Traceability information management method and system based on cloud cut online data

Abstract

The application provides a cloud-based online data traceability information management method and system, relating to the technical field of information management, which comprises the following steps: obtaining the unique traceability identification of a steel plate cutting finished product, querying the cloud cutting online platform for orders, detecting the quality anomaly of the steel plate cutting finished product according to the matched cloud cutting order, tracing the cutting processing data of the cloud cutting online platform according to the unique traceability identification, analyzing the multi-parameter causal traceability of the multi-dimensional cutting quality anomaly according to the cutting process block, and establishing a cutting quality anomaly traceability graph; and dynamically correlating and compensating the cutting quality anomaly traceability graph according to the layout process block and the blanking process block to obtain a cloud cutting anomaly traceability graph. The technical problems of the prior art that the reasons for the steel plate cutting anomaly are difficult to trace and the quality control means are not accurate are solved. The technical effects of accurately locating the root cause of the steel plate cutting quality anomaly and effectively improving the cutting quality control level are achieved.

Description

Technical Field

[0001] This invention relates to the field of information management technology, specifically to a method and system for managing traceability information based on cloud-based online data. Background Technology

[0002] In modern steel plate processing, cutting quality directly affects product performance and the reliability of downstream manufacturing processes. With the development of industrial automation and intelligent manufacturing, steel plate cutting equipment and processes are becoming increasingly complex, involving multi-dimensional factors such as layout, cutting parameters, equipment status, and environmental conditions. These factors interact during processing, causing quality anomalies such as dimensional deviations, cut defects, and finished product deformation to exhibit diverse and dynamic characteristics. Current technologies rely primarily on manual inspection or single-process monitoring, making it difficult to detect anomalies in a timely manner and lacking effective means for full-process tracking and causal analysis. Once an anomaly occurs, only surface corrections can be made, making it difficult to accurately pinpoint the root cause, leading to high anomaly recurrence rates, low production efficiency, and increased scrap rates. Furthermore, current technologies lack the ability to integrate and dynamically correlate multi-stage and multi-factor data, leaving quality control and process optimization without a reliable basis.

[0003] Existing technologies have technical problems such as difficulty in tracing the cause of abnormal steel plate cutting and imprecise quality control methods. Summary of the Invention

[0004] The purpose of this application is to provide a traceability information management method and system based on cloud-based online cutting data, in order to solve the technical problems of difficulty in tracing the cause of abnormal steel plate cutting and inaccurate quality control methods in existing technologies.

[0005] In view of the above problems, this application provides a method and system for managing traceability information based on cloud-based online data.

[0006] The first aspect of this application provides a method for managing traceability information based on cloud-based online cutting data. This method includes: obtaining a unique traceability identifier for the finished steel plate cutting product; querying orders on the cloud-based online cutting platform based on the unique traceability identifier to obtain matching cloud-based cutting orders; performing quality anomaly detection on the finished steel plate cutting product based on the matching cloud-based cutting orders to obtain multi-dimensional cutting quality anomaly characteristics; performing cutting processing data backtracking on the cloud-based online cutting platform based on the unique traceability identifier to obtain the layout process block, cutting process block, and unloading process block corresponding to the finished steel plate cutting product; performing multi-parameter causal traceability analysis on the multi-dimensional cutting quality anomaly characteristics based on the cutting process block to establish a cutting quality anomaly traceability map; and performing upstream and downstream dynamic correlation compensation on the cutting quality anomaly traceability map based on the layout process block and the unloading process block to obtain a cloud-based cutting anomaly traceability map.

[0007] Optionally, the unique traceability identifier is input into the cloud-based online cutting platform to obtain the finished product size data, cut image, and finished product shape image corresponding to the steel plate cut product; the finished product size data is subjected to size consistency detection based on the matched cloud-based cutting order to obtain size deviation anomaly characteristics; the cut quality of the steel plate cut product is subjected to cut quality detection based on the cut image to obtain cut quality anomaly characteristics; the deformation of the steel plate cut product is subjected to deformation detection based on the finished product shape image to obtain cutting deformation anomaly characteristics, and the multi-dimensional cutting quality anomaly characteristics are generated by combining the size deviation anomaly characteristics and the cut quality anomaly characteristics.

[0008] Optionally, the following steps are taken: extracting features of the cut edge region, cut cross-sectional morphology, and cut attachments from the cut image; supervising the training of a deep convolutional neural network based on a cut edge anomaly detection record set to obtain a cut edge anomaly detection network; inputting the cut edge region features into the cut edge anomaly detection network to obtain cut edge anomaly characteristics; identifying cross-sectional morphology anomalies based on the cut cross-sectional morphology features to obtain cut cross-sectional morphology anomaly characteristics; identifying cut attachment anomalies based on the cut attachment features to obtain cut attachment anomaly characteristics; and combining the cut edge anomaly characteristics and the cut cross-sectional morphology anomaly characteristics to generate the cut quality anomaly characteristics.

[0009] Optionally, multi-parameter causal tracing is performed on the dimensional deviation anomaly characteristics based on the cutting process blocks to obtain the dimensional anomaly tracing path; multi-parameter causal tracing is performed on the cut quality anomaly characteristics based on the cutting process blocks to obtain the cut anomaly tracing path; multi-parameter causal tracing is performed on the cutting deformation anomaly characteristics based on the cutting process blocks to obtain the deformation anomaly tracing path; the dimensional anomaly tracing path, the cut anomaly tracing path, and the deformation anomaly tracing path are graphically analyzed to generate the cutting quality anomaly tracing map.

[0010] Optionally, the cutting process blocks are parsed to obtain cutting sequence information and multi-dimensional cutting feature sequences; the size deviation anomaly characteristics are associated and identified according to the multi-dimensional cutting feature sequences to obtain a size anomaly cutting relationship network; based on the size anomaly cutting relationship network, the importance of the association relationships of different parameters is identified to determine the distribution of key factors of size anomalies; the distribution of key factors of size anomalies is reconstructed by order constraints according to the cutting sequence information to obtain the size anomaly order factor distribution; the formation process of the size deviation anomaly characteristics is restored according to the size anomaly order factor distribution to generate a size anomaly tracing path.

[0011] Optionally, dynamic upstream and downstream correlation compensation is performed on the dimensional anomaly tracing path based on the nesting process block and the material unloading process block to obtain a dimensional anomaly correction path; dynamic upstream and downstream correlation compensation is performed on the cut anomaly tracing path based on the nesting process block and the material unloading process block to obtain a cut anomaly correction path; dynamic upstream and downstream correlation compensation is performed on the deformation anomaly tracing path based on the nesting process block and the material unloading process block to obtain a deformation anomaly correction path; the cutting quality anomaly tracing map is optimized based on the dimensional anomaly correction path, the cut anomaly correction path, and the deformation anomaly correction path to generate the cloud cutting anomaly tracing map.

[0012] Optionally, the nesting process block is parsed to obtain nesting layout information and local density features; interference analysis is performed on the size deviation anomaly characteristics based on the nesting layout information and local density features to obtain the size deviation nesting interference path; the material feeding process block is parsed to obtain material feeding sequence information and material feeding operation feature sequence; interference analysis is performed on the size deviation anomaly characteristics based on the material feeding sequence information and material feeding operation feature sequence to obtain the size deviation material feeding interference path; the upstream and downstream correlation reconstruction of the size anomaly tracing path is performed based on the size deviation nesting interference path and the size deviation material feeding interference path to generate the size anomaly correction path.

[0013] Optionally, the multidimensional cutting feature sequence includes a cutting process feature sequence, a cutting equipment feature sequence, and a cutting environment feature sequence.

[0014] Optionally, a cloud-cutting anomaly early warning instruction can be generated based on the cloud-cutting anomaly tracing map.

