Online workpiece classification and quality assessment method and system based on edge computing

Abstract

The present application relates to the field of workpiece evaluation technology, and provides an online workpiece classification and quality evaluation method and system based on edge computing, including a recognition algorithm model, a classification algorithm model and a quality evaluation algorithm model of workpiece categories pre-stored in a central cloud; in response to image information sent by a camera, the edge node recognizes the image through the recognition algorithm model, obtains the corresponding classification algorithm model from the central cloud according to the recognized workpiece category, and configures the classification algorithm program; classifies defects in the image or video information through the classification algorithm program to obtain classification information; packages the classification information and workpiece category information and sends them to the central cloud; the central cloud configures the corresponding quality evaluation algorithm program according to the workpiece category information, and after the configuration is completed, the quality of the workpiece is evaluated through the quality evaluation program according to the classification information. This enables the production line to automatically adapt to the classification, recognition and quality evaluation of workpieces of different production lines during production.

Description

Technical Field

[0001] The present application relates to the field of artifact assessment, and more specifically, to an online artifact classification and quality assessment method and system based on edge computing. Background Art

[0002] The contents of this section merely provide background information related to the present application and may not constitute prior art.

[0003] Edge computing is a distributed computing model that transfers data processing or service deployment from the cloud to the edge of the network, that is, close to the data source or user. This model aims to reduce network latency, improve data security, save bandwidth costs, and improve user experience and service quality. With the rapid development of technologies such as the Internet of Things, mobile Internet, and artificial intelligence, the amount of data has exploded, and network bandwidth and cloud computing resources are difficult to meet this growing demand. Therefore, edge computing came into being. By deploying computing nodes at the edge of the network, data can be processed nearby, thereby improving the efficiency and real-time performance of data processing.

[0004] In the field of industrial manufacturing, workpiece classification and quality assessment constitute an indispensable part of the production process. Currently, there are advanced technologies, such as the invention with publication number CN111080622A, which proposes a neural network training method and a classification and detection method for workpiece surface defects. This scheme uses deep learning technology to accurately locate the defect position on the surface of the workpiece. In this way, the classification results of defects can not only provide rich information in the detection image, but also be used to evaluate test samples and reveal potential influencing factors in the manufacturing process. However, this scheme faces a limitation in practical applications: it is mainly designed to process a single type of workpiece. When it needs to be deployed on multiple different production lines, it is often necessary to manually replace the processing model that matches the processing object, and lacks the ability to automatically adapt to different production line workpieces for classification, identification and quality assessment. Therefore, there is an urgent need for an online workpiece classification and quality assessment method and system based on edge computing. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of this application is to provide an online workpiece classification and quality assessment method and system based on edge computing, which uses edge nodes to identify the workpiece type through edge computing, and then uses corresponding classification models and assessment models according to the type, so that the production line can automatically adapt to different production line workpieces for classification, identification and quality assessment during production.

[0006] The purpose of this application is achieved through the following technical solutions:

[0007] In the first aspect, an embodiment of the present application proposes an online workpiece classification and quality assessment method based on edge computing, including a camera for real-time monitoring of a production line, an edge node connected to multiple cameras, and a central cloud connected to multiple edge nodes. The method includes:

[0008] Based on the input production parameters, the central cloud sends first configuration information to the corresponding edge node according to the product category of the production parameters, and the central cloud pre-stores an identification algorithm model of the workpiece category, a classification algorithm model of multiple workpieces, and a quality assessment algorithm model of multiple workpieces;

[0009] The edge node obtains the recognition algorithm model from the central cloud according to the first configuration information, and configures the recognition algorithm program;

[0010] In response to the image or video information sent by the camera, the edge node identifies the image or video information through the recognition algorithm model, obtains the corresponding classification algorithm model from the central cloud according to the identified workpiece category, and configures the classification algorithm program;

[0011] Classify defects in image or video information through a classification algorithm program to obtain classification information; package the classification information and workpiece category information and send them to the central cloud;

[0012] The central cloud configures the corresponding quality assessment algorithm program according to the workpiece category information. After the configuration is completed, the quality of the workpiece is assessed through the quality assessment program based on the classification information.

