A quality inspection device for large-size nickel sheets

Abstract

The present invention provides a quality inspection device for large-sized nickel sheets, comprising: a support frame; a weighing mechanism, which is arranged on the support frame and is used to obtain quality information of the sheet to be inspected; a visual inspection mechanism; wherein a data acquisition module is arranged on the support frame, and the data acquisition module is located above the weighing mechanism, and is used to collect surface images of the sheet to be inspected; an image processing module is used to perform size recognition processing on the surface image, generate size information of the sheet to be inspected, perform defect detection processing on the surface image, and generate defect detection results; a control module is used to perform surface density calculation processing on the corresponding sheet to be inspected according to the quality information and size information, generate surface density information, and determine that when the defect detection result meets the preset qualified condition and the surface density information is within the preset density range, the corresponding sheet to be inspected is inspected and qualified. The quality inspection device for large-sized nickel sheets provided by the present invention can improve the quality inspection efficiency of nickel sheets.

Description

Technical Field

[0001] The present invention relates to the technical field of large-size sheet material quality inspection, and in particular to a large-size nickel sheet material quality inspection device, method and system. Background Art

[0002] In recent years, with the rapid development of the electronics industry and new energy industry, the demand for large-sized and thin nickel sheets has continued to increase. Such nickel sheets are mainly used in batteries, electronic components and other high-tech products, and have attracted widespread attention due to their good conductivity, corrosion resistance and mechanical strength.

[0003] However, due to the large size and thin thickness of this type of nickel sheet, there are many challenges in the production, processing and application process, especially in quality inspection (quality inspection). At present, the quality inspection of large-sized nickel sheets is mainly surface defect inspection, and the inspection method mainly relies on manual measurement and optical microscope inspection. The existing large-sized nickel sheet inspection method has the problems of low efficiency and insufficient precision. Therefore, there is room for improvement. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a quality inspection device for large-sized nickel sheets, thereby improving the problem of low quality inspection efficiency of existing large-sized nickel sheets.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The present invention provides a quality inspection device for large-size nickel sheets, comprising:

[0007] Support frame;

[0008] A weighing mechanism, arranged on the support frame, for obtaining quality information of the sheet to be tested;

[0009] A visual inspection mechanism, including a data acquisition module, an image processing module and a control module;

[0010] Wherein, the data acquisition module is arranged on the support frame, and the data acquisition module is located above the weighing mechanism, and is used to collect the surface image of the sheet to be detected on the weighing mechanism;

[0011] The image processing module is used to perform size recognition processing on the surface image to generate size information of the sheet to be detected;

[0012] The image processing module is also used to perform defect detection processing on the surface image to generate a defect detection result;

[0013] The control module is used to perform surface density calculation processing on the corresponding sheet material to be detected according to the quality information and the size information to generate surface density information;

[0014] The control module is also used to determine that when the defect detection result meets a preset qualification condition and the surface density information is within a preset density range, the corresponding sheet material to be detected is qualified.

[0015] In one embodiment of the present invention, the support frame includes at least one base, a vertical rod and a horizontal rod, one end of the vertical rod is connected to the base, and one end of the horizontal rod is connected to an end of the vertical rod away from the base;

[0016] The data acquisition module is arranged on the cross bar.

[0017] In one embodiment of the present invention, a transport assembly is provided on one side of the support frame, and the transport assembly includes:

[0018] A transport body, arranged on one side of the base;

[0019] A lifting drive mechanism is arranged on the transport body;

[0020] A lifting bracket is drivingly connected to the lifting drive mechanism, and the lifting bracket is slidably connected to the transport body through the lifting drive mechanism, so as to transport the sheet to be detected so as to load the sheet to be detected onto the weighing mechanism.

[0021] In one embodiment of the present invention, a power wheel mechanism is disposed on the base, the weighing mechanism is disposed on the power wheel mechanism, and the weighing mechanism includes:

[0022] A weighing sensor, fixedly arranged on the housing of the power wheel mechanism;

[0023] A tray, arranged on the weighing sensor, for carrying the sheet to be tested on the lifting bracket;

[0024] A display, fixedly disposed at one end of the tray;

[0025] Wherein, the weighing sensor is used to obtain the quality information of the sheet to be detected in the tray and display it on the display.

[0026] In one embodiment of the present invention, a sliding component is provided on the crossbar, the data acquisition module is fixedly provided on the sliding component, and the data acquisition module slides along the crossbar with the cooperation of the sliding component to acquire picture information of the display.

