A method for managing laser etching label production data based on the Internet of Things

Abstract

The present invention relates to the technical field of electronic digital data processing, and specifically relates to a method for managing production data of laser-etched labels based on the Internet of Things, including: collecting multi-dimensional production data and images of each laser-etched label, and taking the multi-dimensional production data of any laser-etched label as a data point; clustering the data points based on the adaptive Euclidean distance between the data points to obtain multiple clustering clusters, calculating the average Euclidean distance between each data point in the clustering cluster and other data points, and in response to the ratio of the average Euclidean distance to the minimum value of the average Euclidean distance between each data point in the clustering cluster and other data points being greater than a set threshold, the data point is an abnormal data point. The present invention solves the problem that there is still a low detection accuracy in the detection accuracy of laser etching.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing. More specifically, the present invention relates to a method for managing laser etching label production data based on the Internet of Things. Background Art

[0002] Laser etching technology is a precision machining technology widely used in industries such as electronic manufacturing, automotive industry, medical devices, and packaging printing. By using a high-energy laser beam to mark, engrave, or remove on the material surface, permanent identification or patterns can be formed. Compared with traditional marking methods such as mechanical engraving and inkjet printing, laser etching has significant advantages such as non-contact processing, high precision, high contrast, wear resistance, and wide application range, and can achieve high-quality identification on various materials such as metals, plastics, glasses, and ceramics. However, in the actual production process, the laser etching process is affected by various factors such as processing parameters, equipment status, and environmental factors, and problems such as uneven etching depth, edge burrs, blurred characters, and positioning offset are likely to occur, resulting in a decline in product quality. Therefore, how to intelligently manage the data in the laser etching label production process to improve production quality, optimize process parameters, and reduce the defect rate is the focus of current industry attention.

[0003] In recent years, the rapid development of the Internet of Things technology has provided new solutions for intelligent manufacturing. The Internet of Things combines sensors, intelligent terminals, network communication, and data analysis to achieve real-time monitoring and remote control of key data such as production equipment, process parameters, and environmental status. In laser etching label production, the Internet of Things can integrate industrial cameras, photoelectric sensors, temperature and humidity sensors, laser power detection modules, etc. to achieve comprehensive data collection and analysis of the production process, thereby improving the intelligent management level.

[0004] Traditional laser etching quality inspection mainly relies on manual visual inspection or offline inspection equipment, but these methods have many limitations. Manual inspection has low efficiency and is affected by subjective factors, making it difficult to ensure the consistency of inspection results and fine-quality control; offline inspection has latency and cannot detect and correct defects in real time during the etching process, resulting in the accumulation of defects in the production process and affecting the quality of the final product. Therefore, traditional inspection methods are difficult to balance high efficiency and high precision, resulting in problems with low detection accuracy. Summary of the Invention

[0005] To solve the problem of still having low detection accuracy in terms of detection accuracy proposed in the above background art, the present invention provides the following solutions.

[0006] The present invention provides a method for managing production data of laser-engraved labels based on the Internet of Things, including: collecting multi-dimensional production data and images of each laser-engraved label at different times, and taking the multi-dimensional production data of any laser-engraved label as a data point; clustering the data points based on the adaptive Euclidean distance between the data points to obtain multiple clustering clusters, obtaining the average Euclidean distance between each data point in the clustering cluster and other data points, and in response to the ratio of the average Euclidean distance to the minimum value of the average Euclidean distance between each data point in the clustering cluster and other data points being greater than a set threshold, the data point is an abnormal data point; wherein, the adaptive Euclidean distance is , is the texture Euclidean distance between the images corresponding to the th data point and the th data point, is the Euclidean distance between the th data point and the th data point, is the average value of the texture Euclidean distances between the images corresponding to all pairs of data points among all data points; the texture Euclidean distance is: , is the importance degree of the th texture, is the number of texture types, is the sum of the importance degrees of all textures, is the frequency of the th data point corresponding to the image where the th texture appears, is the frequency of the th data point corresponding to the image where the th texture appears, and the importance degree characterizes the regularity degree of the texture.

[0007] Through the above technical solution, by collecting the multi-dimensional production data and images of laser-engraved labels, and performing clustering analysis based on the adaptive Euclidean distance, by analyzing the average distance of data points within the clustering cluster, data points with a large deviation degree are screened out as abnormal data points, avoiding misjudgment that may be caused by simply relying on the traditional Euclidean distance, thereby improving the accuracy and robustness of abnormal detection.

[0008] Further, the importance degree is , where is the th frequency value of the th clustering cluster for the th texture, , , are respectively the th texture in the The The frequency value of the cluster center of the is the total number of clustering clusters, is the total number of frequency values in the is a preset hyperparameter, and the clustering clusters are obtained by acquiring the frequencies of various textures appearing in different images and clustering the frequencies.