[0013] Furthermore, each camera is connected to at least two edge nodes, each edge node stores information sent by a connected camera, each edge node processes information sent by an authorized camera, and a connection comparison table between edge nodes and cameras is pre-stored in the central cloud. The method also includes:

[0014] When the central cloud does not receive the classification information sent by the configured edge node within the first preset time, the central cloud sends a heartbeat packet to the edge node to detect the connection status;

[0015] If the heartbeat packet does not receive a response within the second preset time, the central cloud obtains the identity information of the lost camera connected to the edge node from the connection comparison table and broadcasts the identity information;

[0016] In response to the broadcasted identity information, each edge node compares the identity information with the connected camera. If the comparison results are different, the corresponding edge node remains silent;

[0017] If the comparison results are the same, the corresponding edge node is defined as an associated edge node, and the associated edge node feeds the result back to the central cloud. The central cloud selects any associated edge node to authorize processing of the image or video information sent by the lost camera.

[0018] Furthermore, after authorizing the processing of the image or video information sent by the lost camera, it also includes:

[0019] The central cloud generates the lost node information based on the timestamp and the classification information last sent by the lost edge node;

[0020] The associated edge node locates the lost node in the storage unit according to the lost node information, and processes the image or video information after the lost node in chronological order.

[0021] Furthermore, the step of edge node identifying the image or video information through the recognition algorithm model specifically includes:

[0022] Cut out a target image of the workpiece to be processed from the image or video information;

[0023] Use OCR to identify the target image by number or text. If the number or text is recognized, match the corresponding product category according to the database;

[0024] If no number or text is recognized, the similarities between the image of the workpiece and a plurality of preset workpiece standard templates are calculated respectively, and the workpiece standard template with the highest similarity and meeting a preset first threshold is used as the product category.

[0025] Furthermore, the step of classifying defects of the image or video information by a classification algorithm program specifically includes:

[0026] The contour information of the workpiece is extracted from the target image using an edge detection algorithm, and compared with a preset standard contour. If the similarity is less than a preset second threshold, it is judged as a contour defect, and the contour information and the standard contour are divided into multiple units. The units at the same position of the contour information and the standard contour are compared, and the units with similarity lower than the third threshold are screened out, and the coordinates corresponding to the units are recorded. If the coordinates are discontinuous with the contour of the workpiece, it is judged as a crack;

[0027] If the similarity is greater than a preset second threshold, the color difference between each pixel in the target image and the surrounding pixels is compared. If it exceeds the preset color difference value, it is determined to be a depression or a spot.

[0028] Furthermore, the step of performing quality assessment on the workpiece through a quality assessment program according to the classification information specifically includes:

[0029] The center cloud selects workpieces that meet the preset rework conditions based on the classification information, which are used to calculate the rework rate; the center cloud selects workpieces that do not meet the preset rework conditions based on the classification information, which are used to calculate the scrap rate; the center cloud calculates the appearance defect rate based on the number of workpieces with concave or spot features in the classification information; the center cloud calculates the casting defect rate based on the number of workpieces with crack features in the classification information; the center cloud calculates the size defect rate based on the contour defects in the classification information;

[0030] According to the product category, weights are assigned to each item in the classification information. The rework rate, scrap rate, appearance defect rate, casting defect rate, dimensional defect rate and the corresponding weights are input into the preset formula to obtain the workpiece quality assessment value.

[0031] Furthermore, the preset formula includes:

[0032]

[0033] in, The rework rate, is the weight of the rework rate, The scrap rate, is the weight of scrap rate, is the appearance failure rate, is the weight of appearance failure rate, is the casting defect rate, is the weight of the casting defect rate, is the size defect rate, is the weight of the size defect rate.