[0015] A second aspect of this application provides a traceability information management system based on cloud-cutting online data. This system includes: an order query module for obtaining a unique traceability identifier for the finished steel plate cutting product and querying orders on the cloud-cutting online platform based on the unique traceability identifier to obtain matching cloud-cutting orders; an anomaly detection module for performing quality anomaly detection on the finished steel plate cutting product based on the matching cloud-cutting orders to obtain multi-dimensional cutting quality anomaly characteristics; a data backtracking module for performing cutting processing data backtracking on the cloud-cutting online platform based on the unique traceability identifier to obtain the layout process block, cutting process block, and unloading process block corresponding to the finished steel plate cutting product; a traceability analysis module for performing multi-parameter causal traceability analysis on the multi-dimensional cutting quality anomaly characteristics based on the cutting process block to establish a cutting quality anomaly traceability map; and a correlation compensation module for performing upstream and downstream dynamic correlation compensation on the cutting quality anomaly traceability map based on the layout process block and the unloading process block to obtain a cloud-cutting anomaly traceability map.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] The method provided in this application embodiment obtains a unique traceability identifier for the finished steel plate cutting product, and queries the online cloud cutting platform based on the unique traceability identifier to obtain matching cloud cutting orders; performs quality anomaly detection on the finished steel plate cutting product based on the matching cloud cutting orders to obtain multi-dimensional cutting quality anomaly characteristics; performs cutting processing data backtracking on the online cloud cutting platform based on the unique traceability identifier to obtain the layout process block, cutting process block, and unloading process block corresponding to the finished steel plate cutting product; performs multi-parameter causal tracing analysis on the multi-dimensional cutting quality anomaly characteristics based on the cutting process block to establish a cutting quality anomaly tracing map; and performs upstream and downstream dynamic correlation compensation on the cutting quality anomaly tracing map based on the layout process block and the unloading process block to obtain a cloud cutting anomaly tracing map. Through precise data backtracking on the online cloud cutting platform, combined with multi-dimensional quality anomaly detection and causal tracing analysis, the technical effect of accurately locating the root cause of steel plate cutting quality anomalies and effectively improving the level of cutting quality control is achieved.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application 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 merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the traceability information management method based on cloud-cut online data provided in this application.

[0021] Figure 2 A schematic diagram of the structure of the traceability information management system based on cloud-cut online data provided in this application.

[0022] Figure labeling: Order query module 11, anomaly detection module 12, data backtracking module 13, source tracing and analysis module 14, and correlation compensation module 15. Detailed Implementation

[0023] This application provides a traceability information management method and system based on cloud-based online cutting data, designed to address the technical problems of difficulty in tracing the causes of steel plate cutting anomalies and inaccurate quality control methods in existing technologies. By using a cloud-based online cutting platform for precise data backtracking, combined with multi-dimensional quality anomaly detection and causal source analysis, the method achieves the technical effect of accurately locating the root causes of steel plate cutting quality anomalies and effectively improving the level of cutting quality control.

[0024] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0025] Example 1, as Figure 1 As shown, this application provides a method for managing traceability information based on cloud-based online data, which includes:

[0026] Obtain a unique traceability identifier for the finished steel plate cutting product, and use the unique traceability identifier to query orders on the cloud cutting online platform to obtain matching cloud cutting orders.

[0027] Specifically, the first step is to obtain a unique traceability identifier for each steel plate cut product. This identifier is a unique identifier assigned to each steel plate part during the cutting process. This identifier can be a string of numbers, letters, or symbols, generated by the production management system based on the production batch, product serial number, and cutting number. This unique traceability identifier ensures that each cut product can be uniquely recorded on the cloud-based online cutting platform. For example, encoding methods based on QR codes or radio frequency identification (RFID) tags can be used. Taking QR codes as an example, a specific encoding algorithm encodes relevant information about the cut steel plate, such as production batch, production date, and steel material, into a unique QR code pattern, which is then printed or affixed to the finished product. If RFID tags are used, the relevant information is written into an RFID chip, which is then embedded in a suitable location on the finished product.

[0028] Using the unique traceability identifier as a query condition, the system accesses the database of the cloud-based online cutting platform for order matching. This platform is a cloud-based service platform integrating production data, processing orders, and quality control data, capable of storing and managing information throughout the entire process from order acceptance, scheduling, cutting, processing to final delivery. During the query process, a request is sent to the cloud-based online cutting platform based on the unique traceability identifier to query relevant cloud-based cutting orders and obtain matching orders. These matching orders include detailed information about steel plate cutting, at least including the order number, customer information, steel plate specifications, cutting requirements, and quality standards, enabling targeted quality inspection and traceability analysis.

[0029] By accurately obtaining the unique traceability identifier for each steel plate cut product and querying its order, the entire production process of each product can be precisely traced, ensuring that each cut product has a complete production record available for query, thus ensuring the reliability of quality traceability.

[0030] Based on the matched cloud cutting order, the steel plate cutting finished product is subjected to quality anomaly detection to obtain multi-dimensional cutting quality anomaly characteristics.

[0031] Furthermore, based on the matched cloud cutting order, the steel plate cut finished product is subjected to quality anomaly detection to obtain multi-dimensional cutting quality anomaly characteristics, including: inputting the unique traceability identifier into the cloud cutting online platform to obtain the finished product size data, cut image, and finished product shape image corresponding to the steel plate cut finished product; performing size consistency detection on the finished product size data based on the matched cloud cutting order to obtain size deviation anomaly characteristics; performing cut quality detection on the steel plate cut finished product based on the cut image to obtain cut quality anomaly characteristics; performing deformation detection on the steel plate cut finished product based on the finished product shape image to obtain cutting deformation anomaly characteristics, and combining the size deviation anomaly characteristics and the cut quality anomaly characteristics to generate the multi-dimensional cutting quality anomaly characteristics.

[0032] Specifically, a unique traceability identifier is input into the cloud-based online cutting platform. Database retrieval technology quickly locates and extracts the finished product's dimensions, cut images, and finished product morphology images corresponding to the cut steel plate. For example, the cloud-based online cutting platform's database uses a distributed storage architecture with a data storage capacity of up to PB level, capable of responding to data query requests within 0.5 seconds, ensuring rapid acquisition of the required data. The finished product's dimensions refer to the length, width, and height of the cut steel plate. This data originates from the online platform's product information database and is measured and uploaded in real-time via IoT devices, such as laser measuring instruments. The cut images are obtained through image acquisition technology, such as industrial cameras, capturing high-resolution images of the cut surfaces of the finished steel plate. These images provide information on the quality of the cut surface, edge conditions, and surface integrity. The finished product morphology images include the overall morphology of the cut steel plate, encompassing its surface condition, shape, and appearance features, reflecting deformation issues such as warping and twisting.

[0033] Based on the matching cloud cutting order, the finished product size data is checked for dimensional consistency. The matching cloud cutting order clearly specifies the dimensional requirements of the cut steel plate, such as the specific values ​​of length, width, and thickness, as well as the allowable tolerance range. The obtained finished product size data is compared and analyzed with the standard size data in the matching cloud cutting order to achieve dimensional consistency detection. When the actual finished product size data exceeds the tolerance range, it can be determined that there is a dimensional deviation anomaly, and then the characteristics of the dimensional deviation anomaly are obtained. For example, if the matching cloud cutting order requires the steel plate length to be 1000mm with a tolerance range of ±0.5mm, and the actual measured length is 1001.2mm, then it is determined that the cut steel plate has an abnormal deviation in length. Image processing algorithms are used to process the obtained cut image, and the edge area of ​​the cut is extracted for quality inspection. When there are burrs on the cut edge, the cut surface is not perpendicular, or there are too many attachments, it is determined that the cut quality is abnormal, thus obtaining the characteristics of the cut quality anomaly.

