Intelligent management and control method and system for carbon fiber product production line
By using intelligent management and control methods for carbon fiber product production lines, key processes are identified and merged, and compensation adjustments are made based on process combinability. This solves the problem of poor production line connectivity and improves production efficiency and quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-07
- Publication Date
- 2026-04-03
AI Technical Summary
In existing carbon fiber product production lines, the connections between various processes are not smooth, resulting in low production efficiency and a backlog of defective products, which affects overall production efficiency.
By analyzing historical defective product data, important processes are identified and adjacent processes are merged. Compensation and adjustments are made based on process combinability to optimize the production line process.
This improved process stability, reduced the occurrence of defective products, and achieved a dual improvement in quality and efficiency.
Smart Images
Figure CN121787859A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing, and specifically to an intelligent control method and system for a carbon fiber product production line. Background Technology
[0002] In carbon fiber frame production, allowing a single workstation to independently complete the entire frame layup process can easily lead to serious production problems. Because the layup process is time-consuming, a single workstation can easily become a bottleneck in the production line, causing subsequent workstations to wait and hindering overall efficiency. To avoid this, in actual production, the manufacturing process of carbon fiber products is usually divided into multiple steps, such as layup, curing, and processing. To improve efficiency and reduce waiting time, factories generally use a "time balancing" method to divide workstations, aiming to make the working time of each workstation as similar as possible.
[0003] However, while this method of division solves the time matching problem, it overlooks another crucial aspect: how to smoothly connect the various processes. For example, if the products completed in the previous process cannot be transferred to the next station in a timely manner, or if a product defect occurs in a certain process, causing the entire production line to stop or experience a temporary delay, it will trigger a chain reaction of waiting in subsequent processes, ultimately leading to a backlog of semi-finished products and affecting overall production efficiency. Summary of the Invention
[0004] This invention provides an intelligent control method and system for carbon fiber product production lines to solve existing problems.
[0005] The intelligent control method for a carbon fiber product production line of the present invention adopts the following technical solution: One embodiment of the present invention provides an intelligent control method for a carbon fiber product production line, the method comprising the following steps: The complete production process of carbon fiber products is broken down into process units; Obtain historical defective product data, which includes the defect contribution of each historical defective product in each process unit and the corresponding product defect severity for each historical defective product. Based on historical defective product data, determine the process impact of each process unit; Based on the degree of influence of each process, important processes are identified from the process units, and then the important processes are ranked according to the degree of influence of each important process to obtain the process ranking. Obtain the process combinability of the target process and its adjacent processes in the process ranking; The target process with a process combinability greater than a preset process combinability threshold is combined with adjacent processes to obtain a process group; In each process group, the defects in the products produced by the target process are compensated and adjusted in adjacent processes to obtain an optimized carbon fiber product production line.
[0006] Optionally, obtain historical defective product data, specifically including: Based on historical experience, the defect contribution of each historically defective product in each process unit is obtained, wherein the sum of the defect contribution of each historically defective product in each process unit is 1. Acquire historical defect images of products with historical defects, use a defect recognition model to identify the historical defect images, and obtain the recognition results; The ratio of the number of pixels in the a-th identification result to the number of pixels in the corresponding historical defect image is determined as the product defect degree of the a-th historical defect product. Obtain the product defect level corresponding to each historical defective product.
[0007] Optionally, based on historical defective product data, the process impact of each process unit is determined, specifically including: The product of the defect contribution of the i-th process unit in the j-th historical defective product data and the product defect degree of the i-th process unit in the j-th historical defective product data is determined as the initial process influence degree of the i-th process unit in the j-th historical defective product data. The average value of the initial process influence of the i-th process unit in each historical defective product data is determined as the process influence of the i-th process unit. Obtain the process impact degree of each process unit.
[0008] Optionally, based on the degree of influence of the process, important processes are identified from the process unit, specifically including: Calculate the mean process influence of each process unit; Process units whose process influence is greater than the average process influence are identified as important processes.