[0034] In the second aspect, an online workpiece classification and quality assessment system based on edge computing proposed in an embodiment of the present application includes a camera for real-time monitoring of a production line, an edge node connected to multiple cameras, and a central cloud connected to multiple edge nodes. The system includes:

[0035] The startup module, based on the input production parameters, the central cloud sends the first configuration information to the corresponding edge node according to the product category of the production parameters, and the central cloud pre-stores the recognition algorithm model of the workpiece category, the classification algorithm model of multiple workpieces and the quality assessment algorithm model of multiple workpieces;

[0036] A configuration module, used for the edge node to obtain the recognition algorithm model from the central cloud according to the first configuration information, and configure the recognition algorithm program;

[0037] Recognition module: In response to the image or video information sent by the camera, the edge node recognizes the image or video information, obtains the corresponding classification algorithm model from the central cloud according to the identified workpiece category, and configures the classification algorithm program;

[0038] The classification module classifies defects in image or video information through a classification algorithm program to obtain classification information; the classification information and workpiece category information are packaged and sent to the central cloud;

[0039] The evaluation module is used by the central cloud to configure the corresponding quality evaluation algorithm program according to the workpiece category information. After the configuration is completed, the quality of the workpiece is evaluated through the quality evaluation program according to the classification information.

[0040] In a third aspect, an electronic device proposed in an embodiment of the present application includes a processor, a storage medium, and a computer program, wherein the computer program is stored in the storage medium, and is characterized in that when the computer program is executed by the processor, the online workpiece classification and quality assessment method as in the first aspect is implemented.

[0041] In a fourth aspect, a computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the online workpiece classification and quality assessment method of the first aspect is implemented.

[0042] In summary, the technical solution of the embodiment of the present application has at least the following advantages and beneficial effects:

[0043] The present invention uses the central cloud as the core of data processing and algorithm storage, and pre-stores recognition algorithm models covering a wide range of workpiece categories, classification algorithm models for multiple workpieces, and corresponding quality assessment algorithm models. When the production line starts to operate, the central cloud will identify the product category currently being produced based on the input production parameters, and send the first configuration information to the corresponding edge node. This step ensures that the system can be flexibly configured for different product categories. Then, after receiving the configuration information, the edge node retrieves the corresponding recognition algorithm model from the central cloud and performs localized algorithm program configuration. Through the configured recognition algorithm model, the edge node can accurately identify the workpiece category, and accordingly obtain the classification algorithm model for the category from the central cloud, and perform local configuration again. After the classification algorithm program is ready, the edge node conducts an in-depth analysis of the image or video information, accurately identifies various defects on the workpiece, and generates detailed classification information. Subsequently, the classification information is packaged and sent back to the central cloud together with the workpiece category information. In the central cloud, the system performs a comprehensive and in-depth quality assessment of the workpiece based on the received workpiece category information. This process not only realizes the automatic classification and identification of workpieces from different production lines, but also ensures the accuracy and efficiency of quality assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A flowchart of an online workpiece classification and quality assessment method based on edge computing provided by the present invention;

[0045] Figure 2A flow chart showing how the recognition algorithm model of the present invention recognizes image or video information;

[0046] Figure 3 This is a flow chart of switching edge nodes in the present invention;

[0047] Figure 4 A structural schematic diagram of an online workpiece classification and quality assessment system based on edge computing provided by the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0049] The following reference Figures 1 to 3 This application is described in further detail.

[0050] The embodiment of the present application proposes an online workpiece classification and quality assessment method based on edge computing, such as Figure 1 As shown, including a camera for real-time monitoring of a production line, an edge node connected to multiple cameras, and a central cloud connected to multiple edge nodes, the method includes:

[0051] S101, based on the input production parameters, the central cloud sends first configuration information to the corresponding edge node according to the product category of the production parameters, and the central cloud pre-stores an algorithm model for identifying the workpiece category, a classification algorithm model for multiple workpieces, and a quality assessment algorithm model for multiple workpieces;

[0052] Specifically, since production parameters such as contract numbers or product numbers will have identifiers for corresponding product categories. Therefore, based on the input production parameters, the central cloud can identify the category of workpieces to be produced. Based on this identification result, the central cloud will send the first configuration information to the corresponding edge node, which instructs the edge node on how to prepare for subsequent processing. It is worth noting that the central cloud pre-stores all the algorithm model libraries needed in the factory, including the identification algorithm model of the workpiece category, the classification algorithm model for multiple workpieces, and the quality assessment algorithm model.