[0034] Simultaneously, deformation detection is performed on the cut steel plate based on the finished product image. Deformation detection includes warping, twisting, and localized deformation. A comprehensive 3D scanning technique is used to scan the finished product, acquiring 3D point cloud data. This 3D point cloud data is collected using an industrial-grade laser scanner or structured light scanner, with each scan point corresponding to a coordinate point on the finished product surface. 3D data processing software, such as PolyWorks or MeshLab, processes the acquired 3D point cloud data and compares it with an ideal finished product shape model. The ideal finished product shape model is based on the design drawings in the matching cloud cutting order or a standard model obtained through precise measurement. The software generates a deformation analysis diagram through comparison. During this process, the 3D data processing software calculates the deviation between each point cloud point and the ideal model, identifying the location and degree of surface deformation. Specifically, when warping is present, the height and range of warping are accurately calculated through comparison; when twisting occurs, the angle of twisting can be accurately measured; and for localized deformation, the location and degree of deformation can be determined. Based on the analysis results, the abnormal characteristics of cutting deformation are obtained. After completing the detection of dimensional deviation, cut quality and deformation, the abnormal characteristics of dimensional deviation, cut quality and cutting deformation are integrated to form multi-dimensional cutting quality abnormal characteristics.

[0035] By detecting quality anomalies in steel plate cutting products from multiple dimensions such as size, cut quality, and deformation, multi-dimensional cutting quality anomaly characteristics can be obtained. This allows for the timely detection of various potential quality problems, avoiding the omission of important quality information due to the limitations of single-dimensional detection. It provides a comprehensive and accurate foundation of quality anomaly information for traceability analysis, thereby enabling effective management and control of steel plate cutting quality and improving product quality and production efficiency.

[0036] Furthermore, the steel plate cutting product is subjected to cut quality inspection based on the cut image to obtain cut quality abnormality characteristics, including: extracting cut edge region features, cut cross-sectional morphology features, and cut attachment features from the cut image; performing supervised training on a deep convolutional neural network based on a cut edge abnormality detection record set to obtain a cut edge abnormality detection network; inputting the cut edge region features into the cut edge abnormality detection network to obtain cut edge abnormality characteristics; identifying cross-sectional morphology abnormalities based on the cut cross-sectional morphology features to obtain cut cross-sectional morphology abnormality characteristics; identifying cut attachment abnormalities based on the cut attachment features to obtain cut attachment abnormality characteristics; and combining the cut edge abnormality characteristics and the cut cross-sectional morphology abnormality characteristics to generate the cut quality abnormality characteristics.

[0037] Specifically, the cut image is converted to grayscale. A Gaussian filter is then used to smooth the grayscale cut image, removing noise. Gaussian filtering replaces the value of each pixel in the grayscale cut image with a weighted average of its neighboring pixel values ​​through convolution. The weights are determined by a Gaussian function; for example, a 5×5 Gaussian kernel with a standard deviation of 1.4 is used. The Sobel operator is then used to calculate the gradient magnitude and direction of each pixel in the grayscale cut image. The Sobel operator contains two 3×3 convolution kernels, used to calculate the horizontal and vertical gradients respectively. These kernels are convolved with the grayscale cut image to obtain the horizontal and vertical gradients, and their magnitude and direction are calculated. Finally, along the gradient direction, the gradient magnitude of each pixel is compared with the gradient magnitudes of its neighboring pixels. If the gradient magnitude of a pixel is not a local maximum, it is suppressed and set to 0. Two thresholds are set: a high threshold and a low threshold. Pixels with gradient magnitudes greater than the high threshold are marked as strong edge pixels, and pixels with gradient magnitudes between the high and low thresholds are marked as weak edge pixels. The weak edge pixels are connected to the strong edge pixels through an edge connection algorithm to obtain the contour information of the cut edge. The curvature change between adjacent pixels on the edge contour is calculated. Areas with large curvature changes indicate uneven edges. The gradient change rate of the edge contour is calculated. Areas with large gradient change rates indicate high edge sharpness, forming the cut edge region feature, including information such as cut edge contour, flatness, and sharpness, which is used to reflect whether the cutting process is smooth and whether there are problems such as excessive ablation, burrs, or gaps.

[0038] Image processing algorithms are used to analyze the cross-sectional morphology of the cut, obtaining its morphological features. Specifically, threshold-based segmentation methods, such as the Otsu algorithm, are employed to automatically determine the optimal threshold for segmenting the cut image, separating the cut area from the background. For example, for a grayscale image of the cut, the Otsu algorithm calculates the inter-class variance to find the threshold that maximizes it, thus achieving automatic segmentation. Morphological processing, such as dilation, erosion, opening, and closing operations, is then applied to the segmented cut image to remove small noise points and fill small holes. For instance, a 3×3 structuring element is used for dilation to enlarge the cut area, and a 3×3 structuring element is used for erosion to shrink it. Opening operations remove small noise points by first eroding and then dilating, while closing operations fill small holes by first dilating and then eroding. By calculating the number and area of ​​pixels in the cut section area, the depth of the cut section is obtained. The variance of gray values ​​of pixels in the cut section area is calculated. Areas with smaller variances indicate that the section is relatively flat, thus obtaining the flatness of the cut section. At the same time, edge detection algorithms, such as Canny edge detection, can be used to detect cracks on the cut section. The depth, flatness, and cracks of the cut are integrated to form the morphological features of the cut section.

[0039] Image recognition technology is used to extract the size, shape, and distribution of attachments at the cut, obtaining their features. Specifically, color-based segmentation methods, such as color thresholding, can be used to separate the attachments from the cut image. For example, for oxide attachments, whose color is usually different from the cut background color, a color threshold can be set to extract the oxide attachment region from the cut image. Morphological processing, such as dilation, erosion, opening, and closing operations, is then performed on the segmented attachment image to remove small noise points and fill small holes. The size and shape of the attachments are described by calculating features such as the number of pixels, area, perimeter, and aspect ratio of the attachment regions. For example, the number of pixels and area of ​​the attachment region reflect its size, while the perimeter and aspect ratio reflect its shape. Furthermore, the distribution of the attachments is analyzed by statistically analyzing their position in the cut image. For example, the centroid coordinates of the attachment regions are calculated to determine their distribution position on the cut, and the distances between attachment regions are calculated to understand their distribution density.

[0040] In the process of detecting anomalies at the cut edges, a large number of image samples containing both normal and abnormal cut edges are first collected to establish a cut edge anomaly detection record set. This record set covers different types and severity of cut edge anomalies, and each sample is labeled to clarify its corresponding normal or abnormal category. Using the cut edge anomaly detection record set as supervised training data, preprocessed and uniformly sized images of cut edge regions are input into a deep convolutional neural network. The deep convolutional neural network consists of an input layer, multiple convolutional layers, pooling layers, and fully connected layers. The input layer receives the cut edge region features obtained from the cut image processing. The convolutional layers are used to automatically extract local features such as texture changes, jagged structures, and ablation marks at the cut edges. Each convolutional layer contains multiple convolutional kernels, such as 3×3 or 5×5, and a feature map is generated through sliding calculation. Shallow convolutions mainly learn low-level features, such as edge intensity and direction changes, while deep convolutions gradually learn high-level semantic features, such as continuous burrs, edge breakage patterns, and other abnormal patterns. A pooling layer, such as max pooling, is placed after the convolutional layer to downsample the feature mapping, reduce the feature dimensionality, decrease computation, and enhance the robustness of the deep neural network to slight shifts in edge positions. The pooling window is typically set to 2×2 with a stride of 2. Fully connected layers are used to comprehensively discriminate high-level features and learn the mapping relationship between different combinations of cut edge features and anomaly categories. The output layer uses the Softmax or Sigmoid activation function to output the classification result of the cut edge. During training, the predicted results of the cut edges are obtained through forward propagation, and the predicted results are compared with the corresponding labels. The error is calculated using the cross-entropy loss function, and then the weights and bias parameters of each convolutional kernel are continuously adjusted through the backpropagation algorithm, so that the deep neural network gradually learns the distinction pattern between normal and abnormal features of the cut edge. When the training process converges, a cut edge anomaly detection network for actual cut quality detection is obtained. The features of the cut edge region are input into a trained cut edge anomaly detection network to detect cut edge anomalies. The cut edge anomaly detection network determines whether there are anomalies at the cut edge based on the input features. Anomalies include at least the following categories: burrs, cracks, excessive ablation, irregular cutting, etc.