[0009] Optionally, the process combinability of the target process and its adjacent processes in the process ranking is obtained, specifically including: The target process is determined from the process ranking, and the next process unit in the process unit is determined as the adjacent process; In the historical defective product data, the data of the target process and adjacent processes that meet the preset defect contribution conditions are identified as the target dataset of the target process. Obtain historical data curves for normal products in each process unit, and determine reference curves for adjacent processes from the historical data curves; Obtain the target data curves concentrated in adjacent processes, and calculate the curve deviation of each data point in the target dataset in adjacent processes; Based on the curve deviation of each data point in the target dataset in adjacent processes, the data in the target dataset are sorted to obtain a sorted sequence; Based on the preset curve deviation threshold, low-deviation data and high-deviation data are determined from the sorted sequence; Based on low-deviation and high-deviation data, calculate the process combinability of the target process and adjacent processes; Each important process in the process ranking is identified as the target process, and the process combinability of each target process and its adjacent processes is obtained.
[0010] Optionally, reference curves for adjacent processes can be determined from historical data curves, specifically including: Obtain the c-th curve of the b-th historical normal product in the adjacent process; Calculate the cosine similarity between the c-th curve of the b-th historical normal product in the adjacent process and other curves, where the other curves are the d-th curves of the other historical normal products except the b-th historical normal product in the adjacent process, and the d-th curve and the c-th curve are of the same type. The cosine similarity of the c-th curve of the b-th historical normal product in the adjacent process with other curves is summed to obtain the representative feature value of the c-th curve of the b-th historical normal product in the adjacent process. Obtain the representative characteristic value of each curve in the adjacent process for the b-th historical normal product and calculate the average value to obtain the representative characteristic value of the curve in the adjacent process for the b-th historical normal product. Obtain the curve representative feature value of each historical normal product in adjacent processes; The curve representing the historical normal product with the largest characteristic value in the adjacent process is determined as the reference curve for the adjacent process.
[0011] Optionally, the curve deviation of each data point in the target dataset in adjacent processes is calculated separately, specifically including: Calculate the cosine similarity between the e-th data in the target dataset and the f-th curve in the adjacent process and the c-th curve in the reference curve to obtain the initial cosine similarity, where the f-th curve and the c-th curve are of the same type; Obtain the initial cosine similarity of the e-th data in the target dataset for each curve in adjacent processes and calculate the mean value to obtain the target cosine similarity of the e-th data in the target dataset in adjacent processes. Based on the target cosine similarity of the e-th data in the target dataset to the adjacent processes, calculate the curve deviation of the e-th data in the target dataset to the adjacent processes; Obtain the curve deviation of each data point in the target dataset in adjacent processes.
[0012] Optionally, based on low-deviation and high-deviation data, the process combinability of the target process and adjacent processes is calculated, specifically including: The curve deviations of the low deviation data are sorted to obtain the first deviation sequence. Each low deviation data in the first deviation sequence is replaced with the product defect degree corresponding to the low deviation data to obtain the first defect degree sequence. Calculate the cosine similarity between the first deviation sequence and the first defect degree sequence to obtain the first cosine similarity. The curve deviations of the high deviation data are sorted to obtain the second deviation sequence. Each high deviation data in the second deviation sequence is replaced with the product defect degree corresponding to the high deviation data to obtain the second defect degree sequence. Calculate the cosine similarity between the second deviation sequence and the second defect degree sequence to obtain the second cosine similarity; The process combinability of the target process and its adjacent processes is calculated based on the first cosine similarity and the second cosine similarity.
[0013] Optionally, the process combinability of the target process and adjacent processes is calculated based on the first cosine similarity and the second cosine similarity, specifically including: The difference between the second cosine similarity and the first cosine similarity is defined as the cosine similarity difference. The first cosine similarity is negative and used as the exponent of the natural constant to obtain the similarity coefficient; The product of the cosine similarity difference and the similarity coefficient is used to determine the process combinability between the target process and adjacent processes.