[0053] S102, the edge node obtains the recognition algorithm model from the central cloud according to the first configuration information, and configures the recognition algorithm program. The purpose of this step is to enable the edge node to have preliminary workpiece category recognition capabilities, laying the foundation for subsequent classification and evaluation. The introduction of edge nodes effectively reduces the computing burden of the central cloud and improves the response speed and efficiency of the overall system.

[0054] S103, in response to the image or video information sent by the camera, the edge node identifies the image or video information through the recognition algorithm model, obtains the corresponding classification algorithm model from the central cloud according to the identified workpiece category, and configures the classification algorithm program.

[0055] Specifically, when the camera captures images or video information on the production line, the edge node will immediately process this information using the configured recognition algorithm model to quickly identify the specific category of the workpiece. Based on the recognition results, the edge node will obtain the classification algorithm model that matches the category from the central cloud and configure it accordingly. This dynamic configuration process enables the system to flexibly respond to the processing needs of different categories of workpieces, realizing the intelligent and automated production of the production line.

[0056] Among them, the edge node recognizes the image or video information through the recognition algorithm model, such as Figure 2 As shown, specifically including:

[0057] S201, extracting the target image of the workpiece to be processed from the image or video information. This step aims to accurately extract the workpiece area, remove background interference, and ensure the smooth progress of subsequent recognition work.

[0058] S202, perform number or text recognition on the target image through OCR, and if the number or text is recognized, match the corresponding product category according to the database. OCR is Optical Character Recognition. The introduction of OCR is to improve recognition efficiency, especially for those workpieces with clear markings, so that their categories can be quickly determined, reducing unnecessary calculations and resource consumption.

[0059] S203, if no number or text is recognized, the similarities between the image of the workpiece and a plurality of preset workpiece standard templates are calculated respectively, and the workpiece standard template with the highest similarity and meeting a preset first threshold is used as the product category.

[0060] Specifically, in industrial production, the distinguishing features between workpieces are relatively obvious, so for workpieces without numbers or text information, image matching algorithms are used for identification. In this step, the edge node will calculate the similarity between the image of the workpiece and multiple preset workpiece standard templates. These standard templates are made in advance based on the workpieces and represent the typical features of various types of workpieces. By comparing the similarity between the workpiece image to be processed and the standard template, the system can find the workpiece standard template with the highest similarity and meeting the preset first threshold, thereby determining the product category of the workpiece.

[0061] S104, classify defects of the image or video information through a classification algorithm program to obtain classification information; package the classification information and workpiece category information and send them to the central cloud.

[0062] The step of classifying defects of the image or video information by a classification algorithm program specifically includes:

[0063] Use edge detection algorithms (such as Canny edge detection) to extract the contour information of the workpiece in the target image. The extracted contour information is compared with the preset standard contour, and the integrity of the contour is evaluated by calculating the similarity between the two. If the calculated similarity is less than the preset second threshold, the program will automatically determine that the workpiece has a contour defect. In order to locate the specific location of the defect more accurately, the program will divide the contour information and the standard contour into multiple units and compare these units one by one. By screening out units with a similarity lower than the third threshold and recording the coordinates corresponding to these units, it is convenient to further analyze the continuity of these coordinates with the overall contour of the workpiece. If some coordinates are discontinuous with the contour of the workpiece, the program will determine that there are cracks at these locations.

[0064] On the other hand, if the similarity between the contour information and the standard contour is greater than the preset second threshold, it means that the contour of the workpiece basically meets the standard, and the program will turn to the detection of the surface quality of the workpiece. Compare the color difference between each pixel in the target image and the surrounding pixels. If the color difference between a pixel and its surrounding pixels exceeds the preset color difference value, the program will determine that there is a concave or spot defect at that location.

[0065] S105, the central cloud configures the corresponding quality assessment algorithm program according to the workpiece category information. After the configuration is completed, the quality of the workpiece is assessed through the quality assessment program according to the classification information.

[0066] Among them, the quality assessment program conducts a comprehensive and systematic assessment of the quality of the workpiece based on the detailed classification information provided by the classification algorithm program. This process not only involves the calculation of multiple quality indicators, but also takes into account the specific weight allocation under different product categories to ensure the accuracy and practicality of the assessment results.