[0041] After extracting the cross-sectional morphology features from the cut image, a rule-based recognition method is used to identify cross-sectional morphology anomalies. Based on preset cross-sectional morphology standards derived from process specifications or historical quality data, these standards include allowable deviations in cross-sectional perpendicularity, thresholds for cross-sectional flatness, and tolerances for local concavity and convexity. The extracted cross-sectional morphology features are compared and analyzed. For example, if the cross-sectional perpendicularity deviation exceeds 0.5°, it is determined to be an abnormal cross-sectional morphology, and the abnormal cross-sectional morphology characteristics are obtained. For the features of attachments to the cut, image segmentation technology is used to separate the attachments from the cut image. Then, based on preset attachment judgment rules, such as if the attachment area exceeds 5% of the cut area or has an irregular shape, abnormal attachment characteristics are identified, and the abnormal attachment characteristics are obtained. The abnormal characteristics of the cut edge, the abnormal cross-sectional morphology, and the abnormal attachment are integrated to form a complete set of abnormal cut quality characteristics.

[0042] Through multi-dimensional feature extraction and targeted analysis, the quality of the incision can be comprehensively and accurately evaluated. Edge, cross-sectional morphology and attachment features are extracted from the incision image, and incision edge anomalies are detected. At the same time, anomalies in cross-sectional morphology and attachments are identified to generate incision quality anomaly characteristics, which effectively improves the accuracy and comprehensiveness of incision quality detection and provides a reliable basis for timely detection and handling of incision quality problems.

[0043] Based on the unique traceability identifier, the cutting and processing data of the cloud-based online cutting platform is traced back to obtain the layout process block, cutting process block, and blanking process block corresponding to the finished steel plate cutting product.

[0044] Specifically, the unique traceability identifier of the finished steel plate is used to trace the cutting data of the cloud cutting online platform, and to obtain detailed information of the finished steel plate throughout the entire processing process, including the layout process block, the cutting process block, and the unloading process block.

[0045] The layout process block refers to the arrangement information and local density characteristics of the finished steel plate during the production layout stage. The layout information exists in the form of two-dimensional or three-dimensional graphic data. Through graphic processing algorithms, such as layout algorithms in computer-aided design (CAD) software, the arrangement of parts on the steel plate can be intuitively displayed. The local density characteristics are obtained by calculating the ratio of the total area of ​​the parts to the total area of ​​the steel plate, and are used to describe the spatial distribution of each cutting unit on the steel plate. The cutting process block includes various data from the start to the end of the cutting operation, such as key process data such as cutting equipment operating parameters, cutting path planning, cutting time nodes, and cutting environment. The unloading process block records the operation sequence, handling trajectory, fixture usage information, and unloading sequence characteristics of the finished steel plate from cutting completion to unloading. Related orders and production data are indexed in the database of the cloud-based online cutting platform using a unique traceability identifier. Combined with database retrieval technology, various processing data corresponding to the unique traceability identifier can be quickly traced back, ensuring data integrity and real-time performance.

[0046] By accurately tracing back the cutting and processing data, detailed information on the three process blocks of layout, cutting, and blanking can be obtained. This allows for the establishment of a correlation between quality anomalies and specific process parameters, production operation sequence, and layout information. This enables multidimensional causal analysis of the causes of anomalies, providing a reliable basis for timely improvement measures and effectively improving production efficiency and product quality.

[0047] Based on the cutting process blocks, multi-parameter causal analysis is performed on the multi-dimensional cutting quality anomaly characteristics to establish a cutting quality anomaly tracing map.

[0048] Furthermore, based on the cutting process blocks, multi-parameter causal analysis is performed on the multi-dimensional cutting quality anomaly characteristics to establish a cutting quality anomaly tracing map. This includes: performing multi-parameter causal tracing on the dimensional deviation anomaly characteristics based on the cutting process blocks to obtain the dimensional anomaly tracing path; performing multi-parameter causal tracing on the cut quality anomaly characteristics based on the cutting process blocks to obtain the cut anomaly tracing path; performing multi-parameter causal tracing on the cutting deformation anomaly characteristics based on the cutting process blocks to obtain the deformation anomaly tracing path; and mapping and organizing the dimensional anomaly tracing path, the cut anomaly tracing path, and the deformation anomaly tracing path to generate the cutting quality anomaly tracing map.

[0049] Furthermore, based on the cutting process blocks, multi-parameter causal tracing of the dimensional deviation anomaly characteristics is performed to obtain the dimensional anomaly tracing path, including: parsing the cutting process blocks to obtain cutting sequence information and multi-dimensional cutting feature sequences; identifying the correlations of the dimensional deviation anomaly characteristics based on the multi-dimensional cutting feature sequences to obtain a dimensional anomaly cutting relationship network; identifying the importance of different parameter correlations based on the dimensional anomaly cutting relationship network to determine the distribution of key dimensional anomaly factors; reconstructing the distribution of key dimensional anomaly factors based on the cutting sequence information to obtain a dimensional anomaly sequence factor distribution; and reconstructing the formation process of the dimensional deviation anomaly characteristics based on the dimensional anomaly sequence factor distribution to generate a dimensional anomaly tracing path.

[0050] Furthermore, the multidimensional cutting feature sequence includes a cutting process feature sequence, a cutting equipment feature sequence, and a cutting environment feature sequence.

[0051] Specifically, structured data extraction techniques, such as SQL queries and ETL processing, are used to parse the cutting process blocks, obtaining cutting sequence information and multi-dimensional cutting feature sequences. The cutting sequence information includes the order in which each steel plate is cut, the start and end times of cutting, and tool change records, which are sorted using timestamps and work order numbers to form a complete sequence. The multi-dimensional cutting feature sequences include cutting process feature sequences, cutting equipment feature sequences, and cutting environment feature sequences. The cutting process feature sequences refer to the set of process parameters directly involved in material removal and forming during steel plate cutting, arranged according to the cutting sequence or time dimension. These parameters reflect the settings and dynamic changes of cutting process parameters, including but not limited to laser or plasma power, cutting speed, cutting path trajectory, type and pressure of cutting gas, focal point position, and cutting start / stop frequency. The cutting equipment feature sequences refer to the set of features formed by the changing operating status and performance parameters of the cutting equipment over time during the cutting process. These parameters describe the equipment's health status and execution stability, specifically including machine tool model, equipment runtime, tool or nozzle wear, servo system load, motor current, laser cavity temperature, and equipment vibration amplitude. The cutting environment characteristic sequence refers to the set of characteristics formed by the environmental conditions of the cutting operation during the cutting cycle. It is used to reflect the indirect interference factors of the external environment on the cutting process, specifically including workshop temperature, humidity, environmental vibration intensity, etc.

[0052] The acquired dimensional deviation anomalies are correlated with multidimensional cutting feature sequences. First, a matrix is ​​formed by parsing the multidimensional cutting feature sequences from the cutting process blocks, such as laser power, cutting speed, and machine tool status, and the dimensional deviation anomalies of each steel plate. Each row corresponds to a finished product, and each column corresponds to a cutting parameter. Pearson correlation coefficient or Spearman rank correlation coefficient is used to statistically analyze the relationship between each cutting feature and the dimensional deviation anomaly, calculating the contribution of each cutting feature to the dimensional deviation. Then, the cutting features are used as network nodes, and the dimensional deviation anomalies are used as target nodes. When the correlation coefficient between the cutting parameters and the dimensional deviation anomalies exceeds a threshold, such as 0.3, a graph database, such as Neo4j, is used to establish edges connecting the parameter nodes and the anomaly nodes. The edge weights can be set to the absolute values ​​of the correlation coefficients, forming a dimensional deviation cutting relationship network. By obtaining a dimension anomaly cutting relationship network, we can intuitively reflect the correlation between various cutting parameters and dimension deviations, and clarify which process conditions may lead to anomalies. For example, statistical analysis shows that the correlation coefficient between laser power and dimension deviation is 0.65 and the cutting speed is 0.42. In this case, we can establish two edges in the dimension anomaly cutting relationship network with edge weights of 0.65 and 0.42, respectively, which intuitively reflects that laser power has a greater impact on dimension deviation.

