this method converts each record in the data set into a unique hash value through a hash algorithm, and treats data with the same hash value as duplicate data to quickly determine the repetitiveness of the data.

[0090] Step 3: Classify and organize the processed data and store it in the cloud database as a data source;

[0091] Classification and organization is to organize the data of the same production process together as an analysis database for this production process.

[0092] Step 4: Screen out abnormal data based on the interquartile range method for similar data of the data source and determine the normal range of the corresponding data.

[0093] The interquartile range method is a statistical measure of data variability, which is achieved by dividing the data set into four equal parts. The difference IQR between the first quartile (Q1) and the third quartile (Q3) of the data set is calculated. Any data point less than Q1-1.5*IQR or greater than Q3+1.5*IQR is considered an outlier.

[0094] Furthermore, the defect analysis of carbon fiber image data is performed based on machine vision technology and fuzzy clustering algorithm, including:

[0095] Step 1: Perform grayscale enhancement, image enhancement and edge detection on the preprocessed image data;

[0096] Grayscale enhancement refers to improving the visual effect of an image by adjusting the grayscale value of the image; image enhancement aims to improve the visual effect of an image or provide better image quality for subsequent processing. Common methods include brightness adjustment, contrast enhancement, and noise removal; edge detection is a key step in image analysis and processing, and is used to detect the boundaries between objects and regions in an image. This embodiment uses the Sobel operator, which is a gradient-based edge detection algorithm that detects edges by calculating the gradient of the image in the horizontal and vertical directions.

[0097] Step 2: Apply photometric stereo to the image data to calculate the gradient matrix of the image;

[0098] In this embodiment, the grayscale image is processed using the Sobel operator to calculate the gradient strength and direction of each pixel. The square of the gradient strength of a point is equal to the sum of the squares of the gradient values ​​in the horizontal and vertical directions of the point. The gradient strength is then normalized for the lighting effect of the image in the light change and shadow part.

[0099] Step 3: Input the gradient matrix of the image into the fuzzy C-means clustering algorithm and perform clustering analysis using pixels as data points to distinguish normal areas from areas with possible defects.

[0100] By taking advantage of the fuzzy C-means clustering in image segmentation, more original image information can be retained, and each data point can be determined to belong to a certain cluster through membership, thereby achieving fine segmentation and defect detection of the image. The image is segmented, extracted and binarized through fuzzy C-means clustering, and then the various regions are superimposed to make the edges of the image continuous, thereby realizing defect identification.

[0101] Step 4: Segment the image according to the division between normal areas and areas with possible defects;

[0102] Step 5: Extract features from images of areas where defects may exist, and identify specific defect types based on feature differences.

[0103] For images with possible defective areas, the height, length and rectangularity are extracted based on the edge segmentation in image processing. If the rectangularity is smaller than the normal range, it is judged as a hair ball defect; if the length difference is smaller than the normal range, it is judged as a long-short wire defect; if the height is smaller than the normal range, it is judged as a hair wire defect; if the height of the connected area is lower than the minimum value of the broken wire, it is judged as a broken wire defect.

[0104] Furthermore, the defect types include pilling defects, hairy wire defects, long and short wire defects, and broken wire defects.

[0105] Pilling defect: Pilling refers to the loose aggregation of rovings on carbon fiber yarns, which are usually identified by manual observation or weighing to detect whether there are broken filaments.

[0106] Hair defects: Hair is a thin and long fiber attached to the surface of carbon fiber. Machine vision technology can be used to monitor the hair on each carbon fiber yarn in real time, and save defect images and related information.

[0107] Long and short filament defects: Long and short filaments refer to the inconsistency of the length of the carbon fiber yarn. Although the specific detection method is not described in detail in the information I searched, it can be inferred that through image processing and deep learning algorithms, areas of different lengths can be identified and classified.

