Monitoring system of flexible feeding disc based on visual identification

Through a monitoring system based on visual identification, material defects and equipment failures in the operation of the flexible loading tray feeding mechanism are identified, material wear problems are solved, product quality and production efficiency are improved, and production costs are reduced.

CN119048451AInactive Publication Date: 2024-11-29ZHONGSHAN LEILING TECHNOLOGY CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411101524.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art has failed to effectively monitor and solve the defects in the material during operation during the flexible loading tray feeding mechanism or wear problems caused by vibration during transportation, which affects product quality, production efficiency and production costs.

Method used

Adopt a monitoring system based on visual recognition, including data acquisition module, data analysis module, defect identification module and fault detection module, train artificial intelligence models through image data and operation data, identify material defects and calculate fault evaluation coefficients, and send early warning information to prevent Equipment failure.

Benefits of technology

Effectively identify and deal with material defects, improve product quality and production efficiency, reduce production costs, and ensure the stable operation of flexible loading tray equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048451A_ABST
    Figure CN119048451A_ABST
Patent Text Reader

Abstract

The invention discloses a monitoring system for a flexible feeding tray based on visual identification, relates to the technical field of flexible feeding tray monitoring, and solves the technical problems that the product quality, the production efficiency and the production cost are affected as defects of materials before feeding or defects occur in the transportation process are not considered in the prior art. The method comprises the steps that image data of materials in a target area and operation data of a flexible feeding disc are obtained; processing the historical image data to obtain material defect data; training an artificial intelligence model based on the historical image data and the material defect data to obtain a material defect identification model, identifying the image data based on the material defect identification model to obtain an identification result, and processing the material based on the identification result; calculating a fault evaluation coefficient based on the operation data; and based on the fault evaluation coefficient and a preset threshold value, judging whether to carry out early warning on the flexible charging tray equipment. The technical problem is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of flexible loading tray monitoring, relates to visual recognition technology, and specifically is a flexible loading tray monitoring system based on visual recognition. Background Art

[0002] Flexible feeding tray, also known as flexible vibration tray, flexible vibration tray or flexible feeder, is an auxiliary feeding machine for automated assembly or fully automatic processing equipment. Compared with conventional manual feeding methods, it relatively improves the efficiency of production and processing.

[0003] The prior art (invention patent with application number 2023111532472) discloses an operation monitoring system for a flexible vibration plate feeding mechanism, including a distribution image determination module, a vibration distribution evaluation module and an operation monitoring management module; the distribution image determination module analyzes the material distribution image on the vibration plate to analyze the material distribution coverage uniformity and material direction deviation; the vibration distribution evaluation module is used to train a vibration distribution evaluation model, and import the position coordinates of each material center in the coordinate system where the vibration plate is located, the material distribution coverage uniformity and material direction deviation on the vibration plate into the vibration distribution evaluation model to obtain the vibration distribution evaluation coefficient under the current vibration frequency and vibration phase; and according to whether the vibration distribution evaluation coefficient meets the requirements, the vibration plate feeding mechanism is operated and managed and controlled. This invention patent improves the feeding effect and achieves the stability of the feeding mechanism operation by monitoring the feeding effect and regulating the vibration parameters according to the interference loss degree of the vibration parameters. However, the existing technology does not take into account that the material may have defects before loading, or that the vibration of the vibrating plate may cause severe friction between the material and the plate surface during transportation, accelerating the wear of the material and causing defects, which directly affects product quality, production efficiency and production costs.

