Intelligent adjusting and testing control method and system for industrial production

Abstract

The invention discloses an intelligent adjusting and testing control method and system for industrial production, and relates to the technical field of adjusting and testing control, and the method comprises the steps: obtaining the production flow information of product spare and accessory parts on an SMT production line, setting a product detection point according to the flow information, and sampling the image data of the to-be-detected product spare and accessory parts; acquiring image training data of each product detection point, and generating product part defect point difference label data according to the image training data; constructing an intelligent adjusting and testing model, performing camera resolution grade labeling and sampling frequency grade labeling on the image training data to obtain grade labeling data, and performing real-time training on the intelligent adjusting and testing model through the image training data and the grade labeling data; the intelligent debugging and testing model monitors and controls the acquisition parameters of the product detection point positions and the acquisition parameters of other product detection point positions in real time according to the image data of the product detection point positions; and the generation probability of product spare part defects is obviously reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of debugging control, and specifically to an intelligent debugging control method and system for industrial production. Background Art

[0002] In traditional large-scale production, there has gradually emerged a demand for customized production. Under the limited production capacity of equipment, it has become a common goal for many enterprises to reasonably arrange the processing of production tasks, efficiently, timely, and dynamically plan production schedules, and perform intelligent production scheduling.

[0003] The comparative document CN117033373A, "An Internet of Things detection data management system and method for industrial production", includes an equipment management module, a data acquisition module, a data transmission module, a data storage module, a data access control module, and a data mining and visualization module; uses the equipment management module to deploy and adjust the parameters of various Internet of Things sensing devices in the industrial production environment; uses the data acquisition module to collect the data detected by various deployed sensors; adopts the methods of consortium blockchain and edge computing to distributively store and manage Internet of Things detection data, improving data management efficiency and storage capacity, and ensuring the credibility and traceability of data; realizes access permission control of Internet of Things detection data through smart contracts to ensure data security. The present invention can achieve efficient management and configuration of sensor devices in the industrial production environment, and ensure the timeliness and accuracy of Internet of Things detection data in industrial production.

[0004] The comparative document CN115963800A, "An intelligent factory scheduling method, system and device based on industrial Internet", generates an initial scheduling sequence corresponding to multiple order information; according to the initial scheduling sequence, executes the production tasks corresponding to the multiple order information respectively, and generates a to-be-scheduled production sequence corresponding to the to-be-scheduled order information; performs perturbation detection on the initial scheduling sequence to trigger re-scheduling of the initial scheduling sequence to obtain a re-scheduled target scheduling sequence; determines the physical workshop corresponding to the target scheduling sequence and the virtual workshop corresponding to the physical workshop, and based on the virtual workshop, performs simulation scheduling on the target scheduling sequence to obtain control parameters corresponding to the target scheduling sequence; generates a control command corresponding to the control parameters, and sends the control command to the controller of the physical workshop, so that the controller controls each production device in the physical workshop to perform corresponding production actions according to the control command.

[0005] In existing industrial production, the cameras for collecting image data of product parts to be detected often do not automatically adjust the resolution of the cameras according to whether the product parts are defective, which directly affects the accuracy of defect detection of product parts. To solve the above technical problems, there is provided an intelligent debugging control method and system for industrial production. Summary of the Invention

[0006] In order to solve the above technical problems, the object of the present invention is to provide an intelligent debugging and control method for industrial production, including the following steps: Step S1: Obtain the production process information of product parts and components on the SMT production line, set product detection points according to the process information, and sample the image data of the product parts and components to be detected; Step S2: Obtain the image training data of each product detection point, and generate defect point difference label data of product parts and components according to the image training data; Step S3: Build an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on the image training data, obtain level annotation data, and perform real-time training on the intelligent debugging model through the image training data and the level annotation data; Step S4: The intelligent debugging model monitors and controls the acquisition parameters of the product detection points and the acquisition parameters of other product detection points in real time according to the image data of the product detection points.