[0053] The influence value of each parameter node in the dimensional anomaly cutting relationship network is calculated to account for the anomaly node. This can be achieved using the Shapley value or the edge weight accumulation method. The Shapley value measures the average contribution of each cutting parameter to the anomaly formation under different combinations, while the edge weight accumulation method directly normalizes and accumulates the edge weights connecting the dimensional anomaly nodes. Based on the influence values, key factors are identified, and their distribution across different cutting units is statistically analyzed to form a distribution table of key factors for dimensional anomalies. For example, in the dimensional anomaly cutting relationship network, if the laser power edge weight is 0.65, the cutting speed edge weight is 0.42, and the tool wear edge weight is 0.30, then the influence values ​​are ranked as follows: laser power > cutting speed > tool wear. Laser power is marked as one of the key factors, and its actual value in different cutting units is recorded to form a key factor distribution table. By combining cutting sequence information, the distribution of key factors is reconstructed using sequence constraints. This involves sorting the key factor distribution according to the cutting sequence, and then serializing the position, value, and importance of each key factor in the cutting process over time. A sliding window weighting method is used, with each window covering N consecutive steel plates. The cumulative importance or average value of key factors within each window is calculated, thus smoothing the sequence and highlighting points of anomaly concentration. Sequence constraint reconstruction allows for more accurate positioning of the key factors' roles in the cutting process. The sequence of key factors reconstructed using sequence constraints is then used as time series input to generate a distribution of dimensional anomaly sequence factors. The horizontal axis represents the cutting sequence, and the vertical axis represents the value or weight of the key factors. The anomaly contribution is overlaid to reflect the dynamic evolution of key process parameters in the cutting process and their impact on dimensional deviations. The distribution of dimensional anomaly sequence factors is mapped to the actual dimensional deviation anomaly characteristics. For each steel plate, the anomaly value is matched with the corresponding cutting parameter node to form a causal path node sequence. The causal path generation algorithm or the shortest path weighting method is used to generate the path by accumulating edge weights or influence values. Each path node is marked with the process parameters, cutting sequence and processing unit where the anomaly occurred. The edge weight value represents the factor contribution intensity, thus generating a dimensional anomaly tracing path.

[0054] Simultaneously, multi-parameter causal tracing of cut quality anomalies is performed based on the cutting process blocks to obtain the anomaly tracing path. A similar approach to handling dimensional anomalies is adopted: first, the cutting process blocks are analyzed to obtain relevant data; then, correlation analysis and feature importance assessment methods are used to identify key factors affecting cut quality, such as tool wear and cutting gas flow rate. Taking tool wear as an example, by monitoring tool usage time and the number of cuts, it was found that when tool wear exceeds 0.2mm, the probability of burrs appearing on the cut increases significantly. Then, combined with cutting sequence information, sequence constraint reconstruction is performed to reconstruct the formation process of cut quality anomalies and generate the anomaly tracing path.

[0055] To trace the causal origins of cutting deformation anomalies based on the cutting process blocks, we first analyze the cutting process block data and then use numerical simulation techniques such as finite element analysis to analyze the impact of different cutting parameters, such as cutting speed and cutting depth, on material deformation. For example, finite element simulations revealed that when the cutting depth exceeds one-third of the material thickness, the probability of significant material deformation increases by 70%. Then, by combining correlation analysis and sequence constraint reconstruction, we determine the key factors and formation processes of cutting deformation anomalies and generate a deformation anomaly tracing path. Specific implementation steps can be found in the implementation process of obtaining the dimensional anomaly tracing path, which will not be elaborated here.

[0056] The source paths for size anomalies, cut anomalies, and deformation anomalies are graphically organized and mapped onto the same knowledge graph or causal graph modeling framework. Using graph database technology, such as Neo4j, the source paths for size anomalies, cut anomalies, and deformation anomalies are stored and displayed in the form of nodes and edges. Nodes represent key factors or cutting steps, and edges represent the causal relationships between factors or the order of steps, generating a source path graph for cutting quality anomalies.

[0057] By precisely linking the abnormal characteristics of cutting quality with specific cutting processes, equipment status, and cutting sequence, the cause-and-effect tracing of dimensional deviations, cut quality, and deformation abnormalities can be achieved. By establishing a source map of cutting quality abnormalities, the causes and processes of abnormalities can be intuitively understood, thereby enabling targeted improvement measures such as adjusting cutting parameters, replacing tools, and optimizing the cutting sequence, which can effectively improve the efficiency and quality of steel plate cutting.

[0058] Based on the sorting process block and the unloading process block, the upstream and downstream dynamic correlation compensation of the cutting quality anomaly tracing map is performed to obtain the cloud cutting anomaly tracing map.

[0059] Furthermore, based on the nesting process block and the material unloading process block, the cutting quality anomaly tracing map is dynamically correlated upstream and downstream to obtain a cloud-cut anomaly tracing map. This includes: dynamically correlated upstream and downstream to the dimensional anomaly tracing path based on the nesting process block and the material unloading process block to obtain a dimensional anomaly correction path; dynamically correlated upstream and downstream to the cut anomaly tracing path based on the nesting process block and the material unloading process block to obtain a cut anomaly correction path; dynamically correlated upstream and downstream to the deformation anomaly tracing path based on the nesting process block and the material unloading process block to obtain a deformation anomaly correction path; and optimized based on the dimensional anomaly correction path, the cut anomaly correction path, and the deformation anomaly correction path to generate the cloud-cut anomaly tracing map.

[0060] Furthermore, based on the nesting process block and the material unloading process block, the upstream and downstream dynamic correlation compensation of the size anomaly tracing path is performed to obtain the size anomaly correction path, including: parsing the nesting process block to obtain nesting layout information and nesting local density features; performing interference analysis on the size deviation anomaly characteristics based on the nesting layout information and the nesting local density features to obtain the size deviation nesting interference path; parsing the material unloading process block to obtain material unloading sequence information and material unloading operation feature sequence; performing interference analysis on the size deviation anomaly characteristics based on the material unloading sequence information and the material unloading operation feature sequence to obtain the size deviation material unloading interference path; and performing upstream and downstream correlation reconstruction of the size anomaly tracing path based on the size deviation nesting interference path and the size deviation material unloading interference path to generate the size anomaly correction path.

[0061] Specifically, the nesting process is analyzed by reading nesting files, CAD / CAM nesting data, or process logs from the cloud-based online cutting platform to extract the nesting layout information of the steel plate on the mother plate. This layout information includes the spatial position, relative spacing, nesting direction, and relationships between adjacent parts within the mother plate. This information is then converted into a standardized two-dimensional spatial model, dividing the mother plate into continuous regions with spatial coordinates. Each part corresponds to its geometric contour, center point coordinates, and orientation angle. Based on this two-dimensional spatial model, a spatial neighborhood analysis method is used. Using a single part as the analysis center, a fixed radius or an adaptive neighborhood range is set (e.g., 1.5 times the bounding rectangle of the part as the neighborhood radius). The number of other parts within the neighborhood, minimum spacing, and directional consistency are statistically analyzed to calculate local nesting density characteristics, such as the number of parts per unit square meter, average part spacing, and directional concentration. When the number of parts in a certain area exceeds a preset threshold (e.g., more than 8 parts per unit square meter, or an average spacing less than 10 mm), the area is determined to have high-density nesting characteristics.

[0062] By combining layout information and cutting path data, the cutting trajectory of each part is analyzed segment by segment, discretizing the trajectory into several small line segments or trajectory units. Corresponding cutting process parameters, such as laser power, cutting speed, and cutting time, are matched to each trajectory unit to calculate the heat input per unit length. This heat input can be estimated using the power-to-speed ratio, for example, Q=P / v, where P represents laser power and v represents cutting speed. Based on the principles of heat conduction and diffusion, an empirical heat diffusion model or a simplified two-dimensional heat conduction model is introduced to spatially simulate the heat-affected zone formed by each trajectory unit within the motherboard plane. The heat-affected zone is approximated as a strip-shaped or elliptical region centered on the cutting trajectory, with its width determined by material thickness, thermal conductivity, and heat input. The diffusion radius is determined by looking up tables or empirical formulas. Next, the heat-affected zones of adjacent parts are mapped onto a unified two-dimensional spatial model. Spatial overlay analysis is performed on each heat-affected zone in the high-density layout area to calculate the overlap area or overlap length ratio of the heat-affected zones between adjacent parts. When the overlap area of ​​the heat-affected zone of a certain part with the heat-affected zone of the adjacent parts exceeds a preset threshold, such as 30%, it is determined that the part has a significant risk of thermal interference in the layout stage.