[0009] The above technical solution quantifies the importance of textures by calculating the frequency distribution characteristics of various textures in different clustering clusters and combining the frequency changes of the cluster centers, thereby enhancing the discrimination ability of the texture Euclidean distance, being able to adaptively evaluate the regularity of textures, making the contributions of different textures to the overall difference more reasonable when calculating the adaptive Euclidean distance between data points, and effectively avoiding the interference caused by some texture features to anomaly detection due to large local changes.

[0010] Further, the multi-dimensional production data includes: laser etching speed, laser etching intensity, and etching surface temperature.

[0011] Further, a CCD camera or a CMOS camera is used to collect images of each laser etching label.

[0012] Further, the DBSCAN clustering algorithm is used for clustering.

[0013] The above technical solution improves the clustering ability for data with complex distributions by introducing a density-based clustering method to adaptively identify data clusters with different density distributions. Combining the analysis of the importance of texture features and the calculation of the adaptive Euclidean distance enables the clustering results to more accurately reflect the true associations between data points and effectively distinguish normal data points from abnormal data points.

[0014] Further, the set threshold is 1.6.

[0015] Further, the texture is specifically:

[0016] Obtain the gray-level co-occurrence matrix of the image, and calculate texture features including at least contrast, correlation, energy, and entropy based on the gray-level co-occurrence matrix to characterize the texture of the image.

[0017] The above technical solution extracts the deep texture features of the image by constructing a gray-level co-occurrence matrix, including contrast, correlation, energy, and entropy, which characterize the texture information of the image from multiple dimensions and make the calculation of the texture Euclidean distance more accurate. Contrast can measure the clarity of the texture, correlation reflects the regularity of the texture structure, energy reflects the uniformity of the texture, and entropy is used to describe the complexity of the texture. The combination of these features enables the differences between data points to be characterized more comprehensively.

[0018] Further, it also includes standardizing the multi-dimensional production data.

[0019] Further, it also includes grayscale processing of the image.

[0020] Further, when an abnormal data point is detected, the laser etching label corresponding to the abnormal data point is removed.

[0021] The above technical solution can effectively prevent defective labels from flowing into subsequent processes, improve the consistency and reliability of product quality, by identifying and removing the laser etching labels corresponding to abnormal data points. Through automatic removal of abnormal labels, the workload of manual inspection is reduced, production efficiency is improved, and at the same time, quality problems caused by abnormal labels are avoided, reducing rework and scrap rates. It strengthens the intelligent level of quality control, makes the laser etching process more accurate and stable, and optimizes the reliability and traceability of the entire production process.

[0022] The beneficial effects of the present invention are as follows:

[0023] The present invention combines multi-dimensional production data and image data, and uses adaptive Euclidean distance and texture feature analysis to accurately detect and manage abnormal data points in the production process of laser etching labels. Through cluster analysis and comprehensive evaluation of texture features, abnormal situations that may occur in the production process can be effectively identified, improving the accuracy and reliability of production data. Extracting texture features using the gray-level co-occurrence matrix of the image helps to accurately distinguish the subtle differences between different data points, further enhancing the ability to identify abnormal data points. When an abnormal data point is detected, the corresponding laser etching label is removed in a timely manner, which can prevent defective products from flowing in, improve product quality and production efficiency. The entire process improves the intelligent level of laser etching label production through automated data management and processing, reduces the need for manual intervention, and promotes the optimization and intelligent management of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flowchart schematically showing a method for managing production data of laser etching labels based on the Internet of Things according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] An embodiment of a method for managing production data of laser-etched labels based on the Internet of Things.

[0026] As Figure 1 shown, a flowchart of a method for managing production data of laser-etched labels based on the Internet of Things according to an embodiment of the present invention includes the following steps:

[0027] S1: Collect multi-dimensional production data and its images of each laser-etched label at different times, and regard the multi-dimensional production data of any laser-etched label as a data point.

[0028] In one embodiment, the multi-dimensional production data includes: laser etching speed, laser etching intensity, and etching surface temperature. These key parameters directly determine the etching effect, edge clarity, and uniformity of material removal. In the laser etching process, due to possible fluctuations in the external environment, equipment status, and material properties, the etching effects under the same process parameters may vary. Therefore, it is crucial to perform standardization processing on the collected multi-dimensional production data.

[0029] Exemplarily, Z-score standardization can be adopted for the multi-dimensional production data; through standardization, the deviation caused by different data scales or unit differences can be eliminated, the comparability of data in different batches can be improved, and subsequent analysis and optimization can be made more accurate.