[0027] In one embodiment of the present invention, the sliding assembly comprises:

[0028] An electric slide rail is fixedly arranged on the cross bar;

[0029] A slider is slidably arranged on the electric slide rail, the travel direction of the slider is arranged parallel to the long side of the cross bar, and the slider is fixedly connected to the data acquisition module;

[0030] Wherein, under the control of the control module, the electric slide rail drives the slider to drive the data acquisition module to slide to a preset position to acquire the picture information of the display.

[0031] In one embodiment of the present invention, the image processing module is fixedly mounted on the vertical rod, and the image processing module is also used to perform information recognition processing on the image information of the display to generate quality information of the corresponding sheet to be detected in the tray.

[0032] In one embodiment of the present invention, the control module is disposed on the vertical rod, and the control module includes:

[0033] A control system cabinet is fixedly mounted on the vertical rod;

[0034] An operation screen is fixedly arranged on the vertical rod;

[0035] Among them, the control system cabinet is used to determine that when the surface density information is not within the preset density range or the defect detection result does not meet the preset qualification conditions, a warning message and a re-inspection prompt message are generated and displayed on the operation screen.

[0036] In one embodiment of the present invention, a transport component is disposed on one side of the support frame, and the transport component is a forklift lifting mechanism.

[0037] In one embodiment of the present invention, the support frame is U-shaped.

[0038] In one embodiment of the present invention, the image processing module is a high-performance image processing computer.

[0039] As described above, the present invention provides a quality inspection device, method and system for large-size nickel sheets. A weighing mechanism is set to obtain quality information of the sheet to be inspected. A data acquisition module collects a surface image of the sheet to be inspected on the weighing mechanism, and an image processing module is set to perform size recognition processing on the surface image to generate size information of the sheet to be inspected. The control module calculates the surface density of the sheet to be inspected according to the quality information and the size information to generate the surface density information. The image processing module also performs defect detection processing on the surface image to generate a defect detection result. Finally, the control module determines that when the defect detection result meets the preset qualified conditions and the surface density information is within the preset density range, the corresponding sheet to be inspected is qualified, thereby improving the quality inspection accuracy and quality inspection efficiency of the sheet to be inspected. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0041] Figure 1 It is a schematic diagram of the partial structure of a quality inspection device in one embodiment of the present invention;

[0042] Figure 2 is another partial structural schematic diagram of a quality inspection device in one embodiment of the present invention;

[0043] Figure 3 It is a schematic diagram of the working principle structure of a quality inspection device in one embodiment of the present invention;

[0044] Figure 4 A sample library formed by surface images of a plurality of sheets to be inspected in one embodiment of the present invention;

[0045] Figure 5 For Figure 4 The images obtained after preprocessing some images in the sample library;

[0046] Figure 6 For Figure 5 The defect detection result image generated after performing defect detection processing on the image in;

[0047] Figure 7 It is a schematic flow chart of a quality inspection method for large-size sheets of the present invention;

[0048] Figure 8 It is a schematic diagram of a quality inspection system for large-size sheets of the present invention;

[0049] Fig. 9 for Figure 8 Model structure diagram of convolutional neural network module 40;

[0050] Fig.10 Schematic diagram of a pooled feature map generated by performing maximum pooling on a convolution output feature map;

[0051] Fig.11 Schematic diagram of the corresponding relationship between the fully connected layer and the activation function layer in the output layer;

[0052] Fig.12 Schematic diagram of the process of processing the pooled feature map by the fully connected layer and the activation function layer in the output layer.

[0053] Component number description:

[0054] 10. Quality inspection equipment; 20. Information acquisition module; 30. Preprocessing module; 40. Convolutional neural network module; 50. Result output module;

[0055] 100, support frame; 110, transport assembly; 111, transport body; 112, lifting drive mechanism; 113, lifting bracket;

[0056] 121. base; 122. vertical bar; 123. horizontal bar;

[0057] 130. Power wheel mechanism; 131. Roller; 132. Power body;

[0058] 140. Sliding assembly; 141. Electric slide rail; 142. Sliding block;

[0059] 200, weighing mechanism; 210, weighing sensor; 220, tray; 230, display; 240, bracket;

[0060] 300, visual inspection mechanism; 310, data acquisition module; 320, image processing module;

[0061] 330, control module; 331, control system cabinet; 332, operation screen;

[0062] 340. Data monitoring module; 350. Mobile monitoring terminal. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] See also Figure 1 , Figure 2 and Figure 3 The present invention provides a quality inspection device for large-size sheets, which can be used to integrate weighing and detection of large-size sheets, realize real-time data collection, analysis and closed-loop feedback, and thus improve the quality inspection and analysis efficiency of large-size sheets.