[0108] Broken wire defect: Broken wire refers to the phenomenon of local damage or breakage in carbon fiber yarn. After identifying each carbon fiber area, the length of their minimum circumscribed rectangle can be calculated. If the length of a certain area is shorter than expected, it is judged as a broken wire defect.

[0109] Furthermore, the Apriori algorithm is used to mine association rules of product defects and abnormal data based on the big data module, including:

[0110] Step 1: Each image is regarded as a "transaction", and the image features, defect types and corresponding abnormal data types are regarded as an "item";

[0111] Step 2: Set the minimum support threshold and the minimum confidence threshold, and use the Apriori algorithm to search for frequent itemsets layer by layer;

[0112] Step 3: Generate frequent item sets whose support is greater than the minimum support threshold after iteration;

[0113] Step 4: Mining strong association rules with confidence greater than the minimum confidence threshold based on frequent item sets.

[0114] In this embodiment, Figure 2 As shown, a monitoring method applied to a carbon fiber production line comprises the following steps:

[0115] Step 1: Collect process-related data of each carbon fiber production process and image data of the produced carbon fiber;

[0116] Step 2: Pre-process all data collected by the data acquisition module and match and mark the data by production batch;

[0117] Step 3: Perform abnormal analysis on the processed process-related data based on big data technology, and perform defect analysis on the carbon fiber image data based on machine vision technology and fuzzy clustering algorithm;

[0118] Step 4: Set the minimum support threshold and the minimum confidence threshold, and use the Apriori algorithm to mine association rules of product defects and abnormal data;

[0119] Step 5: Issue early warnings for product defects in the production process and the production process where the matching abnormal data is located.

[0120] In this embodiment, if the support degree of the association rule between a product defect and abnormal data in the production process is greater than 50% and the confidence degree is greater than 70%, it is judged that the association rule is meaningful, and it is judged that there is an association relationship between the product defect and the abnormal data, and early warnings are issued for the production processes where the product defect and the abnormal data are located.

[0121] In this embodiment, the correlation relationships with support greater than 50% and confidence greater than 70% include the correlation between the pilling defect and the pre-oxidation process and the carbonization process, the correlation between the hairy fiber defect and the carbonization process, the correlation between the long and short fiber defect and the raw material spinning process, and the correlation between the broken fiber defect and the raw material quality.

[0122] Furthermore, the minimum support threshold is set to 50%.

[0123] Furthermore, the minimum confidence threshold is set to 70%.

[0124] In association rule mining, the setting of minimum support and minimum confidence are key parameters, which directly affect the quality and relevance of mining results.

[0125] The minimum support refers to the frequency threshold of the item set in all transactions. It is used to measure the minimum importance that the rule needs to meet, that is, the frequency of an item set in the database must exceed this threshold to be considered a frequent item set.

[0126] The minimum confidence level refers to the probability threshold of another item occurring when a certain item occurs. It indicates the minimum reliability that the association rule needs to meet, that is, in all transactions containing the item, the ratio of the 0 item and other items occurring simultaneously must exceed this threshold.

[0127] In this embodiment, a higher minimum support is set to ensure that the discovered rules have strong market universality, and a higher confidence level is set to ensure the accuracy and reliability of rule mining.

[0128] It should be noted that the present application is not limited to the above-mentioned embodiments. The above-mentioned embodiments are only examples, and the embodiments having the same structure as the technical idea and exerting the same effect within the scope of the technical solution of the present application are all included in the technical scope of the present application. In addition, without departing from the scope of the main purpose of the present application, various modifications that can be thought of by those skilled in the art to the embodiments and other methods of combining some of the constituent elements in the embodiments are also included in the scope of the present application.