[0030] In addition, in order to further improve the etching quality monitoring and product consistency, a high-precision CCD camera or CMOS camera is used to collect images of each laser-etched label, and advanced image processing algorithms are used to perform grayscale processing on the collected images. Grayscale processing can not only reduce the complexity of image information but also enhance the contrast of the etching area, making the etching profile and micro-defects clearer, which is convenient for subsequent quality detection, edge recognition, and depth analysis. Combining the image information and the multi-dimensional production data after standardization processing, an efficient quality assessment can be established, realizing precise control of the laser etching process, reducing etching defects, improving the overall production efficiency, and providing solid data support for intelligent detection and process optimization, thereby improving the product qualification rate and consistency and meeting the strict requirements of high-precision manufacturing.

[0031] S2: Cluster the data points based on the adaptive Euclidean distance between the data points to obtain multiple clusters.

[0032] In one embodiment, the DBSCAN algorithm is used for clustering. The adaptive Euclidean distance between the th data point and the th data point is is the​ Data points and The texture Euclidean distance between the images corresponding to the data points is For the Data points and The Euclidean distance between data points is is the average value of the texture Euclidean distance between the corresponding images of all data points;

[0033] By combining adaptive Euclidean distance with density clustering method, the similarity calculation between data points is made more reasonable, which improves the accuracy and robustness of clustering. Adaptive Euclidean distance takes into account the texture features of the image and the differences in multi-dimensional production data at the same time. It measures the similarity of image details through texture Euclidean distance, and combines traditional Euclidean distance to calculate the overall relationship between data points, making the distance metric more in line with the actual situation. The average value of global texture difference is used for normalization adjustment to avoid the imbalance between data of different scales and make clustering more stable. Combined with density clustering method, there is no need to preset the number of clusters, which can effectively identify complex clustering structures and improve the detection ability of abnormal data points.

[0034] The texture Euclidean distance is: , For the The importance of the texture, is the number of texture types, is the sum of the importance of all textures, For the The data point corresponds to the The frequency of the texture. For the The data point corresponds to the The frequency of occurrence of the texture;

[0035] The texture is specifically:

[0036] A gray level co-occurrence matrix of the image is obtained, and texture features including at least contrast, correlation, energy and entropy are calculated based on the gray level co-occurrence matrix to characterize the texture of the image.

[0037] By constructing a gray-level co-occurrence matrix to extract the key texture features of an image and using the texture Euclidean distance to measure the texture similarity between data points, the quantization of image features becomes more comprehensive and accurate. By calculating texture features in multiple dimensions such as contrast, correlation, energy, and entropy, the structure, complexity, and uniformity of the image can be fully characterized, making the texture analysis more representative. Combining the weighted calculation according to the importance of the texture, the contributions of different textures to the distance metric are reasonably allocated, thus reducing the interference of local texture fluctuations on the overall analysis. Further, by calculating the weighted texture differences between data points, different categories of images can be more accurately distinguished during the clustering process, improving the ability to identify abnormal data points.

[0038] The importance degree is , where is the -th frequency value of the -th texture in the -th clustering cluster, , , are the frequency values of the cluster centers of the -th texture in the -rd, -th, and -th clustering clusters respectively, is the total number of clustering clusters, is the total number of frequency values in the -th clustering cluster, is a preset hyperparameter, and the clustering clusters are obtained by acquiring the frequencies of various textures appearing in different images and clustering the frequencies.

[0039] By calculating the frequency distribution of texture features in different clustering clusters and combining the change amount of the cluster center to quantify the importance of the texture, the contribution of the texture features in the similarity calculation becomes more accurate. By analyzing the frequency fluctuation situation within different clustering clusters, the importance of the texture can be dynamically adjusted, giving greater weight to textures with higher regularity and reducing the influence on textures with larger fluctuations, thereby enhancing the stability and discrimination ability of the texture Euclidean distance. Using the exponential function to normalize the frequency deviation can effectively reduce the interference of abnormal points or local extreme values on the calculation result, making the role of texture features in clustering analysis and anomaly detection more reasonable.

[0040] In another embodiment, the importance degree is , where is the -th frequency value of the -th texture in the -th clustering cluster, is the -th texture in the The frequency value of the cluster center of the clusters, is the total number of clusters, For the The total number of frequency values in the clusters, To preset hyperparameters, For the Texture The standard deviation of the frequency values in the clusters is obtained by obtaining the frequencies of occurrence of various textures in different images and clustering the frequencies.

[0041] By calculating the frequency distribution of texture features in different clusters and introducing the standard deviation as a factor to measure texture volatility, the calculation of texture importance is made more refined. Using the standard deviation of the frequency value to adjust the weight of the texture can effectively distinguish between stable textures and textures with large changes, so that when calculating the similarity between data points, the contribution of stable textures is greater, and the influence of textures with strong volatility is appropriately weakened, thereby improving the reliability of the texture Euclidean distance. The application of the exponential function further enhances the robustness to local outliers, making the calculation results smoother and avoiding the interference of a single extreme value on the overall clustering analysis.