[0014] This invention proposes an intelligent control system for a carbon fiber product production line, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the intelligent control method for a carbon fiber product production line as described above.
[0015] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, by analyzing historical data, the degree of influence of different processes on the quality of carbon fiber products is obtained. Combined with the merging of adjacent processes, unit processes that can be merged are identified. Merging these unit processes into process groups, and compensating for product defects within these process groups, improves process stability. While considering process timing, this minimizes the occurrence of defective products, achieving a dual improvement in quality and efficiency. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 A flowchart illustrating an intelligent control method for a carbon fiber product production line according to an embodiment of the present invention; Figure 2 This is a structural diagram of an intelligent control system for a carbon fiber product production line provided in one embodiment of the present invention. Detailed Implementation
[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent control method for a carbon fiber product production line proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0020] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent control method for a carbon fiber product production line provided by the present invention.
[0021] This invention provides an intelligent control method and system for a carbon fiber product production line. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of an intelligent control method for a carbon fiber product production line according to an embodiment of the present invention. The method includes the following steps: S101. The complete production process of carbon fiber products is broken down into process units.
[0022] For example, in different processes of a carbon fiber product production line, corresponding data can be collected by sensors, such as: Laying process: Pressure sensors are used to monitor the pressure data of the laying rollers to ensure that air bubbles are eliminated and the fibers are not damaged; Curing process: Thermocouples are used to monitor real-time temperature data inside the mold and product to ensure precise control of the curing cycle; Prepreg storage: Use temperature and humidity sensors to monitor the environmental data of the cold storage to prevent the prepreg from curing prematurely or failing due to moisture absorption.
[0023] The specific production lines for different products may have different processes. We need to combine the specific processes to obtain the relevant data for each process.
[0024] Furthermore, each process in the carbon fiber product production line can be broken down into its smallest unit, known as a process unit. For example, the layup process can be further divided into: a layup unit specifically responsible for the main parts of the front triangle (frame body), a layup unit specifically responsible for the rear and connecting parts, and a layup unit specifically responsible for complex connecting parts such as the rear fork and bottom bracket. These decomposed processes are recorded as process units.
[0025] S102. Obtain historical defective product data, which includes the defect contribution of each historical defective product in each process unit and the product defect degree corresponding to each historical defective product.
[0026] In this embodiment, obtaining historical defective product data specifically includes: Based on historical experience, the defect contribution of each historically defective product in each process unit is obtained, wherein the sum of the defect contribution of each historically defective product in each process unit is 1. Acquire historical defect images of products with historical defects, use a defect recognition model to identify the historical defect images, and obtain the recognition results; The ratio of the number of pixels in the a-th identification result to the number of pixels in the corresponding historical defect image is determined as the product defect degree of the a-th historical defect product. Obtain the product defect level corresponding to each historical defective product.
[0027] For example, defective carbon fiber products are recorded as historical defective products, and the production data corresponding to these historical defective products is recorded as historical defective product data. Furthermore, the defect contribution of each historical defective product in each process unit can be obtained as follows: The contribution of each smallest process unit to the defect is determined through expert scoring or calculation using historical defective product data. Since many defects originate within the smallest process unit, this unit is used as the smallest unit of analysis to obtain the defect contribution of each unit. Furthermore, the sum of the defect contributions of all process units in each historical defective product data point is 1. This allows for the generation of a defect contribution sequence for each historical defective product data point, where the first element represents the defect contribution of the first smallest process unit, the second element represents the defect contribution of the second smallest process unit, and so on.
[0028] To determine the degree of product defect, image data of each historically defective product can be acquired using a camera, i.e., historical defect images. A trained defect recognition model is then used to identify these historical defect images, yielding the recognition result. The ratio of the number of defective pixels in the recognition result to the total number of pixels in the entire historical defect image is used to determine the degree of product defect for each historically defective product.