[0065] See also Figure 1 and Figure 2 The quality inspection device 10 may include but is not limited to a support frame 100, a weighing mechanism 200 and a visual inspection mechanism 300. The weighing mechanism 200 may be arranged on the support frame 100 to obtain quality information of the sheet to be inspected. The visual inspection mechanism 300 may include but is not limited to a data acquisition module 310, an image processing module 320 and a control module 330.

[0066] The data acquisition module 310 may be disposed on the support frame 100 , and the data acquisition module 310 may be located above the weighing mechanism 200 , and is used to collect a surface image of the sheet to be inspected on the weighing mechanism 200 .

[0067] The image processing module 320 can be used to perform size recognition processing on the surface image to generate size information of the sheet to be detected. The image processing module 320 is also used to perform defect detection processing on the surface image to generate defect detection results.

[0068] The control module 330 can be used to calculate the surface density of the corresponding sheet to be detected according to the quality information and size information to generate surface density information. The control module 330 is also used to determine that when the defect detection result meets the preset qualified condition and the surface density information is within the preset density range, the corresponding sheet to be detected is qualified.

[0069] See also Figure 1 , Figure 2 and Figure 3 , further, a transport assembly 110 may be provided on one side of the support frame 100, and the support frame 100 may include at least one base 121, a vertical rod 122 and a horizontal rod 123. Among them, one end of the vertical rod 122 may be fixedly connected to the base 121, one end of the horizontal rod 123 may be fixedly connected to the end of the vertical rod 122 away from the base 121, and the central axis of the vertical rod 122 is perpendicular to the central axis of the base 121 and the central axis of the horizontal rod 123. That is, the shape of the support frame 100 may be U-shaped, and the U-shaped support frame 100 may be arranged to be rotated 90° counterclockwise. However, it is not limited thereto, and the support frame 100 may also be other shapes.

[0070] Furthermore, the transport assembly 110 may be a combined forklift lifting mechanism to control the overall direction of the support frame 100. However, it is not limited thereto, and the transport assembly 110 may also be other lifting mechanisms as long as it can control the overall direction of the support frame 100.

[0071] The transport assembly 110 may include, but is not limited to, a transport body 111, a lifting drive mechanism 112, and a lifting bracket 113. The transport body 111 may be disposed on one side of the base 121, and the transport body 111 may be disposed in an L shape, and a plurality of moving wheels may be disposed at the bottom end of the transport body 111, so that when the transport assembly 110 is not matched with the support frame 100, the transport assembly 110 can act as a forklift and move. A handle may also be disposed on the vertical rod of the transport body 111 to facilitate towing the transport body 111.

[0072] See also Figure 1 The lifting drive mechanism 112 can be disposed on the transport body 111, and the lifting drive mechanism 112 can be a drive motor or other types of drive mechanisms. The lifting bracket 113 can be connected to the lifting drive mechanism 112 by transmission, and the lifting bracket 113 can be slidably connected to the transport body 111 through the lifting drive mechanism 112 to transport the sheet to be detected, so as to load the sheet to be detected onto the weighing mechanism 200.

[0073] Specifically, a slide groove may be provided on the vertical rod of the transport body 111, and one end of the lifting bracket 113 may be provided in the slide groove and connected to the output shaft of the lifting drive mechanism 112, so as to drive the sheet to be detected to move up and down under the action of the lifting drive mechanism 112, so as to realize the upper and lower sheets of the sheet to be detected. In addition, the central axis of the lifting bracket 113 may be perpendicular to the central axis of the vertical rod on the transport body 111.

[0074] See also Figure 2 and Figure 3 , a power wheel mechanism 130 may be provided on the base 121 to control and drive the support frame 100. The power wheel mechanism 130 may include but is not limited to a plurality of rollers 131 and a power body 132. The roller 131 may be rotatably provided on the base 121, the power body 132 may be fixedly provided on the base 121, and the power body 132 may be transmission-connected with the roller 131 to drive the roller 131, thereby controlling the rotation of the roller 131. The power body 132 may be an electric motor, a hydraulic motor, or a servo motor. However, it is not limited thereto, and the power body 132 may also be other types of driving mechanisms.

[0075] See also Figure 2 and Figure 3There are many ways to connect the power body 132 and the roller 131, and the specific ways depend on the type of the power body 132. For example, see Figure 1 , four rollers 131 may be provided, symmetrically arranged on both sides of the base 121. The power body 132 and the rollers 131 are directly driven, that is, the output shaft of the power body 132 is directly connected to the rollers 131. This method is simple and direct in structure and has high transmission efficiency, but requires that the position layout of the power body 132 must be aligned with the rollers 131. However, it is not limited to this, and the power body 132 and the rollers 131 can also be belt driven, chain driven, or gear driven.