[0063] By integrating the thermal interference determination results with the high-density layout characteristics obtained from spatial neighborhood analysis, it is determined that the heat superposition is caused by excessively small part spacing, consistent arrangement direction, or concentrated cutting path. This establishes a causal link from layout factors, thermal interference characteristics to dimensional deviation anomalies. The relevant factors are then linked in time and space order to form a structured dimensional deviation layout interference path. The dimensional deviation layout interference path includes parameters such as part spacing, layout direction, local density, and degree of overlap of heat-affected zones.

[0064] The cloud-based online cutting platform reads process logs, equipment control records, and manual operation records related to the material cutting process. Structured data is extracted from the material cutting data, and the cutting behavior is parsed using work order numbers, timestamps, and equipment event codes to extract material cutting sequence information. This information includes the order in which parts separate from the motherboard, the start and end times of individual part cutting, and the interval between cutting adjacent parts. Simultaneously, material cutting operation feature sequences are extracted from equipment sensor data and operation records. These sequences include the cutting method, the number and position of support points, clamping status, flipping or secondary movement operations, and whether there are pauses or repetitive operations during the cutting process. These features characterize the changes in the force and constraint states of the parts during the cutting process. Based on the material cutting sequence information, dimensional deviation anomalies are arranged according to the order of material cutting. Analysis is conducted to determine if dimensional deviation anomalies are concentrated in consecutively cut parts. Combining the material cutting operation characteristic sequence, a stress and release interference rule is introduced to determine the stress release behavior of parts during the cutting process due to insufficient support, uneven clamping, or premature detachment from the mother plate. For example, if a part has fewer support points than a preset value or is removed before adjacent parts are fully cut, it is determined that there is a risk of material cutting interference. Furthermore, the above determination results are mapped to the characteristics of dimensional deviation anomalies. When the dimensional deviation value and the high-risk material cutting operation are highly consistent in time and object, it is confirmed that the material cutting operation has an interference effect on the dimensional deviation. Finally, the material cutting sequence nodes, material cutting operation factors, and dimensional deviation anomaly results are connected in a causal relationship to generate a dimensional deviation material cutting interference path. This path clearly indicates the specific material cutting sequence position, operation method, and its influence on the formation of dimensional deviations.

[0065] After obtaining the interference paths for dimensional deviations in the layout and the material cutting, the upstream and downstream connections of the dimensional anomaly tracing path formed by the cutting process are reconstructed. This is specifically achieved by introducing graph theory modeling and a path merging algorithm: First, the three types of paths are uniformly mapped into a directed weighted graph structure. Nodes represent key factors and process nodes affecting dimensional deviations, such as layout density, part spacing, cutting speed, laser power, material cutting sequence, and support status. Edges represent the causal relationships between factors, and edge weights are quantified based on the contribution, correlation coefficient, or risk score obtained from previous analysis. The path merging algorithm is used to fuse the layout interference path, the cutting tracing path, and the material cutting interference path. Nodes that appear repeatedly in multiple paths are merged, and the weights of their associated edges are accumulated or weighted averaged to reflect the superimposed influence of factors at multiple stages. For nodes that exist only in the upstream or downstream paths, directed connections are made to complete them based on their temporal order and causal constraints with the cutting process nodes. Finally, by searching the optimal causal path from the nesting stage to the dimensional deviation anomaly node in the fused directed weighted graph, a dimensional anomaly correction path is generated. This dimensional anomaly correction path can completely connect the key influencing factors in the entire process of nesting, cutting and blanking, and clearly present the formation mechanism of dimensional anomalies from upstream nesting interference, process fluctuations in the process to downstream blanking stress release. It can more accurately reflect the causes and processes of dimensional anomalies, thereby significantly improving the accuracy and reliability of dimensional anomaly tracing results.

[0066] The process of obtaining the cut anomaly correction path and deformation anomaly correction path is basically similar to that of obtaining the dimensional anomaly correction path, except that the processing objects are different: Based on the nesting process block and the blanking process block, the nesting file, CAD / CAM nesting data, and blanking process log are parsed to extract the nesting layout information, local density characteristics, blanking sequence information, and blanking operation feature sequence of the parts on the mother plate. The nesting layout is standardized into a two-dimensional spatial model, and the blanking operation sequence is converted into a time series for interference analysis. In the process of generating the cut anomaly correction path, based on the nesting layout information and local density characteristics, the interference that the close arrangement of parts, the overlap of adjacent cutting trajectories, or the concentration of cutting paths may cause to the cut quality is analyzed. At the same time, combined with the blanking sequence and operation characteristics, the influence of cut burrs, gaps, or edge deformation caused by insufficient support or improper operation during the blanking process is analyzed, thereby forming the cut interference path. Similarly, in the process of generating deformation anomaly correction paths, the influence of thermal superposition effects, part spacing, and arrangement direction on warping, twisting, and local deformation is analyzed using layout information and local density characteristics. Furthermore, the interference of uneven stress release during the material preparation process on finished product deformation is analyzed using the material preparation sequence and operational characteristics, forming a deformation interference path. Subsequently, the original cut anomaly tracing path or deformation anomaly tracing path based on the cutting process is merged with the corresponding interference path. A unified directed weighted graph is constructed using a path merging algorithm. Repeated nodes are merged, and edge weights are accumulated or weighted averaged. For nodes existing only upstream or downstream, directed connections are made to complete them based on time sequence and causal constraints. Finally, by searching for the optimal causal path in the merged directed weighted graph, cut anomaly correction paths and deformation anomaly correction paths are generated. These paths comprehensively present the key factors affecting cut quality or finished product deformation and their order of action throughout the entire process from layout and cutting to material preparation, providing a unified and visualized tracing basis for anomaly location, process optimization, and early warning.

[0067] After obtaining the correction paths for size anomalies, cut anomalies, and deformation anomalies, the three types of correction paths are used as input to optimize the cutting quality anomaly tracing map. Specifically, the three correction paths are mapped to the nodes and edges of the cutting quality anomaly tracing map, and the causal connections and weights of the original nodes in the cutting quality anomaly tracing map are updated and strengthened. The nodes represent the key process factors and abnormal features in the cutting process, the nesting process, and the material feeding process, and the edges represent the causal relationships between different factors and their contribution intensity to the formation of anomalies. During the optimization process, for key nodes appearing in the three correction paths, such as part spacing, cutting speed, support status, and heat-affected zone superposition, if a corresponding node already exists in the cutting quality anomaly tracing map, its edge weights are updated by accumulation or weighted averaging to reflect the superposition effect of multi-stage factors. For example, in the three correction paths, for the key node of part spacing, the edge weight of the influence of part spacing on cutting quality anomalies is calculated to be 0.3 in the first correction path, 0.4 in the second correction path, and 0.2 in the third correction path. If the edge weights are updated by accumulation, the edge weights corresponding to part spacing in the three paths are directly added together, i.e., 0.3 + 0.4 + 0.2 = 0.9. Then, the updated edge weight of part spacing in the cutting quality anomaly tracing map will be 0.9. If a weighted average method is used, each path is first assigned a weight based on its importance. Assuming the weights of the three paths are 0.2, 0.5, and 0.3 respectively, the edge weight after weighted average is 0.3×0.2 + 0.4×0.5 + 0.2×0.3 = 0.32. If a new node in a path does not yet exist in the cutting quality anomaly tracing map, the node and its directed edges are completed according to the time sequence and causal constraints, forming a continuous causal link between upstream layout factors, cutting parameters, and downstream material feeding factors. After the above optimization, the generated cloud cutting anomaly tracing map can completely and intuitively present the causal relationships between various anomalies and their key influencing factors throughout the entire process from layout and cutting to material feeding.