[0007] Further, the process of obtaining the production process information of product parts and components on the SMT production line, setting product detection points according to the process information, and sampling the image data of the product parts and components to be detected includes: Obtain the process flow characteristics of industrial equipment on the SMT production line to which the product parts and components to be detected belong, extract process information according to the process flow characteristics, split the SMT production line to which the product parts and components to be detected belong according to the process information, and divide it into several process flow subsequences; Set product detection points on each process flow subsequence. The product detection points are used to obtain the image data of the product parts and components to be detected that have completed the process flow processing in each process flow subsequence according to preset acquisition parameters and mark the sampling time, set the sampling period, and the acquisition parameters include sampling frequency and camera resolution.

[0008] Further, the process of obtaining the image training data of each product detection point and generating defect point difference label data of product parts and components according to the image training data includes: Obtain the standard product parts and components image data with different camera resolutions and the defective product parts and components image data with different camera resolutions of each process flow subsequence, and use the standard product parts and components image data with different camera resolutions and the defective product parts and components image data with different camera resolutions as image training data; Annotate the defective key points of the defective product parts and components image data with different camera resolutions in the image training data, and perform Gaussian kernel convolution on the defective key points in the image training data to generate defect point difference label data of product parts and components.

[0009] Further, the process of labeling defect key points for the defective product spare part image data with different camera resolutions in the image training data includes: Obtain standard product spare part image data with the same camera resolution as the defective product spare part image data, convert the defective product spare part image data and the standard product spare part image data into grayscale images, subtract the pixel values at corresponding positions of the grayscale image of the defective product spare part image data and the grayscale image of the standard product spare part image data to obtain a difference value, convert the difference value into the pixel value of a binary image, and generate a binary image; Set two pixel position traversal pointers, preset a pixel threshold, and start traversing simultaneously from the first pixel position and the last pixel position in the binary image area; Mark the pixel position area in the binary image where the pixel value is greater than the preset pixel threshold as a defect key point.

[0010] Further, the process of constructing an intelligent debugging model based on a deep convolutional neural network, performing camera resolution level labeling and sampling frequency level labeling on the image training data, obtaining level labeling data, and performing real-time training on the intelligent debugging model through the image training data and the level labeling data includes: Construct an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level labeling and sampling frequency level labeling on the product spare part defect point difference label data in the image training data, and obtain level labeling data; Input the image training data and the level labeling data into the intelligent debugging model for training until the loss function training is stable, save the model parameters, test the energy consumption model through a test set until it meets the preset requirements, and output the intelligent debugging model.

[0011] Further, the process of the intelligent debugging model performing real-time monitoring and control on the acquisition parameters of the product detection point according to the image data of the product detection point includes: Input the image data of the to-be-detected product spare part sampled at the previous moment in the current sampling period of the product detection point into the intelligent debugging model to generate the camera resolution and sampling frequency at the current moment of the product detection point, and compare the camera resolution and sampling frequency with the preset camera resolution and preset sampling frequency of the product detection point for consistency; If the camera resolution and sampling frequency are inconsistent with the preset camera resolution and preset sampling frequency of the product detection point, it proves that the image data of the to-be-detected product spare part at the previous moment is a defective image; The product detection point obtains the image data of the product parts to be detected at the current moment according to the camera resolution and sampling frequency at the current moment, obtains the defect point difference label data of the image data of the product parts to be detected according to the image data of the product parts to be detected at the current moment and the corresponding standard product parts image data, obtains the total difference value of all defect points of the image data according to the defect point difference label data, sets a difference value threshold, and compares the total difference value with the difference value threshold; If the total difference value is greater than the difference value threshold, the product monitoring point is marked as a product unqualified point, and the camera resolution and sampling frequency of other product detection points are adjusted.

[0012] Further, the process of the intelligent debugging model for real-time monitoring and control of the acquisition parameters of other product detection points according to the image data of the product detection points includes: Obtain the assembly structure of each industrial device on the SMT production line, and obtain the assembly sequence and assembly relationship between each process subsequence according to the assembly structure of each industrial device; Construct an assembly directed graph, use each process subsequence as a node of the assembly directed graph, use the assembly sequence and assembly relationship between each process subsequence as the connection relationship between nodes, and use the sampled image data, camera resolution and sampling frequency of the product detection points corresponding to each process subsequence as supplementary nodes of the nodes; Obtain other product detection points that have an assembly relationship with the product unqualified point according to the assembly relationship of each node in the assembly directed graph, and obtain other product detection points whose assembly sequence is before the product unqualified point among the other product detection points according to the assembly sequence of each node in the assembly directed graph, and convert the camera resolution and sampling frequency of the other product detection points into the camera resolution and sampling frequency of the product unqualified point.