[0076] See also Figure 2 and Figure 3 , the weighing mechanism 200 can be arranged on the power body 132, and the weighing mechanism 200 can include but is not limited to a weighing sensor 210, a tray 220 and a display 230. Among them, the weighing sensor 210 can be fixedly arranged on the shell of the power body 132, and the weighing sensor 210 can be connected to the control module 330 for communication to upload the quality information to the control module 330. The weighing sensor 210 is also used to obtain the quality information of the sheet to be detected in the tray 220 and display it on the display 230. The weighing sensor 210 can be a thousandths piezoelectric sensor or a thousandths capacitive sensor. However, it is not limited to this, and the weighing sensor 210 can also be other types of sensors.

[0077] See also Figure 2 and Figure 3 Furthermore, a tray 220 can be disposed on the weighing sensor 210 to carry the sheet to be detected on the lifting bracket 113. The size of the tray 220 is larger than the size of the sheet to be detected so as to carry the sheet to be detected. The material of the tray 220 has high strength and good stability to bear heavy objects without deformation, and the tray 220 can be a lightweight detection tray to improve the quality detection accuracy of the sheet to be detected.

[0078] The tray 220 may be directly placed on the weighing sensor 210 and directly contact the weighing sensor 210 . However, the present invention is not limited thereto. To improve the stability of the sheet to be detected in the tray 220 , a bracket 240 may be provided at the bottom of the tray 220 .

[0079] For example, see Figure 3 The vertical cross-section of the bracket 240 is a trapezoid, the shorter bottom side of the bracket 240 is fixedly connected to the weighing sensor 210, and the longer bottom side of the bracket 240 is fixedly connected to the tray 220 to improve the stability of the tray 220 carrying the sheet to be detected.

[0080] The display 230 can be fixedly arranged at one end of the tray 220, and the display 230 can be connected to the weighing sensor 210 for communication, so as to display the quality information of the sheet to be detected in the tray 220. The display 230 can be a high-resolution LCD display or LED display, and the communication method between the display 230 and the weighing sensor 210 can be wired communication or wireless communication.

[0081] See also Figure 3 and Figure 4 The data acquisition module 310 can be arranged on the crossbar 123, and the support frame 100 can also be provided with a sliding assembly 140, and the sliding assembly 140 is arranged on the crossbar 123. The data acquisition module 310 can be fixedly arranged on the sliding assembly 140, and the data acquisition module 310 can slide along the crossbar 123 with the cooperation of the sliding assembly 140 to collect image information of the display 230. The data acquisition module 310 is also used to collect surface images of multiple sheets to be detected on the weighing mechanism 200 to form a Figure 4 The sample library shown.

[0082] Furthermore, the sliding assembly 140 may include but is not limited to an electric slide rail 141 and a slider 142. The electric slide rail 141 may be fixedly disposed on the cross bar 123. The slider 142 may be slidably disposed on the electric slide rail 141, the travel direction of the slider 142 is parallel to the long side of the cross bar 123, and the slider 142 is fixedly connected to the data acquisition module 310. Under the control of the control module 330, the electric slide rail 141 drives the slider 142 to drive the data acquisition module 310 to slide to a preset position to collect image information of the display 230. Specifically, the preset position is a position on the cross bar 123 corresponding to the top of the display 230.

[0083] See also Figure 2 and Figure 3, the image processing module 320 can be fixedly arranged on the vertical rod 122, and the image processing module 320 can be respectively connected to the data acquisition module 310 and the control module 330 for communication, so as to perform information recognition processing on the image information of the display 230, and generate the quality information of the corresponding sheet to be detected in the tray 220. The image processing module 320 is also used to perform defect detection processing on the surface image and generate defect detection results. Among them, the image processing module 320 can be a high-performance image processing computer, and a neural network model for image detection can be set in the image processing module 320. For example, a convolutional neural network model (CNN) can be set in the image processing module 320, and a deep convolutional generative adversarial network (DCGAN) can also be set. However, it is not limited to this, and other types of neural network models can also be set in the image processing module 320, as long as the image information can be processed for defect detection.

[0084] See also Figure 4 , Figure 5 and Figure 6 Further, performing defect detection processing on the image information may include preprocessing the image information to obtain preprocessed image information. Then, performing multi-parameter defect detection on the preprocessed image information to generate a detection result. For example, see Figure 5 , Figure 5 for Figure 4 The surface images of the first row of sheets to be tested in the sample library are displayed in the corresponding way after preprocessing. Figure 6 for Figure 5 Specifically, preprocessing the image information may include performing denoising, image enhancement, geometric transformation, and image segmentation on the image information in sequence.