[0067] Specifically, as the core of data processing, the central cloud selects workpieces that meet the preset rework conditions based on the classification information. Although these workpieces have certain defects, they can still meet the use requirements after rework and repair. Based on these screening results, the program calculates the rework rate, which reflects the proportion of workpieces that require additional time and resources to repair in the production process. At the same time, the central cloud also selects workpieces that do not meet the preset rework conditions. These workpieces are considered scrap due to serious defects or high repair costs. By counting these scraps, the program calculates the scrap rate, which is directly related to production efficiency and cost control.

[0068] In addition to rework and scrap, the quality assessment program also focuses on the appearance and structural quality of the workpiece. Based on the number of workpieces with concave or spot features in the classification information, the program calculates the appearance defect rate, which reflects the overall level of workpiece surface quality. Based on the number of workpieces with crack features in the classification information, the program calculates the casting defect rate, which reveals possible quality problems in the casting process.

[0069] In addition, contour defects, as another common quality problem, are also included in the scope of quality assessment. The program calculates the dimensional defect rate based on the number of contour defects in the classification information, which reflects the accuracy of workpiece dimensional control.

[0070] After calculating each quality indicator, the quality assessment program also assigns weights to each item in the classification information according to the product category. Different product categories may have different requirements for quality indicators, so the weight allocation can ensure that the assessment results are more in line with actual needs. Finally, the program inputs the rework rate, scrap rate, appearance defect rate, casting defect rate, dimensional defect rate and the corresponding weights into the preset formula, and calculates the workpiece quality assessment value. The preset formula is as follows:

[0071]

[0072] in, The rework rate, is the weight of the rework rate, The scrap rate, is the weight of scrap rate, is the appearance failure rate, is the weight of appearance failure rate, is the casting defect rate, is the weight of the casting defect rate, is the size defect rate, is the weight of the size defect rate.

[0073] In some embodiments of the present invention, based on the above technical solution, this embodiment introduces a new network connection management and data processing mechanism to further enhance the stability and reliability of the system. Including each camera establishing a connection with at least two edge nodes, this design is intended to improve the redundancy and fault tolerance of the system. When an edge node fails to work properly due to a fault or network problem, the camera can still send information through other edge nodes to ensure the continuity and integrity of the data. At the same time, each edge node stores the information sent by the connected camera and has the ability to process the information sent by the authorized camera. This distributed storage and processing method not only reduces the burden on the central cloud, but also improves the response speed and data processing capabilities of the system.

[0074] At the same time, in order to effectively manage these complex connection relationships, the central cloud has a pre-stored connection comparison table with edge nodes and cameras. This comparison table records in detail which edge nodes each camera is connected to, and which camera information each edge node is responsible for processing. This provides a clear connection view for the central cloud, facilitating efficient connection management and information scheduling.

[0075] like Figure 3 As shown, the specific method includes the following steps:

[0076] S301, when the central cloud does not receive the classified information sent by the configured edge node within the first preset time, the central cloud sends a heartbeat packet to the edge node to detect the connection status. Specifically, in the actual operation process, due to network failure, equipment failure or configuration error, the edge node may be unable to send classified information to the central cloud. In order to promptly discover and solve such problems, the embodiment of the present invention introduces a heartbeat packet detection mechanism. When the central cloud does not receive the classified information sent by the configured edge node within the first preset time, it will actively send a heartbeat packet to the edge node to detect its connection status. If the heartbeat packet does not receive a response within the second preset time, the central cloud will consider that the edge node is in a lost state.

[0077] S302, if the heartbeat packet does not receive a response within the second preset time, the central cloud obtains the identity information of the lost camera connected to the edge node from the connection comparison table and broadcasts the identity information.

[0078] S303, in response to the broadcasted identity information, each edge node compares the identity information with the connected camera, and if the comparison results are different, the corresponding edge node remains silent;

[0079] Steps S302 and S303 are intended to let other edge nodes know which cameras may not be able to send information normally due to the loss of connection with the edge node. After receiving the broadcast identity information, each edge node will compare it with the connected camera. If the comparison results are different, it means that the edge node is not connected to the lost camera, so it will remain silent and not participate in subsequent processing.

[0080] S304, if the comparison results are the same, the corresponding edge node is defined as an associated edge node, and the associated edge node feeds back the result to the central cloud. The central cloud selects any associated edge node to authorize processing of the image or video information sent by the lost camera.