[0068] By integrating the upstream and downstream influences of the nesting and blanking processes into the cutting quality anomaly tracing map, the causes of various anomalies such as dimensional deviations, cut quality, and deformation can be fully and comprehensively presented in the map. This realizes the transformation of anomaly tracing from a single cutting process to a dynamic correlation analysis of the entire process, multiple stages, and multiple factors. This improves the accuracy and reliability of steel plate cutting anomaly identification, as well as the effectiveness and pertinence of production process optimization and anomaly early warning, effectively enhancing the quality management capabilities in the steel plate cutting process.

[0069] Furthermore, the method also includes: generating a cloud cutting anomaly early warning command based on the cloud cutting anomaly tracing map.

[0070] Specifically, the cloud cutting anomaly tracing map is dynamically analyzed, quantifying the node and edge information in the map into assessable anomaly risk indicators. Nodes represent key process factors and corresponding anomaly characteristics during layout, cutting, and material preparation, such as insufficient part spacing, excessive cutting speed, insufficient support, or overlapping heat-affected zones. Edges represent the causal relationships between factors and their contribution to anomaly formation. Warning thresholds are set, such as a single node's risk contribution exceeding 0.6 or a path's cumulative weight exceeding 0.8, to identify high-risk nodes and key causal links. Simultaneously, by combining time-series information, the occurrence time of high-risk nodes or paths in the processing flow is sorted and predicted, automatically generating targeted cloud cutting anomaly warning instructions. These instructions include warning categories, affected processes, possible process parameter anomalies, and suggested intervention measures. Warning categories include dimensional deviation anomalies, cut anomalies, or deformation anomalies, with corresponding intervention measures such as adjusting cutting speed, increasing support, or optimizing layout, thereby achieving early identification and accurate warning of potential anomalies.

[0071] By extracting and transforming multi-stage, multi-factor abnormal information from the cloud-based abnormality tracing map into executable early warning instructions, the production management system can respond promptly before or in the early stages of anomalies. This not only improves the timeliness and accuracy of anomaly detection but also provides actionable decision-making basis for process adjustment, quality control, and production safety. As a result, proactive monitoring and intelligent management of the entire steel plate cutting process are achieved, further improving the efficiency of steel plate cutting quality control and process optimization.

[0072] Example 2 is based on the same inventive concept as the traceability information management method based on cloud-cutting online data in the previous examples, such as... Figure 2 As shown, this application provides a traceability information management system based on cloud-based online data, wherein the traceability information management system based on cloud-based online data includes:

[0073] The order query module 11 is used to obtain the unique traceability identifier of the finished steel plate cutting product and query the online cutting platform based on the unique traceability identifier to obtain matching online cutting orders; the anomaly detection module 12 is used to perform quality anomaly detection on the finished steel plate cutting product based on the matching online cutting order to obtain multi-dimensional cutting quality anomaly characteristics; the data backtracking module 13 is used to perform cutting processing data backtracking on the online cutting platform based on the unique traceability identifier to obtain the layout process block, cutting process block, and unloading process block corresponding to the finished steel plate cutting product; the traceability analysis module 14 is used to perform multi-parameter causal traceability analysis on the multi-dimensional cutting quality anomaly characteristics based on the cutting process block to establish a cutting quality anomaly traceability map; the correlation compensation module 15 is used to perform upstream and downstream dynamic correlation compensation on the cutting quality anomaly traceability map based on the layout process block and the unloading process block to obtain the online cutting anomaly traceability map.

[0074] Furthermore, the anomaly detection module 12 is also used to: input the unique traceability identifier into the cloud cutting online platform to obtain the finished product size data, cut image, and finished product shape image corresponding to the steel plate cut finished product; perform size consistency detection on the finished product size data according to the matched cloud cutting order to obtain size deviation anomaly characteristics; perform cut quality detection on the steel plate cut finished product according to the cut image to obtain cut quality anomaly characteristics; perform deformation detection on the steel plate cut finished product according to the finished product shape image to obtain cutting deformation anomaly characteristics, and generate the multi-dimensional cutting quality anomaly characteristics by combining the size deviation anomaly characteristics and the cut quality anomaly characteristics.

[0075] Furthermore, the anomaly detection module 12 is also used to: extract features of the cut edge region, the cut cross-sectional morphology features, and the features of the attachments to the cut based on the cut image; perform supervised training on a deep convolutional neural network based on the cut edge anomaly detection record set to obtain a cut edge anomaly detection network; input the cut edge region features into the cut edge anomaly detection network to obtain cut edge anomaly characteristics; perform cross-sectional morphology anomaly identification based on the cut cross-sectional morphology features to obtain cut cross-sectional morphology anomaly characteristics; perform cut attachment anomaly identification based on the attachment features to obtain cut attachment anomaly characteristics; and combine the cut edge anomaly characteristics and the cut cross-sectional morphology anomaly characteristics to generate the cut quality anomaly characteristics.

[0076] Furthermore, the source tracing and analysis module 14 is also used to: perform multi-parameter causal tracing of the dimensional deviation anomaly characteristics based on the cutting process blocks to obtain the dimensional anomaly source path; perform multi-parameter causal tracing of the cut quality anomaly characteristics based on the cutting process blocks to obtain the cut anomaly source path; perform multi-parameter causal tracing of the cutting deformation anomaly characteristics based on the cutting process blocks to obtain the deformation anomaly source path; and perform graphing and sorting of the dimensional anomaly source path, the cut anomaly source path, and the deformation anomaly source path to generate the cutting quality anomaly source graph.

[0077] Furthermore, the source tracing and analysis module 14 is also used to: parse the cutting process block to obtain cutting sequence information and multi-dimensional cutting feature sequence; identify the correlation between the size deviation anomaly characteristics according to the multi-dimensional cutting feature sequence to obtain a size anomaly cutting relationship network; identify the importance of the correlation between different parameters based on the size anomaly cutting relationship network to determine the distribution of key factors of size anomaly; reconstruct the distribution of key factors of size anomaly according to the cutting sequence information to obtain the size anomaly sequence factor distribution; and restore the formation process of the size deviation anomaly characteristics according to the size anomaly sequence factor distribution to generate a size anomaly source tracing path.

[0078] Furthermore, the correlation compensation module 15 is also used to: perform upstream and downstream dynamic correlation compensation on the dimensional anomaly tracing path based on the nesting process block and the material unloading process block to obtain a dimensional anomaly correction path; perform upstream and downstream dynamic correlation compensation on the cut anomaly tracing path based on the nesting process block and the material unloading process block to obtain a cut anomaly correction path; perform upstream and downstream dynamic correlation compensation on the deformation anomaly tracing path based on the nesting process block and the material unloading process block to obtain a deformation anomaly correction path; and optimize the cutting quality anomaly tracing map based on the dimensional anomaly correction path, the cut anomaly correction path, and the deformation anomaly correction path to generate the cloud cutting anomaly tracing map.

[0079] Furthermore, the correlation compensation module 15 is also used to: parse the layout process block to obtain layout information and local density features; perform interference analysis on the size deviation anomaly characteristics based on the layout information and local density features to obtain the size deviation layout interference path; parse the material feeding process block to obtain material feeding sequence information and material feeding operation feature sequence; perform interference analysis on the size deviation anomaly characteristics based on the material feeding sequence information and material feeding operation feature sequence to obtain the size deviation material feeding interference path; and perform upstream and downstream correlation reconstruction on the size anomaly tracing path based on the size deviation layout interference path and the size deviation material feeding interference path to generate the size anomaly correction path.

[0080] Furthermore, the traceability analysis module 14 is also used to: the multidimensional cutting feature sequence includes a cutting process feature sequence, a cutting equipment feature sequence, and a cutting environment feature sequence.

[0081] Furthermore, the system is also used to: generate a cloud-cutting anomaly early warning command based on the cloud-cutting anomaly tracing map.