[0085] De-noising the image information can remove the noise in the image to improve the clarity and quality of the image, and the specific method of the de-noising process can be Gaussian blur or median filtering. That is, the specific steps of de-noising the image information are not limited.

[0086] Performing image enhancement processing on the denoised image information can enhance the useful information in the image, suppress useless information, and make the image more suitable for analysis. The image enhancement processing method can be a histogram equalization method, a contrast adjustment method, or a gamma correction method. However, it is not limited to this, and other types of image enhancement methods can also be used, that is, the specific method of performing image enhancement processing on the image information can also be not limited.

[0087] Performing geometric transformation on the enhanced image information can correct possible distortion in the image, thereby facilitating subsequent image detection. The geometric transformation may include, but is not limited to, rotating, scaling, and translating the image information. The transformation method may be an affine transformation or a perspective transformation.

[0088] Image segmentation refers to dividing the image after image enhancement into multiple meaningful regions for subsequent analysis. The image segmentation method can be threshold segmentation, region growing, watershed algorithm, etc.

[0089] Furthermore, multi-parameter defect detection of pre-processed image information may include, but is not limited to, flatness detection, mesh detection, crease detection, scratch detection, oil stain detection, and color difference detection of pre-processed image information. Flatness detection can be used to detect the flatness of the image surface and identify the unevenness or wavy shape of the surface. For example, it can be evaluated by analyzing the grayscale change. Mesh detection can be used to detect and calculate the mesh count of the material surface grid to ensure that it meets the specifications.

[0090] For example, methods such as Fourier transform or frequency domain analysis can be used. Crease detection can be used to detect creases or wrinkles on the surface of a material. For example, edge detection can be combined with morphological processing to identify and locate the crease area. Scratch detection can be used to detect scratches on the surface of an image and identify their location and severity. For example, edge detection techniques (such as the Canny operator) can be used to identify elongated scratch areas. Oil stain detection can be used to detect oil stains or other foreign matter on the surface of a material. For example, color segmentation or threshold segmentation techniques can be used to identify color abnormal areas (such as HSV color space conversion). Color difference detection can be used to detect color consistency on the surface of an image and identify color difference areas. For example, this can be achieved through color space conversion (such as Lab color space) or color histogram comparison.

[0091] See also Figure 2 and Figure 3 The control module 330 can be disposed on the vertical rod 122, and the control module 330 can be respectively connected to the weighing mechanism 200 and the image processing module 320 for communication. The control module 330 is used to obtain the quality information of the sheet to be detected in the weighing mechanism 200 and the size information of the sheet to be detected identified in the image processing module 320, and perform surface density calculation processing on the corresponding sheet to be detected according to the quality information and the size information to generate surface density information.

[0092] Specifically, the control module 330 is connected in communication with the weighing sensor 210 in the weighing mechanism 200 to obtain the quality information of the sheet to be detected detected by the weighing sensor 210. The control module 330 is also connected in communication with the image processing module 320, and the control module 330 and the image processing module 320 can communicate by wire or by wireless. The control module 330 can calculate the area information of the sheet to be detected based on the acquired size information, and can calculate the surface density information of the sheet to be detected based on the area information and the quality information.

[0093] Further, the control module 330 may include but is not limited to a control system cabinet 331 and an operation screen 332. The control system cabinet 331 may be fixedly mounted on the vertical rod 122. The operation screen 332 may be fixedly mounted on the vertical rod 122. The control system cabinet 331 is used to determine that when the defect detection result meets the preset qualified condition and the surface density information is within the preset density range, the corresponding sheet to be detected is qualified.

[0094] When the control system cabinet 331 determines that the defect detection result does not meet the preset qualified conditions or the surface density information is not within the preset density range, the control system cabinet 331 can generate a warning message and a re-inspection prompt message, and display it on the operation screen 332 to prompt the staff to re-inspect the corresponding sheet to be detected. However, it is not limited to this. When the control system cabinet 331 determines that the defect detection result does not meet the preset qualified conditions and the surface density information is not within the preset density range, the control system cabinet 331 can generate a warning message and a sheet change prompt message, and display it on the operation screen 332 to prompt the staff to change the sheet.

[0095] Furthermore, re-inspection of the sheet to be inspected means placing the sheet to be inspected back on the tray 220 to repeatedly obtain the quality information of the sheet to be inspected, and issuing operation instructions to the control system cabinet 331 through the operation screen 332 to control the data acquisition module 310 to repeatedly collect the surface image of the sheet to be inspected. The image processing module 320 performs size recognition processing and defect detection processing on the surface image respectively to generate the size information of the sheet to be inspected and the defect detection result. The control system cabinet 331 repeats the above judgment steps. If the defect detection result does not meet the preset qualified conditions and / or the surface density information is not within the preset density range, a warning message and a sheet change prompt message are generated and displayed in the operation screen 332 to prompt the staff to perform the sheet change process. If it is determined that the defect detection result meets the preset qualified conditions and the surface density information is within the preset density range, the sheet to be inspected is inspected and qualified.