Claims

1. A monitoring system for a carbon fiber production line, characterized in that: include: A data acquisition module is used to collect process-related data of each carbon fiber production process and image data of the produced carbon fiber; The data processing module is used to pre-process all the data collected by the data collection module and match and mark the data according to production batches; The big data module is used to analyze the processed process-related data based on big data technology and filter the abnormal data to obtain the normal range of the corresponding data; The data analysis module is used to perform defect analysis on the image data of carbon fiber based on machine vision technology and fuzzy clustering algorithm, and then filter the abnormal data in the process-related data with the same mark through the big data module. Finally, the Apriori algorithm is used to mine the association rules between product defects and abnormal data based on the big data module to find the abnormal data of the production process that matches the product defects and send them to the early warning module; The early warning module is used to receive product defect data and corresponding production process abnormal data and issue early warning prompts respectively.

2. A monitoring system for a carbon fiber production line according to claim 1, characterized in that: The process-related data specifically include the type of raw material, oxidation time, oxygen concentration, temperature at the oxidation process collection time point, mass before carbonization, mass after carbonization, electrolyte pH value, and processing current value.

3. The monitoring system for a carbon fiber production line according to claim 1, characterized in that: The preprocessing includes performing missing value processing and data normalization on process-related data, and removing noise from image data.

4. The monitoring system for a carbon fiber production line according to claim 1, characterized in that: The method of analyzing the processed process-related data based on big data technology and screening the abnormal data therein to obtain the normal range of the corresponding data specifically includes: Step 1: Collect production process-related data from the company's internal database, the Internet, and carbon fiber production standards; Step 2: Preprocess the collected data and perform hashing to remove duplicates; Step 3: Classify and organize the processed data and store it in the cloud database as a data source; Step 4: Screen out abnormal data based on the interquartile range method for similar data of the data source and determine the normal range of the corresponding data.

5. The monitoring system for a carbon fiber production line according to claim 1, characterized in that: The defect analysis of carbon fiber image data based on machine vision technology and fuzzy clustering algorithm specifically includes: Step 1: Perform grayscale enhancement, image enhancement and edge detection on the preprocessed image data; Step 2: Apply photometric stereo to the image data to calculate the gradient matrix of the image; Step 3: Input the gradient matrix of the image into the fuzzy C-means clustering algorithm and perform clustering analysis using pixels as data points to distinguish normal areas from areas with possible defects. Step 4: Segment the image according to the division between normal areas and areas with possible defects; Step 5: Extract features from images of areas where defects may exist, and identify specific defect types based on feature differences.

6. The monitoring system for a carbon fiber production line according to claim 5, characterized in that: The defect types include hair ball defects, hairy wire defects, long and short wire defects, and broken wire defects.

7. The monitoring system for a carbon fiber production line according to claim 1, characterized in that: The method of mining association rules of product defects and abnormal data based on the big data module by using the Apriori algorithm specifically includes: Step 1: Each image is regarded as a "transaction", and the image features, defect types and corresponding abnormal data types are regarded as an "item"; Step 2: Set the minimum support threshold and the minimum confidence threshold, and use the Apriori algorithm to search for frequent itemsets layer by layer; Step 3: Generate frequent item sets whose support is greater than the minimum support threshold after iteration; Step 4: Mining strong association rules with confidence greater than the minimum confidence threshold based on frequent item sets.

8. A monitoring method applied to a carbon fiber production line, characterized in that: The following steps are involved: Step 1: Collect process-related data of each carbon fiber production process and image data of the produced carbon fiber; Step 2: Pre-process all data collected by the data acquisition module and match and mark the data by production batch; Step 3: Perform abnormal analysis on the processed process-related data based on big data technology, and perform defect analysis on the carbon fiber image data based on machine vision technology and fuzzy clustering algorithm; Step 4: Set the minimum support threshold and the minimum confidence threshold, and use the Apriori algorithm to mine association rules of product defects and abnormal data; Step 5: Issue early warnings for product defects in the production process and the production process where the matching abnormal data is located.

9. A monitoring method for a carbon fiber production line according to claim 8, characterized in that: The minimum support threshold is set to 30% to 70%.

10. The monitoring method for a carbon fiber production line according to claim 8, characterized in that: The minimum confidence threshold is set to 50% to 80%.

Images