[0004] Therefore, the present invention proposes a monitoring system for a flexible loading tray based on visual recognition to solve the above problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a monitoring system for a flexible loading tray based on visual recognition, which is used to solve the technical problems that the prior art does not take into account the defects of the material before loading, or the vibration of the vibrating tray during transportation causes severe friction between the material and the tray surface, accelerating the wear of the material and causing defects, thereby directly affecting product quality, production efficiency and production cost.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides a monitoring system for a flexible loading tray based on visual recognition, comprising: a data acquisition module, a data analysis module, a defect recognition module and a fault detection module;

[0007] Data acquisition module: used to obtain image data of materials in the target area and operation data of the flexible loading tray;

[0008] Data analysis module: obtains material defect data by processing historical image data;

[0009] Defect recognition module: trains an artificial intelligence model based on historical image data and material defect data to obtain a material defect recognition model; and

[0010] Identify the image data based on the material defect recognition model, obtain the recognition result, and process the material based on the recognition result;

[0011] Fault detection module: calculate the fault assessment coefficient based on the operating data; determine whether to issue an early warning to the flexible loading tray equipment based on the fault assessment coefficient and the preset threshold; if yes, send an early warning message to the user end; if not, continue to detect and judge; the user end includes: a mobile phone or a computer.

[0012] Preferably, the data analysis module communicates with and / or is electrically connected to the data acquisition module and the defect identification module respectively; the fault detection module communicates with and / or is electrically connected to the data acquisition module.

[0013] Preferably, the step of obtaining image data of materials in the target area includes:

[0014] The image data of the target area is collected in real time by an image acquisition device; wherein the image acquisition device includes: a camera or a video camera; the target area refers to the disk surface area of ​​the flexible loading disk; and the image data refers to the image of the material on the disk surface.

[0015] Preferably, the operation data of the flexible loading tray includes:

[0016] The operation data of the flexible loading tray is collected by a data collection device; wherein the operation data includes: the motor temperature of the flexible loading tray device, the vibration frequency of the loading tray and the vibration amplitude of the loading tray.

[0017] Preferably, the material defect data is obtained by processing the historical image data, including:

[0018] Based on the historical image data, the corresponding material defect types are matched from the database; wherein the database is an existing data set storing various types of material defects; the material defect types include: shape mismatch, size mismatch or material surface damage;

[0019] The historical image data is classified according to the types of material defects to obtain material defect data; wherein the historical image data is the historical data of the material image data of the target area.

[0020] Preferably, the training of the artificial intelligence model based on historical image data and material defect data includes:

[0021] Integrate historical image data into standard input data, and integrate material defect data corresponding to the historical image data into standard output data;

[0022] The artificial intelligence model is trained based on standard input data and standard output data to obtain a material defect recognition model.

[0023] Preferably, the identifying the image data based on the material defect identification model to obtain the identification result includes:

[0024] By inputting the image data of the target area collected in real time into the material defect recognition model, the material defect type corresponding to the image data recognized by the material defect recognition model is obtained.

[0025] It should be noted that the material defect type corresponding to the image data identified by the material defect recognition model is the recognition result.

[0026] The present invention obtains image data of materials in a target area; and, obtains material defect data by processing historical image data; trains an artificial intelligence model based on the historical image data and the material defect data to obtain a material defect recognition model; and, recognizes the image data based on the material defect recognition model to obtain a recognition result, thereby solving the technical problems that the prior art does not take into account defects in the material before loading, or that during transportation, the vibration of the vibrating plate causes severe friction between the material and the plate surface, which accelerates the wear of the material and causes defects, thereby directly affecting product quality, production efficiency and production cost.

[0027] Preferably, the processing of the material based on the identification result includes:

[0028] S1: converting the image data corresponding to the material defect type in the recognition result into a black-and-white binary image by a binarization method; wherein the white in the black-and-white binary image represents the defect area and the black represents the background; the binarization method includes: a global threshold or an adaptive threshold;

[0029] S2: Based on OpenCV, connected component analysis is performed on all black and white binary images to obtain the connected components of the independent white areas of each black and white binary image;

[0030] S3: Calculate the actual area of ​​the connected component based on the number of pixels of each connected component; the specific calculation formula is as follows:

[0031] Aact=(Apix×Dact×Pmm^2) / (W×H);

[0032] Where Aact is the actual area of ​​the connected component, Apix is ​​the pixel area; Dact is the known actual size reference; Pmm is the pixel resolution, and W×H is the total number of pixels in the image;

[0033] S4: Determine whether the actual area of ​​the connected component is greater than a preset area threshold; if yes, determine that the material corresponding to the connected component is a serious defect; if no, determine that the material corresponding to the connected component is a slight defect; wherein the preset area threshold is set according to the qualification requirements of the actual material.