[0096] It is worth further explaining that, when the control system cabinet 331 determines that the sheet to be inspected has passed the inspection, the control system cabinet 331 can also generate a number prompt information and display it on the operation screen 332 to remind the operator to perform label number processing on the sheet that has passed the quality inspection. However, it is not limited to this. After the operator performs label number processing on the sheet that has passed the quality inspection, the operator can also confirm that the numbering is completed on the human-computer interaction interface of the operation screen 332, and then the control system cabinet 331 continues to control the sliding component 140 to drive the data acquisition module 310 to slide the label number directly above the position, and collect the surface image of the quality inspection-qualified sheet with the label number. The image processing module 320 performs number recognition processing on the surface image with the label number information to generate the label number information of the quality inspection-qualified sheet. The control system cabinet 331 can store and process the quality information, size information, surface density information, defect detection results and label number information corresponding to the quality inspection-qualified sheet to generate a test record report.

[0097] See also Figure 2 and Figure 3 The visual inspection mechanism 300 also includes a data monitoring module (Supervisory Control and Data Acquisition, SCADA) 340 and a mobile monitoring terminal 350. The data monitoring module 340 can be set on the vertical rod 122, and the data monitoring module 340 is connected to the control system cabinet 331 for communication, so as to obtain the inspection record report of the qualified sheet in the control system cabinet 331, and send it to the mobile monitoring terminal 350 for remote monitoring, thereby realizing the digital quality inspection of large-size sheets and improving the quality inspection efficiency of sheets.

[0098] It can be seen that in the above scheme, the quality information of the sheet to be inspected is obtained by setting a weighing mechanism, the data acquisition module collects the surface image of the sheet to be inspected on the weighing mechanism, and the image processing module is set to perform size recognition processing on the surface image to generate the size information of the sheet to be inspected, and the control module calculates the surface density of the sheet to be inspected according to the quality information and the size information to generate the surface density information, and the image processing module also performs defect detection processing on the surface image to generate a defect detection result. Finally, according to the control module, it is determined that when the defect detection result meets the preset qualified conditions and the surface density information is within the preset density range, the corresponding sheet to be inspected is qualified, thereby improving the quality inspection accuracy and quality inspection efficiency of the sheet to be inspected.

[0099] Furthermore, by setting up a control system cabinet, the quality information, size information, surface density information, defect detection results and label number information corresponding to the qualified sheet materials are stored and processed to generate an inspection record report, and the inspection record report is sent to the mobile monitoring terminal for remote monitoring, which can facilitate direct observation and monitoring by the operator and improve the quality inspection efficiency of the sheet materials to be inspected.

[0100] See also Figure 3 and Figure 7 The present invention also provides a quality inspection method for large-size sheets, which can be applied to the above-mentioned quality inspection device 10 to realize digital quality inspection of large-size sheets, thereby improving the quality inspection efficiency of large-size sheets. The quality inspection method for large-size sheets can include steps S10 to S50, which are described in detail as follows.

[0101] Step S10: Obtain quality information and surface image of the sheet to be tested.

[0102] Step S20: Perform size recognition processing on the surface image to generate size information of the sheet to be detected.

[0103] Step S30: perform defect detection processing on the surface image to generate a defect detection result.

[0104] Step S40: performing surface density calculation processing on the corresponding sheet material to be detected according to the quality information and the size information to generate surface density information.

[0105] Step S50: When the defect detection result meets the preset qualified condition and the surface density information is within the preset density range, the corresponding sheet to be detected is qualified.

[0106] See also Figures 8 to 12The present invention also provides a quality inspection system for large-size sheets, which can be applied to the quality inspection method described above. The quality inspection system for large-size sheets can include an information acquisition module 20, a preprocessing module 30, a convolutional neural network module 40 and a result output module 50.

[0107] In an exemplary embodiment, the information acquisition module 20 may be used to acquire quality information and a surface image of the sheet to be inspected.

[0108] In an exemplary embodiment, the pre-processing module 30 may be used to perform size recognition processing on the surface image to generate size information of the sheet to be detected.

[0109] In an exemplary embodiment, the convolutional neural network module 40 may be used to perform defect detection processing on a surface image to generate a defect detection result.

[0110] Furthermore, the convolutional neural network module 40 can be used to extract image features of the surface image and predict the category to which the image features belong. The convolutional neural network module 40 is also used to continuously adjust network parameters through training data, ultimately achieving automatic image feature recognition, and classifying and saving these features.