[0013] An intelligent debugging control system for industrial production includes a monitoring center, and the monitoring center is communicatively connected with a data acquisition module, a data processing module, an intelligent debugging model construction module and an intelligent control module; The data acquisition module is used to obtain the production process information of product parts on the SMT production line, set product detection points according to the process information, and sample the image data of the product parts to be detected; The data processing module is used to obtain the image training data of each product detection point, and generate product part defect point difference label data according to the image training data; The intelligent debugging model construction module is used to construct an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on image training data to obtain level annotation data, and perform real-time training on the intelligent debugging model through the image training data and the level annotation data; The intelligent control module is used to perform real-time monitoring and control on the acquisition parameters of the product detection points and the acquisition parameters of other product detection points through the intelligent debugging model.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. The intelligent debugging model automatically generates a resolution level according to the image data sampling result of the product detection point, significantly improving the accuracy of defect detection of the product parts to be detected at the product detection point.

[0015] 2. The intelligent debugging model converts the camera resolution and sampling frequency of other product detection points whose assembly sequence is before the product unqualified point. That is, the reason for the defect of the product parts at the product unqualified point may be related to this point, and may also be caused by the tiny defects that may be generated and accumulated at other product detection points whose assembly sequence is before the product unqualified point. By increasing the camera resolution and sampling frequency of other product detection points whose assembly sequence is before the product unqualified point, the tiny defects generated by the product parts due to industrial equipment reasons can be detected in advance, so as to perform regulation and maintenance operations on the relevant industrial equipment, significantly reducing the probability of product part defects. Description of the Drawings

[0016] Figure 1 It is a schematic diagram of an intelligent debugging control method for industrial production according to an embodiment of the present application.

[0017] Figure 2 It is a schematic diagram of an intelligent debugging control system for industrial production according to an embodiment of the present application. Detailed Embodiments

[0018] The following clearly and completely describes the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0019] As Figure 1 shown, an intelligent debugging control method for industrial production includes the following steps: Step S1: Obtain the production process information of product spare parts on the SMT production line, set product detection points according to the process information, and sample the image data of the product spare parts to be detected. Step S2: Obtain the image training data of each product detection point, and generate product spare part defect point difference label data according to the image training data. Step S3: Build an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on the image training data, obtain level annotation data, and perform real-time training on the intelligent debugging model through the image training data and the level annotation data. Step S4: The intelligent debugging model monitors and controls the acquisition parameters of the product detection point and the acquisition parameters of other product detection points in real time according to the image data of the product detection point.

[0020] It should be further noted that in the specific implementation process, the process of obtaining the production process information of product spare parts on the SMT production line, setting product detection points according to the process information, and sampling the image data of the product spare parts to be detected includes: Obtain the process flow characteristics of the industrial equipment on the SMT production line to which the product spare parts to be detected belong, extract the process information according to the process flow characteristics, and split the SMT production line to which the product spare parts to be detected belong according to the process information into several process flow subsequences. Set product detection points on each process flow subsequence. The product detection points are used to obtain the image data of the product spare parts to be detected that have completed the process flow processing in each process flow subsequence according to the preset acquisition parameters and mark the sampling time, and set the sampling period. The acquisition parameters include the sampling frequency and the camera resolution.

[0021] It should be further noted that in the specific implementation process, the process of obtaining the image training data of each product detection point and generating product spare part defect point difference label data according to the image training data includes: Obtain the standard product spare part image data with different camera resolutions and the product spare part image data with defects with different camera resolutions of each process flow subsequence, and use the standard product spare part image data with different camera resolutions and the product spare part image data with defects with different camera resolutions as the image training data. Annotate the defect key points of the product spare part image data with defects with different camera resolutions in the image training data, and perform Gaussian kernel convolution on the defect key points in the image training data to generate product spare part defect point difference label data.