[0034] It should be noted that the connected component refers to a set of pixel points that have the same attributes (pixel values) and are spatially adjacent in image processing; the known actual size reference refers to the actual width or height of a calibrated area in a known image, and its actual physical size has been measured and recorded before the image is taken, such as placing a standard ruler or reference object of known size in the image (such as a 10cm×10cm calibration plate); the pixel area is the number of pixels in the connected component; the pixel resolution refers to the number of pixels contained per millimeter; the total number of pixels of the image is the width multiplied by the height of the image.

[0035] The present invention converts the image data corresponding to the material defect type in the identification result into a black and white binary image through a binarization method, calculates the actual area of ​​the white defect area, and divides the defect degree of the material into severe defects and minor defects based on the actual area of ​​the defect area and a preset area threshold, which is conducive to arranging different processing flows according to different quality levels. For example, materials with minor defects can enter the repair process, while materials with severe defects are directly scrapped, which solves the problem that the prior art does not divide the degree of material defects.

[0036] Preferably, the calculation of the fault assessment coefficient based on the operating data includes:

[0037] The motor temperature is marked as T, the vibration frequency of the loading tray is marked as L, and the vibration amplitude of the loading tray is marked as F;

[0038] The fault assessment coefficient is calculated by the formula: P = A×ln(T)×tanh(T)+B×(L-ZL)^2+C×(F-ZF)^2; wherein A, B, and C are proportional coefficients, ZL is the vibration frequency of the flexible loading tray equipment under normal operation, ZF is the vibration amplitude of the flexible loading tray equipment under normal operation, ln(*) is a logarithmic function with the natural number e as the base, and tanh(*) is a hyperbolic tangent function.

[0039] Preferably, judging whether to issue an early warning to the flexible loading tray device based on the fault assessment coefficient and the preset threshold value includes:

[0040] Determine whether the fault assessment coefficient is greater than the preset threshold; if yes, generate a warning message; if not, continue to detect and judge.

[0041] It should be noted that the preset threshold is set according to the normal operating parameters of the actual flexible loading tray equipment.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The prior art does not take into account that the material may have defects before loading, or that the vibration of the vibrating plate may cause severe friction between the material and the plate surface during transportation, which accelerates the wear of the material and causes defects, thereby directly affecting product quality, production efficiency and production cost; the present invention obtains image data of the material in the target area; and obtains material defect data by processing historical image data; trains an artificial intelligence model based on the historical image data and material defect data to obtain a material defect recognition model; and recognizes the image data based on the material defect recognition model to obtain a recognition result, thereby solving the above problems.

[0044] 2. The present invention obtains the operating data of the flexible loading tray; calculates the fault assessment coefficient based on the operating data; determines whether to issue an early warning to the flexible loading tray equipment based on the fault assessment coefficient and a preset threshold; if yes, sends an early warning message to the user end; if not, continues to detect and judge; solves the problem that the prior art does not monitor the operating status of the flexible loading tray equipment, resulting in equipment failure and thus causing loading efficiency.

[0045] 3. The present invention converts the image data corresponding to the type of material defect in the identification result into a black-and-white binary image through a binarization method; marks the white area as the defect area, and performs connected component analysis on all black-and-white binary images based on OpenCV to obtain the connected components of the independent white areas of each black-and-white binary image; and calculates the actual size area of ​​the connected component based on the number of pixels of each connected component, and divides the degree of defect of the material into severe defect and minor defect based on the actual area of ​​the defect area and a preset area threshold, which is conducive to arranging different processing flows according to different quality levels. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0047] Figure 1 A schematic diagram of a system module according to an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a specific process of an embodiment of the present invention;

[0049] Figure 3 It is a schematic diagram of fault warning judgment of the flexible loading tray equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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.