[0111] For example, the convolutional neural network module 40 is a deep learning model, which is composed of a data input layer (input layer), a convolutional calculation layer (convolutional layer), a pooling layer (pooling layer), an output layer, etc. For a model structure diagram, please refer to Fig. 9 .

[0112] First, when the model structure of the convolutional neural network module 40 is Fig. 9 When , the data input layer can be used to receive the surface image and convert the image into a pixel matrix through image processing to generate an initial pixel matrix.

[0113] The convolution calculation layer can perform feature extraction processing on the initial pixel matrix through convolution operation to generate the convolution layer output feature map. Specifically, this model can set multiple convolution kernel filters as needed, each convolution kernel can perform convolution operation on the initial pixel matrix separately, and follow the principle of "from left to right, from top to bottom" to extract the local features of the initial pixel matrix, and the parameters of the convolution kernel can be automatically obtained through continuous training and learning.

[0114] Furthermore, in order to improve the convolution effect, the initial pixel matrix can be subjected to feature extraction processing by adopting the boundary pixel padding method (Padding) with the same parameters based on a deep learning framework. The deep learning framework may be a tensor learning framework (TensorFlow), a Keras framework, or the like.

[0115] The formula for calculating the width and height parameters of the convolution layer output feature map after a round of convolution operation of the initial pixel matrix through the general convolution formula is as follows:

[0116]

[0117]

[0118] Among them, W output It can be expressed as the width of the convolutional layer output feature map, W input It can be expressed as the width of the initial pixel matrix, W filter It can be expressed as the width of the convolution kernel, H output It can be expressed as the height of the convolutional layer output feature map, H input It can be expressed as the height of the initial pixel matrix, H filter It can be expressed as the height of the convolution kernel, S can represent the step size of the convolution kernel, and P can represent the number of padding pixel circles added to the edge of the image.

[0119] Secondly, the pooling layer is used to maximize the extraction of the core features of the convolutional layer output feature map to generate a pooled feature map. The pooling layer can not only compress the convolutional layer output feature map, but also reduce the parameters involved in the model operation to improve the computational efficiency. Specifically, the pooling layer can use the maximum pooling method to downsample the convolutional layer output feature map to obtain a pooled feature map. The specific operation process can be found in Fig.10 .

[0120] Furthermore, the output layer can be composed of a fully connected layer and an activation function layer (softmax layer). Among them, the fully connected layer mainly acts as a classifier, which can be used to compress the pooled feature map and integrate and classify the compressed feature image obtained after compression. In addition, each neuron in the fully connected layer is connected to all neurons in the previous layer. These neurons are connected to the previous layer through weight parameters (weights) and biases (biases) to calculate new output values. The output value is usually an unnormalized score (logits) and is represented as the target score. The connection relationship diagram of the fully connected layer and the activation function layer in the output layer can be found in Fig.11 .

[0121] The scores (logits) output by the fully connected layer are passed to the Softmax layer, which can convert these unnormalized scores into probability distributions, where each output value represents the probability of belonging to a certain category, to generate defect detection results. The Softmax function converts each output value into a probability value between 0 and 1, and ensures that the sum of all output probabilities is 1. For details, please refer to Fig.12, which is a schematic diagram of the process of processing the pooled feature map by the fully connected layer and the activation function layer in the output layer.

[0122] For example, after the input surface image is processed by the convolution calculation layer and the pooling layer, a pooling feature map of size 4x4 is obtained. Then this pooling feature map can be flattened into a vector of size 16. This vector is input into a fully connected layer, which has 10 neurons, each neuron represents a category. The output of the fully connected layer is a 10-dimensional vector, each element represents the unnormalized score (logits) of the corresponding category. Finally, this 10-dimensional vector is input into the Softmax layer, and the output is a 10-dimensional probability distribution, each element represents the probability that the input image belongs to the corresponding category. Through the above process, the convolutional neural network model (CNN) can extract features from the input image and finally output the probability that the image belongs to each category to generate a defect detection result. In an exemplary embodiment, the result output module 50 can be used to calculate the surface density of the corresponding sheet to be detected according to the quality information and size information to generate the surface density information. However, it is not limited to this. The result output module 50 is also used to determine that the corresponding sheet to be detected is qualified when the defect detection result meets the preset qualified condition and the surface density information is within the preset density range.