[0022] It should be further noted that in the specific implementation process, the process of annotating defect key points for the defective product spare part image data with different camera resolutions in the image training data includes: Obtain standard product spare part image data consistent with the camera resolution of the defective product spare part image data, convert the defective product spare part image data and the standard product spare part image data into grayscale images, subtract the pixel values at corresponding positions of the grayscale image of the defective product spare part image data and the grayscale image of the standard product spare part image data to obtain a difference value, convert the difference value into the pixel value of a binary image, and generate a binary image; Set two pixel position traversal pointers, preset a pixel threshold, and start traversing simultaneously from the first pixel position and the last pixel position in the binary image area; Mark the pixel position area in the binary image where the pixel value is greater than the preset pixel threshold as the defect key point.

[0023] It should be further noted that in the specific implementation process, based on a deep convolutional neural network, an intelligent debugging model is constructed to perform camera resolution level annotation and sampling frequency level annotation on the image training data, obtain level annotation data, and the process of training the intelligent debugging model in real time through the image training data and the level annotation data includes: Based on a deep convolutional neural network, an intelligent debugging model is constructed, and the product spare part defect point difference label data in the image training data is subjected to camera resolution level annotation and sampling frequency level annotation to obtain level annotation data; Input the image training data and the level annotation data into the intelligent debugging model for training until the loss function training is stable, save the model parameters, test the energy consumption model through a test set until it meets the preset requirements, and output the intelligent debugging model.

[0024] It should be further noted that in the specific implementation process, for example, the image data of the product has no defect, the camera resolution level is level 0, and the corresponding resolution is 1440*900; level 1: the corresponding resolution is 1680*1050, level 2: the corresponding resolution is 1920*1200; the intelligent debugging model automatically generates a resolution level according to the image data sampling result of the product detection point, significantly improving the accuracy of defect detection of the product spare parts to be detected at the product detection point.

[0025] It should be further noted that in the specific implementation process, the calculation formula of the loss function is

[0026] where Represents the value at the position (i, j) of the defect key point. The higher the value, the more likely it is to be a defect key point; Indicates the pixel value at the human defect key point (i, j), and N represents the number of defect key points; 、 Represents hyperparameters.

[0027] It should be further noted that in the specific implementation process, the process of the intelligent debugging model for real-time monitoring and control of the acquisition parameters of the product detection point according to the image data of the product detection point includes: Input the image data of the spare parts of the product to be detected sampled at the previous moment in the current sampling period of the product detection point into the intelligent debugging model to generate the camera resolution and sampling frequency at the current moment of the product detection point, and compare the camera resolution and sampling frequency with the preset camera resolution and preset sampling frequency of the product detection point for consistency; If the camera resolution and sampling frequency are inconsistent with the preset camera resolution and preset sampling frequency of the product detection point, it proves that the image data at the previous moment of the spare parts of the product to be detected is a defective image; The product detection point obtains the image data of the spare parts of the product to be detected at the current moment according to the camera resolution and sampling frequency at the current moment, obtains the defect point difference label data of the image data of the spare parts of the product to be detected according to the image data of the spare parts of the product to be detected at the current moment and the corresponding standard product spare parts image data, obtains the total difference value of all defect points of the image data according to the defect point difference label data, sets a difference value threshold, and compares the total difference value with the difference value threshold; If the total difference value is greater than the difference value threshold, mark the product monitoring point as an unqualified product point, and then adjust the camera resolution and sampling frequency of other product detection points.

[0028] It should be further noted that in the specific implementation process, the process of the intelligent debugging model for real-time monitoring and control of the acquisition parameters of other product detection points according to the image data of the product detection point includes: Obtain the assembly structures of each industrial device on the SMT production line, and obtain the assembly sequence and assembly relationship between each process subsequence according to the assembly structures of each industrial device; Construct an assembly directed graph, use each process subsequence as a node of the assembly directed graph, use the assembly sequence and assembly relationship between each process subsequence as the connection relationship between nodes, and use the sampled image data, camera resolution and sampling frequency of the product detection point corresponding to each process subsequence as supplementary nodes of the nodes; Obtain other product detection points that have an assembly relationship with the unqualified product points according to the assembly relationships of each node in the assembly directed graph, and obtain other product detection points among the above-mentioned other product detection points whose assembly order is before the unqualified product point according to the assembly order of each node in the assembly directed graph, and convert the camera resolution and sampling frequency of the above-mentioned other product detection points into the camera resolution and sampling frequency of the unqualified product point.