[0051] See also Figure 1-Figure 2 , the first aspect of the present invention provides a monitoring system for a flexible loading tray based on visual recognition, including: a data acquisition module, a data analysis module, a defect recognition module and a fault detection module;

[0052] Data acquisition module: used to obtain image data of materials in the target area and operation data of the flexible loading tray;

[0053] Data analysis module: obtains material defect data by processing historical image data;

[0054] Defect recognition module: trains an artificial intelligence model based on historical image data and material defect data to obtain a material defect recognition model; and recognizes image data based on the material defect recognition model to obtain recognition results, and processes the material based on the recognition results;

[0055] Obtain image data of materials in the target area, including:

[0056] The image data of the target area is collected in real time by an image acquisition device; wherein the image acquisition device includes: a camera or a video camera; the target area refers to the disk surface area of ​​the flexible loading disk; and the image data refers to the image of the material on the disk surface.

[0057] By processing historical image data, material defect data is obtained, including:

[0058] Based on the historical image data, the corresponding material defect types are matched from the database; wherein the database is an existing data set storing various types of material defects; the material defect types include: shape mismatch, size mismatch or material surface damage;

[0059] The historical image data is classified according to the types of material defects to obtain material defect data; wherein the historical image data is the historical data of the material image data of the target area.

[0060] Training AI models based on historical image data and material defect data, including:

[0061] Integrate historical image data into standard input data, and integrate material defect data corresponding to the historical image data into standard output data;

[0062] The artificial intelligence model is trained based on standard input data and standard output data to obtain a material defect recognition model.

[0063] Based on the material defect recognition model, the image data is recognized to obtain the recognition results, including:

[0064] By inputting the image data of the target area collected in real time into the material defect recognition model, the material defect type corresponding to the image data recognized by the material defect recognition model is obtained.

[0065] Process the material based on the identification results, including:

[0066] S1: converting the image data corresponding to the material defect type in the recognition result into a black-and-white binary image by a binarization method; wherein the white in the black-and-white binary image represents the defect area and the black represents the background; the binarization method includes: a global threshold or an adaptive threshold;

[0067] S2: Based on OpenCV, connected component analysis is performed on all black and white binary images to obtain the connected components of the independent white areas of each black and white binary image;

[0068] S3: Calculate the actual area of ​​the connected component based on the number of pixels of each connected component; the specific calculation formula is as follows:

[0069] Aact=(Apix×Dact×Pmm^2) / (W×H);

[0070] Where Aact is the actual area of ​​the connected component, Apix is ​​the pixel area; Dact is the known actual size reference; Pmm is the pixel resolution, and W×H is the total number of pixels in the image;

[0071] S4: Determine whether the actual area of ​​the connected component is greater than a preset area threshold; if yes, determine that the material corresponding to the connected component is a serious defect; if no, determine that the material corresponding to the connected component is a slight defect; wherein the preset area threshold is set according to the qualification requirements of the actual material.

[0072] For example, suppose you need to monitor a flexible loading tray with an area of ​​50cm×50cm. Use an industrial camera installed above for real-time image acquisition. The camera resolution is 1280×960 pixels and the pixel resolution is 0.4mm / pixel. Collect one frame of image per second to obtain the real-time status of the material. The specific process is as follows:

[0073] 1. Data analysis module example;

[0074] Assume that a historical image dataset contains 5,000 material images from the past month. Through database matching and preliminary screening, it is known that there are three types of material defects in these images: shape mismatch (1,000 images), size mismatch (1,500 images), and material surface damage (2,500 images). These images are classified by defect type, and a 10 cm × 10 cm calibration plate is prepared for each defect type as a size reference.

[0075] 2. Defect identification module example;

[0076] Data preparation: From 5,000 historical images, select images containing defects, convert each image to a uniform size (e.g., 256x256 pixels), and classify labels according to known defect types.

[0077] Model training: Train a convolutional neural network (CNN) model using a deep learning framework such as TensorFlow or PyTorch. Split the image dataset into a training set (80%), a validation set (10%), and a test set (10%). The training process involves adjusting the model parameters to minimize the gap between the predicted labels and the actual labels.

[0078] Model evaluation: Evaluate the model’s precision, recall, and F1 score on the test set to ensure that the model can accurately identify defects on unseen images.