[0081] If the comparison result is the same, it means that the edge node is connected to the lost camera, and it will be defined as an associated edge node. After confirming its identity, the associated edge node will feedback the result to the central cloud. After receiving the feedback, the central cloud will select any associated edge node to authorize the processing of the image or video information sent by the lost camera. This step ensures that even if an edge node loses connection, the camera it is responsible for can still continue to send information through other associated edge nodes, thereby ensuring the continuity and stability of the system.

[0082] Furthermore, after authorizing the processing of the image or video information sent by the lost camera, it also includes: the central cloud generates the lost node information based on the timestamp and the classification information sent by the lost edge node for the last time; the associated edge node locates the lost node in the storage unit according to the lost node information, and processes the image or video information after the lost node in chronological order.

[0083] Among them, the application of timestamps and the generation and processing mechanism of lost node information are introduced, aiming to more finely manage the state transition and data continuity of edge nodes, ensuring that the system can efficiently and accurately process image or video information even when the edge nodes are lost.

[0084] Specifically, when the central cloud successfully authorizes a certain associated edge node to process the image or video information sent by the lost camera, the process does not end. In order to ensure the integrity and continuity of the data, the central cloud will generate a detailed lost node information based on the timestamp and the classification information last sent by the lost edge node. This information not only contains the identity of the lost edge node, but also records the timestamp of its last normal operation and the content of the classification information sent. The introduction of the timestamp enables the system to track the change points of the working status of the lost node, providing a key time reference for subsequent data processing.

[0085] Next, the authorized associated edge node will locate the relevant data of the lost node in its storage unit based on the lost node information. Since each edge node stores the information sent by the connected camera, the associated edge node can quickly find the image or video information related to the lost node. More importantly, the associated edge node will process this information in chronological order to ensure the continuity and integrity of the data.

[0086] Through this mechanism, the system can quickly switch to the backup node and resume data processing when the edge node loses connection, thus greatly reducing the system downtime caused by network failure or equipment failure. At the same time, due to the introduction of timestamps and lost node information, the system can accurately track the status changes of each node, providing strong data support for subsequent troubleshooting and performance optimization.

[0087] In addition, this mechanism also improves the flexibility and scalability of the system. As the business scale expands and the number of cameras increases, the system can easily add more edge nodes to share the processing pressure without large-scale transformation of the central cloud. This distributed processing method not only improves the processing capacity of the system, but also reduces the impact of a single node failure on the entire system.

[0088] Based on the same concept, Figure 4 As shown, an embodiment of the present invention provides an online workpiece classification and quality assessment system based on edge computing, including a camera for real-time monitoring of a production line, an edge node connected to multiple cameras, and a central cloud connected to multiple edge nodes. The system includes:

[0089] In the start module 401, based on the input production parameters, the central cloud sends first configuration information to the corresponding edge node according to the product category of the production parameters, and the central cloud pre-stores an algorithm model for identifying the workpiece category, a classification algorithm model for multiple workpieces, and a quality assessment algorithm model for multiple workpieces;

[0090] Configuration module 402, used for the edge node to obtain the recognition algorithm model from the central cloud according to the first configuration information, and configure the recognition algorithm program;

[0091] Identification module 403, in response to the image or video information sent by the camera, the edge node identifies the image or video information, obtains the corresponding classification algorithm model from the central cloud according to the identified workpiece category, and configures the classification algorithm program;

[0092] The classification module 404 classifies defects of the image or video information through a classification algorithm program to obtain classification information; the classification information and workpiece category information are packaged and sent to the central cloud;

[0093] The evaluation module 405 is used for the central cloud to configure the corresponding quality evaluation algorithm program according to the workpiece category information. After the configuration is completed, the quality of the workpiece is evaluated through the quality evaluation program according to the classification information.

[0094] Based on the same concept, an embodiment of the present invention provides an electronic device, which includes a processor, a storage medium and a computer program. The computer program is stored in the storage medium, and when the computer program is executed by the processor, an online workpiece classification and quality assessment method is implemented.

[0095] Based on the same concept, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned online workpiece classification and quality assessment method is implemented.