[0029] It should be further noted that in the specific implementation process, the intelligent debugging model converts the camera resolution and sampling frequency of other product detection points whose assembly order is before the unqualified product point. That is, in addition to being related to this point, the reason for the defect of the product parts at the unqualified product point may also be caused by the tiny defects that may be generated and accumulated at other product detection points whose assembly order is before the unqualified product point. By increasing the camera resolution and sampling frequency of other product detection points whose assembly order is before the unqualified product point, the tiny defects generated by the product parts due to industrial equipment reasons can be detected in advance, so as to perform regulation and maintenance operations on the relevant industrial equipment, and significantly reduce the probability of product part defects.

[0030] As Figure 2 shown, an intelligent debugging control system for industrial production includes a monitoring center, and the monitoring center is communicatively connected to a data acquisition module, a data processing module, an intelligent debugging model construction module, and an intelligent control module; The data acquisition module is used to obtain the production process information of product parts on the SMT production line, set product detection points according to the process information, and sample the image data of the product parts to be detected; The data processing module is used to obtain the image training data of each product detection point, and generate product part defect point difference label data according to the image training data; The intelligent debugging model construction module is used to construct an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on the image training data, obtain level annotation data, and perform real-time training on the intelligent debugging model through the image training data and the level annotation data; The intelligent control module is used to perform real-time monitoring and control on the acquisition parameters of the product detection points and the acquisition parameters of other product detection points through the intelligent debugging model.

[0031] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent debugging control method for industrial production, characterized in that, It includes the following steps: Step S1: Obtain the production process information of product spare parts on the SMT production line, set product detection points according to the process information, and sample the image data of the product spare parts to be detected; Step S2: Obtain the image training data of each product detection point, and generate product spare part defect point difference label data according to the image training data; Step S3: Build an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on the image training data, obtain level annotation data, and perform real-time training on the intelligent debugging model through the image training data and the level annotation data; Step S4: The intelligent debugging model monitors and controls the acquisition parameters of the product detection points and the acquisition parameters of other product detection points in real time according to the image data of the product detection points.

2. The intelligent debugging control method for industrial production according to claim 1, wherein The process of obtaining the production process information of product spare parts on the SMT production line, setting product detection points according to the process information, and sampling the image data of the product spare parts to be detected includes: Obtain the process flow characteristics of the industrial equipment on the SMT production line to which the product spare parts to be detected belong, extract process information according to the process flow characteristics, split the SMT production line to which the product spare parts to be detected belong according to the process information, and divide it into several process flow subsequences; Set product detection points on each process flow subsequence. The product detection points are used to obtain the image data of the product spare parts to be detected that have completed the process flow processing in each process flow subsequence according to the preset acquisition parameters and mark the sampling time, set the sampling period, and the acquisition parameters include the sampling frequency and the camera resolution.

3. An intelligent debugging control method for industrial production according to claim 2, characterized in that The process of obtaining the image training data of each product detection point and generating product spare part defect point difference label data according to the image training data includes: Obtain the standard product spare part image data with different camera resolutions and the defective product spare part image data with different camera resolutions of each process flow subsequence, and use the standard product spare part image data with different camera resolutions and the defective product spare part image data with different camera resolutions as image training data; Label the defect key points of the defective product spare part image data with different camera resolutions in the image training data, and perform Gaussian kernel convolution on the defect key points in the image training data to generate product spare part defect point difference label data.

4. An intelligent debugging and control method for industrial production according to claim 3, characterized in that The process of labeling the defect key points of the defective product spare part image data with different camera resolutions in the image training data includes: Obtain the standard product spare part image data with the same camera resolution as the defective product spare part image data, convert the defective product spare part image data and the standard product spare part image data into grayscale images, subtract the pixel values at the corresponding positions of the grayscale image of the defective product spare part image data and the grayscale image of the standard product spare part image data to obtain a difference value, convert the difference value into the pixel value of a binary image, and generate a binary image. Set two pixel position traversal pointers, preset a pixel threshold, and start traversing simultaneously from the first pixel position and the last pixel position of the binary image area; Mark the pixel position area in the binary image where the pixel value is greater than the preset pixel threshold as defect key points.