[0042] S3: Calculate the ratio of the average Euclidean distance of the data points within the cluster to the minimum average Euclidean distance within the cluster to identify abnormal data points.

[0043] In one embodiment, the average Euclidean distance between each data point in the cluster and other data points is calculated, and in response to a data point whose ratio of the average Euclidean distance to the minimum value of the average Euclidean distance between each data point in the cluster and other data points is greater than a set threshold, the data point is an abnormal data point;

[0044] The threshold value may be set to 1.6, and may also be set according to actual conditions.

[0045] When an abnormal data point is detected, the laser etched label corresponding to the abnormal data point is removed.

[0046] The solution of the present invention realizes data management and anomaly detection in the production process of laser-etched labels by collecting multi-dimensional production data and images of laser-etched labels, and combining the adaptive Euclidean distance and clustering analysis methods. By accurately calculating the texture features of the images and using clustering techniques to identify the relationships between data points, abnormal data points can be effectively identified and eliminated, improving the quality control level of production data. In addition, the application of standardization processing and grayscale processing helps to enhance the consistency and stability of the data, thereby improving the accuracy and reliability of the entire production process. Through this method, abnormal situations in the production process can be detected and processed in a timely manner, reducing the occurrence of unqualified labels and ensuring the high quality and production efficiency of laser-etched labels.

[0047] In the description of this specification, "a plurality of" and "several" mean at least two, such as two, three or more, etc., unless otherwise specifically defined.

[0048] Although this specification has shown and described multiple embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, alterations and alternative ways without departing from the spirit and scope of the present invention. It should be understood that various alternative solutions to the embodiments of the present invention described herein may be employed in the practice of the present invention.

Claims

1. A method for managing laser etching label production data based on the Internet of Things, characterized in that, Including: Collecting multi-dimensional production data and its images of each laser-etched label at different times, and taking the multi-dimensional production data of any laser-etched label as a data point; Clustering the data points based on the adaptive Euclidean distance between the data points to obtain multiple clustering clusters, obtaining the average Euclidean distance between each data point in the clustering cluster and other data points, and responding to the data point whose ratio of the average Euclidean distance to the minimum value of the average Euclidean distance between each data point in the clustering cluster and other data points is greater than the set threshold as an abnormal data point; Among them, the adaptive Euclidean distance is , is the texture Euclidean distance between the images corresponding to the th data point and the th data point, is the Euclidean distance between the th data point and the th data point, is the average value of the texture Euclidean distances between the images corresponding to all pairs of data points; The texture Euclidean distance is as follows: , is the importance degree of the -th texture, is the number of texture types, is the sum of the importance degrees of all textures, is the frequency of the -th texture appearing in the image corresponding to the -th data point, is the frequency of the -th texture appearing in the image corresponding to the -th data point, and the importance degree represents the regularity degree of the texture; The importance level is , where is the -th frequency value of the -th texture in the -th clustering cluster, , , are the frequency values of the cluster centers of the -th texture in the -th, -th, and -th clustering clusters respectively, is the total number of clustering clusters, is the -th total number of frequency values in the clustering cluster, is a preset hyperparameter, and the clustering clusters are obtained by acquiring the frequencies of various textures appearing in different images and clustering the frequencies.

2. The method for managing production data of laser etching labels based on the Internet of Things according to claim 1, wherein, The multi-dimensional production data includes: laser etching speed, laser etching intensity, and etching surface temperature.

3. A method for managing laser etching label production data based on the Internet of Things according to claim 1, characterized in that, It is beneficial to use a CCD camera or a CMOS camera to collect images of each laser-etched label.

4. A method for managing laser etching label production data based on the Internet of Things according to claim 1, characterized in that, Use the DBSCAN clustering algorithm for clustering.

5. A method for managing production data of laser etching labels based on the Internet of Things according to claim 1, characterized in that The set threshold is 1.

6.

6. The method for managing production data of laser etching labels based on the Internet of Things according to claim 1, characterized in that, The texture is specifically: Obtaining the gray-level co-occurrence matrix of the image, and calculating texture features including at least contrast, correlation, energy, and entropy based on the gray-level co-occurrence matrix to characterize the texture of the image.

7. A method for managing laser etching label production data based on the Internet of Things according to claim 1, characterized in that, It also includes normalizing the multi-dimensional production data.

8. A method for managing production data of laser etching labels based on the Internet of Things according to claim 1, characterized in that, It also includes grayscaling the image.

9. A method for managing laser etching label production data based on the Internet of Things according to claim 1, characterized in that When an abnormal data point is detected, the laser-etched label corresponding to the abnormal data point is removed.

Images