[0096] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An online workpiece classification and quality assessment method based on edge computing, characterized in that: The method comprises: comprising: a camera for real-time monitoring of a production line, an edge node connected to a plurality of the cameras, and a central cloud connected to the plurality of the edge nodes; Based on the input production parameters, the central cloud sends first configuration information to the corresponding edge node according to the product category of the production parameters, and the central cloud pre-stores an identification algorithm model of the workpiece category, a classification algorithm model of multiple workpieces, and a quality assessment algorithm model of multiple workpieces; The edge node obtains the recognition algorithm model from the central cloud according to the first configuration information, and configures the recognition algorithm program; In response to the image or video information sent by the camera, the edge node identifies the image or video information through the recognition algorithm model, obtains the corresponding classification algorithm model from the central cloud according to the workpiece category obtained by the recognition, and configures the classification algorithm program; wherein, the step of the edge node identifying the image or video information through the recognition algorithm model specifically includes: intercepting the target image of the workpiece to be processed in the image or video information; performing numbering or text recognition on the target image through OCR, and if the number or text is recognized, matching the corresponding product category according to the database; if the number or text is not recognized, respectively calculating the similarity between the image of the workpiece and multiple preset workpiece standard templates, and taking the workpiece standard template with the highest similarity and meeting the preset first threshold as the product category; The image or video information is subjected to defect classification by the classification algorithm program to obtain classification information; the classification information and workpiece category information are packaged and sent to the central cloud; wherein the step of performing defect classification on the image or video information by the classification algorithm program specifically includes: extracting contour information of the workpiece from the target image by using an edge detection algorithm, and comparing it with a preset standard contour; if the similarity is less than a preset second threshold, it is judged as a contour defect; the contour information and the standard contour are divided into a plurality of units; the units at the same position of the contour information and the standard contour are compared, and the units with a similarity lower than a third threshold are screened out; the coordinates corresponding to the units are recorded; if the coordinates are discontinuous with the contour of the workpiece, it is judged as a crack; if the similarity is greater than the preset second threshold, the color difference between each pixel in the target image and the surrounding pixels is compared; if the color difference exceeds the preset color difference value, it is judged as a depression or spot; The central cloud configures the corresponding quality assessment algorithm program according to the workpiece category information. After the configuration is completed, the workpiece is quality assessed according to the classification information through the quality assessment program; wherein, the step of quality assessment of the workpiece according to the classification information through the quality assessment program specifically includes: the central cloud screens out the workpieces that meet the preset rework conditions according to the classification information for calculating the rework rate; screens out the workpieces that do not meet the preset rework conditions according to the classification information for calculating the scrap rate; calculates the appearance unqualified rate according to the number of workpieces containing depressions or spot features in the classification information; calculates the casting defective rate according to the number of workpieces with crack features in the classification information; calculates the size defective rate according to the contour defects in the classification information; assigns weights to each item in the classification information according to the product category, and inputs the rework rate, scrap rate, appearance unqualified rate, casting defective rate, size defective rate and corresponding weights into a preset formula to obtain the workpiece quality assessment value; the preset formula is: in, The rework rate, is the weight of the rework rate, The scrap rate, is the weight of scrap rate, is the appearance failure rate, is the weight of appearance failure rate, is the casting defect rate, is the weight of the casting defect rate, is the size defect rate, is the weight of the size defect rate.

2. The online workpiece classification and quality assessment method according to claim 1, characterized in that: Each of the cameras is connected to at least two edge nodes, each edge node stores information sent by a connected camera, each edge node processes information sent by an authorized camera, and a connection comparison table between edge nodes and cameras is pre-stored in the central cloud. The method further includes: When the central cloud does not receive the classification information sent by the configured edge node within the first preset time, the central cloud sends a heartbeat packet to the edge node to detect the connection status; If the heartbeat packet does not receive a response within the second preset time, the central cloud obtains the identity information of the lost camera connected to the edge node from the connection comparison table and broadcasts the identity information; In response to the broadcasted identity information, each edge node compares the identity information with the connected camera, and if the comparison result is different, the corresponding edge node remains silent; If the comparison results are the same, the corresponding edge node is defined as an associated edge node, and the associated edge node feeds back the result to the central cloud. The central cloud selects any of the associated edge nodes to authorize processing of the image or video information sent by the lost camera.