5. The intelligent debugging control method for industrial production according to claim 4, characterized in that, The process of constructing an intelligent debugging model based on a deep convolutional neural network, performing camera resolution level annotation and sampling frequency level annotation on image training data to obtain level annotation data, and performing real-time training on the intelligent debugging model through the image training data and the level annotation data includes: Construct an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on the product spare part defect point difference label data in the image training data to obtain level annotation data; Input the image training data and the level annotation data into the intelligent debugging model for training until the loss function training is stable, save the model parameters, test the energy consumption model through a test set until it meets the preset requirements, and output the intelligent debugging model.

6. The intelligent debugging control method for industrial production according to claim 5, wherein, The process of the intelligent debugging model performing real-time monitoring and control on the acquisition parameters of the product detection point according to the image data of the product detection point includes: Input the image data of the product spare parts to be detected sampled at the previous moment in the current sampling period of the product detection point into the intelligent debugging model to generate the camera resolution and sampling frequency at the current moment of the product detection point, and compare the camera resolution and sampling frequency with the preset camera resolution and preset sampling frequency of the product detection point; If the camera resolution and sampling frequency are inconsistent with the preset camera resolution and preset sampling frequency of the product detection point, it proves that the image data of the product spare parts to be detected at the previous moment is a defective image; The product detection point obtains the image data of the product spare parts to be detected at the current moment according to the camera resolution and sampling frequency at the current moment, obtains the defect point difference label data of the image data of the product spare parts to be detected according to the image data of the product spare parts to be detected at the current moment and the corresponding standard product spare part image data, obtains the sum of the difference values of all defect points of the image data according to the defect point difference label data, sets a difference value threshold, and compares the sum of the difference values with the difference value threshold; If the sum of the difference values is greater than the difference value threshold, mark the product monitoring point as a product unqualified point, and then adjust the camera resolution and sampling frequency of other product detection points.

7. An intelligent debugging control method for industrial production according to claim 6, characterized in that The process of the intelligent debugging model performing real-time monitoring and control on the acquisition parameters of other product detection points according to the image data of the product detection point includes: Obtain the assembly structures of various industrial equipment on the SMT production line, and obtain the assembly sequence and assembly relationship between each process subsequence according to the assembly structures of the industrial equipment; Construct an assembly directed graph, taking each process subsequence as a node of the assembly directed graph, taking the assembly sequence and assembly relationship between each process subsequence as the connection relationship between nodes, and taking the sampled image data of the product detection points corresponding to each process subsequence, as well as the camera resolution and sampling frequency as supplementary nodes of the nodes; Obtain other product detection points that have an assembly relationship with the product unqualified points according to the assembly relationship of each node in the assembly directed graph, and obtain other product detection points whose assembly sequence is before the product unqualified point among the other product detection points according to the assembly sequence of each node in the assembly directed graph, and convert the camera resolution and sampling frequency of the other product detection points into the camera resolution and sampling frequency of the product unqualified point.

8. An intelligent debugging and control system for industrial production, specifically applied to an intelligent debugging and control method for industrial production according to any one of claims 1 to 7, including a monitoring center, characterized in that, The monitoring center is communicatively connected to a data acquisition module, a data processing module, an intelligent debugging model construction module, and an intelligent control module; The data acquisition module is used to obtain the product parts production process information on the SMT production line, set product detection points according to the process information, and sample the image data of the product parts to be detected; The data processing module is used to obtain the image training data of each product detection point, and generate product parts defect point difference label data according to the image training data; The intelligent debugging model construction module is used to construct an intelligent debugging model based on a deep convolutional neural network, perform camera resolution level annotation and sampling frequency level annotation on the image training data, obtain level annotation data, and perform real-time training on the intelligent debugging model through the image training data and the level annotation data; The intelligent control module is used to perform real-time monitoring and control on the acquisition parameters of the product detection points and the acquisition parameters of other product detection points through the intelligent debugging model.

Images