[0029] For example, the obtained historical defective product data can be: {[Historical defective product data 1: defect contribution 11, defect contribution 12, ..., defect contribution 1n, product defect degree 1], [Historical defective product data 2: defect contribution 21, defect contribution 22, ..., defect contribution 2n, product defect degree 2], [Historical defective product data 3: defect contribution 31, defect contribution 32, ..., defect contribution 3n, product defect degree 3], ..., [Historical defective product data m: defect contribution m1, defect contribution m2, ..., defect contribution mn, product defect degree m]}.
[0030] S103. Based on historical defective product data, determine the process impact degree of each process unit.
[0031] In this embodiment, the process impact of each process unit is determined based on historical defective product data, specifically including: The product of the defect contribution of the i-th process unit in the j-th historical defective product data and the product defect degree of the i-th process unit in the j-th historical defective product data is determined as the initial process influence degree of the i-th process unit in the j-th historical defective product data. The average value of the initial process influence of the i-th process unit in each historical defective product data is determined as the process influence of the i-th process unit. Obtain the process impact degree of each process unit.
[0032] For example, for each process unit, the degree of defect of each product in each historical defective product data can be used as the weight of the product contribution. The process influence of each process unit can be obtained by weighting and averaging all the product defect degrees in each historical defective product data.
[0033] S104. Based on the process influence, identify the important processes from the process units, and rank the important processes according to the process influence of each important process to obtain the process ranking.
[0034] In this embodiment, important processes are identified from the process units based on their impact, specifically including: Calculate the mean process influence of each process unit; Process units whose process influence is greater than the average process influence are identified as important processes.
[0035] For example, the average process influence of all process units is calculated, and process units with a process influence greater than the average process influence are identified as important processes. That is, these important processes are likely to have a significant impact on the quality of carbon fiber products during the production process.
[0036] Optionally, important processes can be sorted according to their process influence, either in ascending or descending order, without any specific restrictions.
[0037] S105. Obtain the process combinability of the target process and its adjacent processes in the process ranking.
[0038] In this embodiment, obtaining the process combinability of the target process and its adjacent processes in the process ranking specifically includes: The target process is determined from the process ranking, and the next process unit in the process unit is determined as the adjacent process; In the historical defective product data, the data of the target process and adjacent processes that meet the preset defect contribution conditions are identified as the target dataset of the target process. Obtain historical data curves for normal products in each process unit, and determine reference curves for adjacent processes from the historical data curves; Obtain the target data curves concentrated in adjacent processes, and calculate the curve deviation of each data point in the target dataset in adjacent processes; Based on the curve deviation of each data point in the target dataset in adjacent processes, the data in the target dataset are sorted to obtain a sorted sequence; Based on the preset curve deviation threshold, low-deviation data and high-deviation data are determined from the sorted sequence; Based on low-deviation and high-deviation data, calculate the process combinability of the target process and adjacent processes; Each important process in the process ranking is identified as the target process, and the process combinability of each target process and its adjacent processes is obtained.
[0039] Determine reference curves for adjacent processes from historical data curves, specifically including: Obtain the c-th curve of the b-th historical normal product in the adjacent process; Calculate the cosine similarity between the c-th curve of the b-th historical normal product in the adjacent process and other curves, where the other curves are the d-th curves of the other historical normal products except the b-th historical normal product in the adjacent process, and the d-th curve and the c-th curve are of the same type. The cosine similarity of the c-th curve of the b-th historical normal product in the adjacent process with other curves is summed to obtain the representative feature value of the c-th curve of the b-th historical normal product in the adjacent process. Obtain the representative characteristic value of each curve in the adjacent process for the b-th historical normal product and calculate the average value to obtain the representative characteristic value of the curve in the adjacent process for the b-th historical normal product. Obtain the curve representative feature value of each historical normal product in adjacent processes; The curve representing the historical normal product with the largest characteristic value in the adjacent process is determined as the reference curve for the adjacent process.