3. The online workpiece classification and quality assessment method according to claim 2, characterized in that: After the authorization to process the image or video information sent by the lost camera, the method further includes: The central cloud generates lost node information according to the timestamp and the classification information last sent by the lost edge node; The associated edge node locates the lost node in the storage unit according to the lost node information, and processes the image or video information after the lost node in chronological order.

4. An online workpiece classification and quality assessment system based on edge computing, characterized in that: The system includes a camera for real-time monitoring of a production line, an edge node connected to a plurality of the cameras, and a central cloud connected to a plurality of the edge nodes. The system includes: The startup module, based on the input production parameters, the central cloud sends the first configuration information to the corresponding edge node according to the product category of the production parameters, and the central cloud pre-stores the recognition algorithm model of the workpiece category, the classification algorithm model of multiple workpieces and the quality assessment algorithm model of multiple workpieces; A configuration module, used for the edge node to obtain the recognition algorithm model from the central cloud according to the first configuration information, and configure the recognition algorithm program; The recognition module responds to the image or video information sent by the camera, and the edge node recognizes the image or video information, obtains the corresponding classification algorithm model from the central cloud according to the workpiece category obtained by the recognition, and configures the classification algorithm program; wherein, the step of the edge node recognizing the image or video information through the recognition algorithm model specifically includes: intercepting the target image of the workpiece to be processed from the image or video information; performing numbering or text recognition on the target image through OCR, and if the number or text is recognized, matching the corresponding product category according to the database; if the number or text is not recognized, respectively calculating the similarity between the image of the workpiece and multiple preset workpiece standard templates, and taking the workpiece standard template with the highest similarity and meeting the preset first threshold as the product category; A classification module, which classifies defects of the image or video information through the classification algorithm program to obtain classification information; packages the classification information and workpiece category information and sends them to the central cloud; wherein the step of classifying defects of the image or video information through the classification algorithm program specifically includes: extracting contour information of the workpiece from the target image using an edge detection algorithm, and comparing it with a preset standard contour; if the similarity is less than a preset second threshold, it is judged as a contour defect; the contour information and the standard contour are divided into multiple units, the units at the same position of the contour information and the standard contour are compared, and the units with similarity lower than a third threshold are screened out, and the coordinates corresponding to the units are recorded; if the coordinates are discontinuous with the contour of the workpiece, it is judged as a crack; if the similarity is greater than the preset second threshold, the color difference between each pixel in the target image and the surrounding pixels is compared; if it exceeds the preset color difference value, it is judged as a depression or spot; An evaluation module is used for the central cloud to configure a corresponding quality evaluation algorithm program according to the workpiece category information. After the configuration is completed, the workpiece is quality evaluated according to the classification information through the quality evaluation program; wherein the step of quality evaluating the workpiece according to the classification information through the quality evaluation program specifically includes: the central cloud screens out workpieces that meet the preset rework conditions according to the classification information for calculating the rework rate; screens out workpieces that do not meet the preset rework conditions according to the classification information for calculating the scrap rate; calculates the appearance unqualified rate according to the number of workpieces containing depressions or spot features in the classification information; calculates the casting defective rate according to the number of workpieces with crack features in the classification information; calculates the size defective rate according to the contour defects in the classification information; assigns weights to each item in the classification information according to the product category, and inputs the rework rate, scrap rate, appearance unqualified rate, casting defective rate, size defective rate and corresponding weights into a preset formula to obtain a workpiece quality evaluation value; the preset formula is: in, The rework rate, is the weight of the rework rate, The scrap rate, is the weight of scrap rate, is the appearance failure rate, is the weight of appearance failure rate, is the casting defect rate, is the weight of the casting defect rate, is the size defect rate, is the weight of the size defect rate.

5. An electronic device comprising a processor, a storage medium and a computer program, wherein the computer program is stored in the storage medium, wherein: When the computer program is executed by a processor, the online workpiece classification and quality assessment method according to any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the online workpiece classification and quality assessment method according to any one of claims 1 to 3 is implemented.

Images