[0123] In the description of this specification, the description with reference to the terms "this embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0124] The embodiments of the present invention disclosed above are only used to help illustrate the present invention. The embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A quality inspection device for large-size nickel sheets, characterized in that: include: The support frame comprises at least one base, a vertical rod and a horizontal rod, wherein one end of the vertical rod is fixedly connected to the base, and one end of the horizontal rod is fixedly connected to one end of the vertical rod away from the base; A weighing mechanism, arranged on the support frame, for obtaining quality information of the sheet to be tested; A visual inspection mechanism, including a data acquisition module, an image processing module and a control module; Wherein, the data acquisition module is arranged on the support frame, and the data acquisition module is located above the weighing mechanism, and is used to collect the surface image of the sheet to be detected on the weighing mechanism; The image processing module is used to perform size recognition processing on the surface image to generate size information of the sheet to be detected. The image processing module is also used to perform defect detection processing on the surface image to generate defect detection results. A convolutional neural network model for image detection is set in the image processing module; the convolutional neural network model includes a data input layer, a convolution calculation layer, a pooling layer and an output layer; the data input layer is used to receive a surface image, and convert the surface image into a pixel matrix through image processing to generate an initial pixel matrix; the convolution calculation layer is based on a deep learning framework, and uses a boundary pixel filling method with the Same parameter to perform feature extraction processing on the initial pixel matrix to generate a convolution layer output feature map; the pooling layer uses a maximum pooling method to downsample the convolution layer output feature map to obtain a pooling feature map; the output layer includes a fully connected layer and an activation function layer; the fully connected layer is used to compress the pooling feature map, and perform model integration and classification on the compressed feature image obtained after compression to calculate the target score; the activation function layer is used to convert the target score into a probability distribution to generate the defect detection result; wherein, the formulas for the parameters of the width and height of the convolution layer output feature map are as follows: W output Represented as the width of the convolutional layer output feature map, W input Represented as the width of the initial pixel matrix, W filter Represented as the width of the convolution kernel, H output Represented as the height of the convolutional layer output feature map, H input Represented as the height of the initial pixel matrix, H filter It is represented as the height of the convolution kernel, S is the step size of the convolution kernel, and P is the number of padding pixels added to the edge of the image; The control module is arranged on the vertical rod, and the control module comprises: A control system cabinet is fixedly mounted on the vertical rod; An operation screen is fixedly arranged on the vertical rod; Wherein, the control system cabinet is used to perform surface density calculation processing on the corresponding sheet material to be detected according to the quality information and the size information to generate surface density information; The control system cabinet is used to determine that when the surface density information is not within a preset density range or the defect detection result does not meet a preset qualified condition, a warning message and a re-inspection prompt message are generated and displayed on the operation screen; The control system cabinet is also used to determine that when the defect detection result meets a preset qualification condition and the surface density information is within a preset density range, the corresponding sheet material to be detected is qualified.

2. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: The data acquisition module is arranged on the cross bar.

3. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: A transport assembly is provided on one side of the support frame, and the transport assembly comprises: A transport body, arranged on one side of the base; A lifting drive mechanism is arranged on the transport body; A lifting bracket is drivingly connected to the lifting drive mechanism, and the lifting bracket is slidably connected to the transport body through the lifting drive mechanism, so as to transport the sheet to be detected so as to load the sheet to be detected onto the weighing mechanism.

4. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: The base is provided with a power wheel mechanism, the weighing mechanism is provided on the power wheel mechanism, and the weighing mechanism comprises: A weighing sensor, fixedly arranged on the housing of the power wheel mechanism; A tray, arranged on the weighing sensor, for carrying the sheet to be tested on the lifting bracket; A display, fixedly disposed at one end of the tray; Wherein, the weighing sensor is used to obtain the quality information of the sheet to be detected in the tray and display it on the display.

5. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: The cross bar is provided with a sliding component, the data acquisition module is fixedly provided on the sliding component, and the data acquisition module slides along the cross bar with the cooperation of the sliding component to acquire the picture information of the display.

6. The quality inspection equipment for large-size nickel sheets according to claim 5, characterized in that: The sliding assembly comprises: An electric slide rail is fixedly arranged on the cross bar; A slider is slidably arranged on the electric slide rail, the travel direction of the slider is arranged parallel to the long side of the cross bar, and the slider is fixedly connected to the data acquisition module; Wherein, under the control of the control module, the electric slide rail drives the slider to drive the data acquisition module to slide to a preset position to acquire the picture information of the display.

7. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: The image processing module is fixedly arranged on the vertical rod, and the image processing module is also used to perform information recognition processing on the image information of the display to generate quality information of the corresponding sheet to be detected in the tray.

8. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: A transport component is arranged on one side of the support frame, and the transport component is a forklift lifting mechanism.

9. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: The support frame is in a U shape.

10. The quality inspection equipment for large-size nickel sheets according to claim 1, characterized in that: The image processing module is a high-performance image processing computer.

Images