[0040] Calculate the curve deviation of each data point in the target dataset in adjacent processes, specifically including: Calculate the cosine similarity between the e-th data in the target dataset and the f-th curve in the adjacent process and the c-th curve in the reference curve to obtain the initial cosine similarity, where the f-th curve and the c-th curve are of the same type; Obtain the initial cosine similarity of the e-th data in the target dataset for each curve in adjacent processes and calculate the mean value to obtain the target cosine similarity of the e-th data in the target dataset in adjacent processes. Based on the target cosine similarity of the e-th data in the target dataset to the adjacent processes, calculate the curve deviation of the e-th data in the target dataset to the adjacent processes; Obtain the curve deviation of each data point in the target dataset in adjacent processes.
[0041] Based on low-deviation and high-deviation data, the process combinability of the target process and adjacent processes is calculated, specifically including: The curve deviations of the low deviation data are sorted to obtain the first deviation sequence. Each low deviation data in the first deviation sequence is replaced with the product defect degree corresponding to the low deviation data to obtain the first defect degree sequence. Calculate the cosine similarity between the first deviation sequence and the first defect degree sequence to obtain the first cosine similarity. The curve deviations of the high deviation data are sorted to obtain the second deviation sequence. Each high deviation data in the second deviation sequence is replaced with the product defect degree corresponding to the high deviation data to obtain the second defect degree sequence. Calculate the cosine similarity between the second deviation sequence and the second defect degree sequence to obtain the second cosine similarity; The process combinability of the target process and its adjacent processes is calculated based on the first cosine similarity and the second cosine similarity.
[0042] Based on the first cosine similarity and the second cosine similarity, the process combinability of the target process and adjacent processes is calculated, specifically including: The difference between the second cosine similarity and the first cosine similarity is defined as the cosine similarity difference. The first cosine similarity is negative and used as the exponent of the natural constant to obtain the similarity coefficient; The product of the cosine similarity difference and the similarity coefficient is used to determine the process combinability between the target process and adjacent processes.
[0043] For example, determining the target process from the process ranking can be done by identifying the most important process with the greatest impact as the target process, and then identifying the next process after the target process as the adjacent process. For instance, if the most important process with the greatest impact is the 5th unit process in the unit process, then the 6th unit process is the adjacent process.
[0044] Optionally, the preset defect contribution condition can be a threshold set based on historical experience, or a threshold obtained based on expert scoring. In a preferred embodiment, the preset defect contribution condition can be that the defect contribution of the target process is greater than 0.7, and the defect contribution of adjacent processes is less than 0.05. Data that meets the preset defect contribution condition is determined as the target dataset for the target process.
[0045] Simultaneously, data on defect-free carbon fiber products from historical production processes are acquired and recorded as historical normal product data. Historical data curves for each process unit using these historical normal products are obtained, and data curves for adjacent processes are selected from these historical data curves. The data curves for adjacent processes can contain at least one data curve, such as temperature, pressure, and humidity curves, etc., with different curve types.
[0046] Taking the temperature curves of adjacent processes in a historical data curve as an example, calculate the sum of the cosine similarities between this temperature curve and other temperature curves. Similarly, calculate the sum of the cosine similarities between the pressure curve and the humidity curve. Averaging the sums of the cosine similarities of the temperature curve, pressure curve, and humidity curve can be considered as the cosine similarity between this historical data curve and each of the other historical data curves. In other words, this represents the curve characteristic value of the historical normal product in adjacent processes corresponding to this historical data curve. Obtain the curve characteristic value of each historical normal product in adjacent processes, and determine the curve of the historical normal product in adjacent processes with the largest curve characteristic value as the reference curve for adjacent processes.