[0082] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The traceability information management method and specific examples based on cloud-cutting online data in the aforementioned embodiment one are also applicable to the traceability information management system based on cloud-cutting online data in this embodiment. Through the foregoing detailed description of the traceability information management method based on cloud-cutting online data, those skilled in the art can clearly understand the traceability information management system based on cloud-cutting online data in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0083] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0084] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for managing traceability information based on cloud-based online data, characterized in that, The method includes: Obtain a unique traceability identifier for the finished steel plate cutting product, and use the unique traceability identifier to query orders on the cloud cutting online platform to obtain matching cloud cutting orders; Based on the matched cloud cutting order, the steel plate cutting finished product is subjected to quality anomaly detection to obtain multi-dimensional cutting quality anomaly characteristics; Based on the unique traceability identifier, the cutting and processing data of the cloud cutting online platform is traced back to obtain the layout process block, cutting process block and unloading process block corresponding to the steel plate cutting finished product; Based on the cutting process blocks, multi-parameter causal source analysis is performed on the multi-dimensional cutting quality anomaly characteristics to establish a cutting quality anomaly source map. Based on the sorting process block and the unloading process block, the upstream and downstream dynamic correlation compensation of the cutting quality anomaly tracing map is performed to obtain the cloud cutting anomaly tracing map.

2. The traceability information management method based on cloud-cutting online data as described in claim 1, characterized in that, Based on the matched cloud cutting order, the steel plate cutting finished product is subjected to quality anomaly detection to obtain multi-dimensional cutting quality anomaly characteristics, including: Input the unique traceability identifier into the cloud-cutting online platform to obtain the finished product size data, cut image and finished product shape image corresponding to the steel plate cut product; Based on the matched cloud cutting order, perform a size consistency check on the finished product size data to obtain abnormal size deviation characteristics; The cut quality of the steel plate is inspected based on the cut image to obtain abnormal characteristics of the cut quality. Deformation detection is performed on the steel plate cut product based on the finished product morphology image to obtain abnormal cutting deformation characteristics. The multidimensional cutting quality abnormal characteristics are generated by combining the abnormal size deviation characteristics and the abnormal cut quality characteristics.

3. The traceability information management method based on cloud-cutting online data as described in claim 2, characterized in that, The steel plate is cut to a finished product based on the cut image to detect cut quality abnormalities, including: Based on the incision image, extract the features of the incision edge region, the incision cross-sectional morphology, and the features of the incision attachments; The deep convolutional neural network is trained under supervision based on the cut edge anomaly detection record set to obtain the cut edge anomaly detection network. The features of the cut edge region are input into the cut edge anomaly detection network to obtain the abnormal characteristics of the cut edge. Based on the cross-sectional morphological characteristics of the cut, cross-sectional morphology anomalies are identified to obtain the characteristics of the cross-sectional morphology anomalies. Based on the characteristics of the attachment material at the cut, anomalies in the cut are identified, and the characteristics of the abnormal attachment at the cut are obtained. The abnormal characteristics of the cut quality are generated by combining the abnormal characteristics of the cut edge and the abnormal characteristics of the cut cross-section morphology.

4. The traceability information management method based on cloud-cut online data as described in claim 1, characterized in that, Based on the cutting process blocks, a multi-parameter causal analysis is performed on the multi-dimensional cutting quality anomaly characteristics to establish a cutting quality anomaly tracing map, including: Based on the cutting process blocks, multi-parameter causal tracing of the abnormal size deviation characteristics is performed to obtain the size abnormality tracing path; Based on the cutting process blocks, multi-parameter causal tracing of abnormal cut quality characteristics is performed to obtain the abnormal cut source path; Based on the cutting process blocks, multi-parameter causal tracing of the abnormal characteristics of cutting deformation is performed to obtain the abnormal deformation tracing path. The source paths for the size anomalies, the cut anomalies, and the deformation anomalies are mapped and organized to generate a source map for the cutting quality anomalies.

5. The traceability information management method based on cloud-cutting online data as described in claim 4, characterized in that, Based on the cutting process blocks, multi-parameter causal tracing is performed on the dimensional deviation anomaly characteristics to obtain the dimensional anomaly tracing path, including: The cutting process blocks are analyzed to obtain cutting order information and multi-dimensional cutting feature sequences; Based on the multidimensional cutting feature sequence, the abnormal size deviation characteristics are associated and identified to obtain a cutting relationship network with abnormal size; Based on the aforementioned size anomaly cutting relationship network, the importance of the correlation between different parameters is identified, and the distribution of key factors of size anomalies is determined. Based on the cutting sequence information, the distribution of key factors for size anomalies is reconstructed by performing sequence constraints to obtain the distribution of sequential factors for size anomalies. Based on the distribution of factors contributing to the size anomaly sequence, the formation process of the size deviation anomaly characteristics is reconstructed, generating a path for tracing the size anomaly source.

6. The traceability information management method based on cloud-cutting online data as described in claim 1, characterized in that, Based on the sorting process block and the unloading process block, the cutting quality anomaly tracing map is dynamically correlated and compensated upstream and downstream to obtain the cloud cutting anomaly tracing map, including: Based on the sorting process block and the unloading process block, the upstream and downstream dynamic correlation compensation of the size anomaly tracing path is performed to obtain the size anomaly correction path; Based on the sorting process block and the unloading process block, the upstream and downstream dynamic correlation compensation of the cut anomaly tracing path is performed to obtain the cut anomaly correction path; Based on the sorting process block and the unloading process block, the deformation anomaly tracing path is dynamically correlated and compensated upstream and downstream to obtain the deformation anomaly correction path; The cutting quality anomaly tracing map is optimized based on the size anomaly correction path, the cut anomaly correction path, and the deformation anomaly correction path to generate the cloud cutting anomaly tracing map.

7. The traceability information management method based on cloud-cutting online data as described in claim 6, characterized in that, Based on the nesting process block and the unloading process block, dynamic upstream and downstream correlation compensation is performed on the dimensional anomaly tracing path to obtain the dimensional anomaly correction path, including: The sorting process blocks are analyzed to obtain sorting layout information and local density characteristics; Based on the layout information and local density characteristics of the layout, interference analysis is performed on the abnormal characteristics of size deviation to obtain the size deviation layout interference path; The material feeding process block is analyzed to obtain material feeding sequence information and material feeding operation feature sequence; Based on the material feeding sequence information and the material feeding operation feature sequence, interference analysis is performed on the abnormal characteristics of the size deviation to obtain the material feeding interference path with size deviation. Based on the dimensional deviation layout interference path and the dimensional deviation material cutting interference path, the upstream and downstream correlation reconstruction of the dimensional anomaly tracing path is performed to generate the dimensional anomaly correction path.

8. The traceability information management method based on cloud-cutting online data as described in claim 5, characterized in that, The multidimensional cutting feature sequence includes a cutting process feature sequence, a cutting equipment feature sequence, and a cutting environment feature sequence.

9. The traceability information management method based on cloud-cutting online data as described in claim 1, characterized in that, Based on the cloud shedding anomaly tracing map, a cloud shedding anomaly early warning command is generated.

10. A traceability information management system based on cloud-based online data, characterized in that: The step of implementing the traceability information management method based on cloud-cutting online data according to any one of claims 1 to 9, wherein the traceability information management system based on cloud-cutting online data comprises: The order query module is used to obtain the unique traceability identifier of the steel plate cutting finished product, and to query the cloud cutting online platform for orders based on the unique traceability identifier to obtain matching cloud cutting orders; Anomaly detection module is used to perform quality anomaly detection on the steel plate cut finished product according to the matching cloud cutting order, and obtain multi-dimensional cutting quality anomaly characteristics; The data backtracking module is used to backtrack the cutting and processing data of the cloud cutting online platform based on the unique traceability identifier, and to obtain the layout process block, cutting process block and unloading process block corresponding to the steel plate cutting finished product; The source tracing and analysis module is used to perform multi-parameter causal source tracing and analysis on the multi-dimensional cutting quality anomaly characteristics based on the cutting process blocks, and to establish a cutting quality anomaly source tracing map; The correlation compensation module is used to perform upstream and downstream dynamic correlation compensation on the cutting quality anomaly tracing map based on the sorting process block and the material feeding process block, and obtain the cloud cutting anomaly tracing map.

Images