[0047] In the preferred embodiment described above, the selection criterion for the target dataset is that the defect contribution of adjacent processes is less than 0.05, meaning that the defects are mainly caused by the target process rather than adjacent processes. If significant deviations are found in the process parameters (such as temperature, pressure, and other curve data) of adjacent processes in the defect data, it may indicate that the adjustment feedback system or workers have taken compensatory measures (such as adjusting parameters to correct deviations) for the semi-finished product defects of the target process, resulting in data deviations from the normal range. In this case, the remedial capability can be quantitatively assessed by statistically analyzing the correlation between the degree of data deviation of adjacent processes and the final defect rate: if the deviation is large but the defect rate is low, the remediation is effective; if the deviation and the defect rate are not significantly correlated, the remediation may be ineffective or there may be over-intervention. Based on this probability assessment result, it can be further determined whether to merge the two processes into one process unit.
[0048] Therefore, we can obtain the initial cosine similarity between each curve in adjacent processes and the reference curves for each data point in the target dataset, thus obtaining the initial cosine similarity. Then, we calculate the average of the initial cosine similarities of all curves in adjacent processes for each data point in the target dataset to obtain the target cosine similarity for each data point in the target dataset across adjacent processes.
[0049] Based on the target cosine similarity of the e-th data in the target dataset to the adjacent processes, the curve deviation of the e-th data in the target dataset to the adjacent processes can be calculated as follows: (1-target cosine similarity) is taken as the curve deviation of the e-th data in the adjacent processes.
[0050] The target dataset is sorted based on the curve deviation to obtain a sorted sequence. The sorting method can be ascending or descending; no specific restriction is imposed here. The preset curve deviation threshold can be obtained by segmenting the sorted sequence using the Otsu thresholding method, or by setting it based on practical experience.
[0051] Data that is less than the preset curve deviation threshold can be recorded as low deviation data, and data that is greater than or equal to the preset curve deviation threshold can be recorded as high deviation data.
[0052] If in low-deviation data, the variation in deviation and defect severity is small or almost nonexistent; but in high-deviation data, the larger the deviation, the smaller the defect severity tends to be, it indicates that adjacent processes can remedy the target process.
[0053] Based on the first cosine similarity and the second cosine similarity, the process combinability of the target process and adjacent processes is calculated. The calculation formula can be:
[0054] in, Indicates process combinatoriality, Indicates the second cosine similarity. Indicates the first cosine similarity. Represents the natural constant.
[0055] This indicates the degree to which adjacent processes remedy the target process. The larger the value, the more likely the next process can remedy the process corresponding to the target process, and thus the more likely the process corresponding to the next process can be combined with the process corresponding to the target process into a single process unit, that is, combining these two processes into one process.
[0056] Using the same method, each important process in the process ranking can be identified as a target process, and the process combinability of each target process and its adjacent processes can be obtained.
[0057] S106. Combine the target process with a process combinability greater than the preset process combinability threshold with adjacent processes to obtain a process group.
[0058] Optionally, the value of the preset process combinability threshold can be set according to the actual situation or historical experience. There is no specific numerical limit here. In a preferred embodiment, it can be 0.7.
[0059] S107. In the adjacent processes of each process group, the product defects produced by the target process are compensated and adjusted to obtain the optimized carbon fiber product production line.
[0060] For example, the two processes in each process group are denoted as the first process and the second process. After the first process, an inspection device is installed, such as to obtain a product image captured by a camera, and the degree of defect of the product at this time is obtained through a defect detection model.
[0061] The system then automatically determines the following based on the measurement results: Pass: Continue with standard procedures; Minor defects: Send adjustment instructions to the equipment in the second process step and activate the compensation plan; Major defects: Record the data and isolate or transport the product to the semi-finished product area, without further processing, and restart the process flow directly from the beginning of the production line. Upon receiving the instructions, the second process step automatically adjusts the process parameters based on the original parameters. For example, in carbon fiber production, if air bubbles are detected in the layup, the system automatically adjusts the curing process: reducing the heating rate and extending the holding time. The final product achieves the required porosity, saving a product that would otherwise be scrapped.
[0062] In summary, in this embodiment of the invention, by analyzing historical data, the degree of influence of different processes on the quality of carbon fiber products is obtained. Combined with the merging of adjacent processes, unit processes that can be merged are identified. Merging these unit processes into process groups, and compensating for product defects within these process groups, can improve process stability. While considering process time connections, it minimizes the occurrence of defective products, achieving a dual improvement in quality and efficiency.
[0063] This invention also proposes an intelligent control system for a carbon fiber product production line; please refer to [link / reference]. Figure 2 The diagram shows a structural diagram of an intelligent control system for a carbon fiber product production line according to an embodiment of the present invention. The system includes: a process splitting module 101, a data processing module 102, and an optimization compensation module 103.
[0064] The process splitting module 101 is used to break down the complete production process of carbon fiber products into process units; The data processing module 102 is used to acquire historical defective product data, which includes the defect contribution of each historical defective product in each process unit and the corresponding product defect degree of each historical defective product; based on the historical defective product data, the process influence of each process unit is determined; based on the process influence, important processes are identified from the process units, and the important processes are ranked according to the process influence of each important process to obtain the process ranking; the process combinability of the target process and its adjacent processes in the process ranking is obtained; the target process and its adjacent processes with a process combinability greater than a preset process combinability threshold are combined to obtain a process group; The optimization compensation module 103 is used to compensate and adjust the product defects produced by the target process in adjacent processes of each process group, so as to obtain an optimized carbon fiber product production line.
[0065] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer equipment can be divided into different functional modules to complete all or part of the functions described above. In addition, the intelligent control system for a carbon fiber product production line and the intelligent control method for a carbon fiber product production line provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.
[0066] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0067] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for intelligent control of a carbon fiber product production line, characterized in that, include: The complete production process of carbon fiber products is broken down into process units; Obtain historical defective product data, which includes the defect contribution of each historical defective product in each process unit and the corresponding product defect severity for each historical defective product. Based on historical defective product data, determine the process impact of each process unit; Based on the degree of influence of each process, important processes are identified from the process units, and then the important processes are ranked according to the degree of influence of each important process to obtain the process ranking. Obtain the process combinability of the target process and its adjacent processes in the process ranking; The target process with a process combinability greater than a preset process combinability threshold is combined with adjacent processes to obtain a process group; In each process group, the defects in the products produced by the target process are compensated and adjusted in adjacent processes to obtain an optimized carbon fiber product production line.
2. The intelligent control method for a carbon fiber product production line according to claim 1, characterized in that, The acquisition of historical defective product data specifically includes: Based on historical experience, the defect contribution of each historically defective product in each process unit is obtained, wherein the sum of the defect contribution of each historically defective product in each process unit is 1. Acquire historical defect images of products with historical defects, use a defect recognition model to identify the historical defect images, and obtain the recognition results; The ratio of the number of pixels in the a-th identification result to the number of pixels in the corresponding historical defect image is determined as the product defect degree of the a-th historical defect product. Obtain the product defect level corresponding to each historical defective product.
3. The intelligent control method for a carbon fiber product production line according to claim 1, characterized in that, The determination of the process impact degree of each process unit based on historical defective product data specifically includes: The product of the defect contribution of the i-th process unit in the j-th historical defective product data and the product defect degree of the i-th process unit in the j-th historical defective product data is determined as the initial process influence degree of the i-th process unit in the j-th historical defective product data. The average value of the initial process influence of the i-th process unit in each historical defective product data is determined as the process influence of the i-th process unit. Obtain the process impact degree of each process unit.
4. The intelligent control method for a carbon fiber product production line according to claim 1, characterized in that, The process of identifying important processes from process units based on their impact includes: Calculate the mean process influence of each process unit; Process units whose process influence is greater than the average process influence are identified as important processes.
5. The intelligent control method for a carbon fiber product production line according to claim 1, characterized in that, The process combinability of the target process and its adjacent processes in the process ranking specifically includes: The target process is determined from the process ranking, and the next process unit in the process unit is determined as the adjacent process; In the historical defective product data, the data of the target process and adjacent processes that meet the preset defect contribution conditions are identified as the target dataset of the target process. Obtain historical data curves for normal products in each process unit, and determine reference curves for adjacent processes from the historical data curves; Obtain the target data curves concentrated in adjacent processes, and calculate the curve deviation of each data point in the target dataset in adjacent processes; Based on the curve deviation of each data point in the target dataset in adjacent processes, the data in the target dataset are sorted to obtain a sorted sequence; Based on the preset curve deviation threshold, low-deviation data and high-deviation data are determined from the sorted sequence; Based on low-deviation and high-deviation data, calculate the process combinability of the target process and adjacent processes; Each important process in the process ranking is identified as the target process, and the process combinability of each target process and its adjacent processes is obtained.
6. The intelligent control method for a carbon fiber product production line according to claim 5, characterized in that, The step of determining the reference curve for adjacent processes from historical data curves specifically includes: Obtain the c-th curve of the b-th historical normal product in the adjacent process; Calculate the cosine similarity between the c-th curve of the b-th historical normal product in the adjacent process and other curves, where the other curves are the d-th curves of the other historical normal products except the b-th historical normal product in the adjacent process, and the d-th curve and the c-th curve are of the same type. The cosine similarity of the c-th curve of the b-th historical normal product in the adjacent process with other curves is summed to obtain the representative feature value of the c-th curve of the b-th historical normal product in the adjacent process. Obtain the representative characteristic value of each curve in the adjacent process for the b-th historical normal product and calculate the average value to obtain the representative characteristic value of the curve in the adjacent process for the b-th historical normal product. Obtain the curve representative feature value of each historical normal product in adjacent processes; The curve representing the historical normal product with the largest characteristic value in the adjacent process is determined as the reference curve for the adjacent process.
7. The intelligent control method for a carbon fiber product production line according to claim 5, characterized in that, The step of calculating the curve deviation of each data point in the target dataset in adjacent processes specifically includes: Calculate the cosine similarity between the e-th data in the target dataset and the f-th curve in the adjacent process and the c-th curve in the reference curve to obtain the initial cosine similarity, where the f-th curve and the c-th curve are of the same type; Obtain the initial cosine similarity of the e-th data in the target dataset for each curve in adjacent processes and calculate the mean value to obtain the target cosine similarity of the e-th data in the target dataset in adjacent processes. Based on the target cosine similarity of the e-th data in the target dataset to the adjacent processes, calculate the curve deviation of the e-th data in the target dataset to the adjacent processes; Obtain the curve deviation of each data point in the target dataset in adjacent processes.
8. The intelligent control method for a carbon fiber product production line according to claim 5, characterized in that, The calculation of the process combinability of the target process and adjacent processes based on low-deviation data and high-deviation data specifically includes: The curve deviations of the low deviation data are sorted to obtain the first deviation sequence. Each low deviation data in the first deviation sequence is replaced with the product defect degree corresponding to the low deviation data to obtain the first defect degree sequence. Calculate the cosine similarity between the first deviation sequence and the first defect degree sequence to obtain the first cosine similarity. The curve deviations of the high deviation data are sorted to obtain the second deviation sequence. Each high deviation data in the second deviation sequence is replaced with the product defect degree corresponding to the high deviation data to obtain the second defect degree sequence. Calculate the cosine similarity between the second deviation sequence and the second defect degree sequence to obtain the second cosine similarity; The process combinability of the target process and its adjacent processes is calculated based on the first cosine similarity and the second cosine similarity.
9. The intelligent control method for a carbon fiber product production line according to claim 8, characterized in that, The step of calculating the process combinability of the target process and adjacent processes based on the first cosine similarity and the second cosine similarity specifically includes: The difference between the second cosine similarity and the first cosine similarity is defined as the cosine similarity difference. The first cosine similarity is negative and used as the exponent of the natural constant to obtain the similarity coefficient; The product of the cosine similarity difference and the similarity coefficient is used to determine the process combinability between the target process and adjacent processes.
10. An intelligent control system for a carbon fiber product production line, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent control method for a carbon fiber product production line as described in any one of claims 1-9.