[0079] 3. Examples of material defect identification and treatment;

[0080] Real-time image acquisition and recognition: From the real-time acquired images, a material image with suspected surface damage was found. The image size is 1280×960 pixels, and it is known that the actual area of ​​the 10cm×10cm calibration plate in the image is 100 square centimeters.

[0081] Binarization: Apply the global threshold binarization method to convert the suspected surface damage area into a black and white binary image. Assume that after binarization, a connected component is obtained, which consists of 1200 pixels.

[0082] Connected component analysis and area calculation:

[0083] The actual area of ​​the connected component is calculated by the formula: Aact = (Apix × Dact × Pmm^2) / (W × H); assuming that Apix = 1200 (pixels), Dact = 10 cm (known actual size reference, assuming the width of the calibration plate here), Pmm = 0.4 mm / pixel (pixel resolution), W × H = 1280 × 960 (total number of pixels in the image);

[0084] Substitute into the formula: Aact = (1200 × 10 × 0.4^2) / (1280 × 960) ≈ 0.078 (square centimeters);

[0085] Defect degree judgment: Assuming that the preset area threshold is 1 square centimeter, since the calculated actual area (0.078) is smaller than the preset area threshold (1), the material defect corresponding to the connected component is judged to be a slight defect.

[0086] 4. Subsequent processing;

[0087] Materials with minor defects will be sent to the repair station for surface treatment or reprocessing to restore their quality to qualified standards. Materials with serious defects (i.e., those with actual area exceeding the preset threshold) will be marked as unqualified and directly removed from the production line to avoid entering the subsequent manufacturing process, thus ensuring the quality of the final product.

[0088] Through this AI-based real-time monitoring and intelligent processing system, production efficiency has been significantly improved, and increased production costs and quality problems caused by material defects have been reduced.

[0089] Fault detection module: calculate the fault assessment coefficient based on the operating data; determine whether to issue an early warning to the flexible loading tray equipment based on the fault assessment coefficient and the preset threshold; if yes, send an early warning message to the user end; if not, continue to detect and judge; the user end includes: a mobile phone or a computer.

[0090] See also Figure 3 , the operating data of the flexible loading tray includes:

[0091] The operation data of the flexible loading tray is collected by a data collection device; wherein the operation data includes: the motor temperature of the flexible loading tray device, the vibration frequency of the loading tray and the vibration amplitude of the loading tray.

[0092] The fault assessment coefficients are calculated based on the operating data, including:

[0093] The motor temperature is marked as T, the vibration frequency of the loading tray is marked as L, and the vibration amplitude of the loading tray is marked as F;

[0094] The fault assessment coefficient is calculated by the formula: P = A×ln(T)×tanh(T)+B×(L-ZL)^2+C×(F-ZF)^2; wherein A, B, and C are proportional coefficients, ZL is the vibration frequency of the flexible loading tray equipment under normal operation, ZF is the vibration amplitude of the flexible loading tray equipment under normal operation, ln(*) is a logarithmic function with the natural number e as the base, and tanh(*) is a hyperbolic tangent function.

[0095] Based on the fault assessment coefficient and the preset threshold, it is determined whether to issue an early warning to the flexible loading tray equipment, including:

[0096] Determine whether the fault assessment coefficient is greater than the preset threshold; if yes, generate a warning message; if not, continue to detect and judge.

[0097] For example, suppose you are monitoring a flexible loading tray device that is equipped with sensors to collect key operating data in real time, including motor temperature, vibration frequency and vibration amplitude of the loading tray. The data acquisition device collects data once a minute for real-time analysis by the fault detection module. The details are as follows:

[0098] 1. Run data collection;

[0099] The operating data of the flexible loading tray equipment is collected through temperature sensors, vibration frequency meters and displacement sensors; it is assumed that the collected data is as follows:

[0100] Motor temperature (T): currently measured at 60°C.

[0101] Loading tray vibration frequency (L): currently measured at 5Hz.

[0102] Feeding tray vibration amplitude (F): currently measured is 0.5mm.

[0103] Normal operating parameters:

[0104] Normal vibration frequency (ZL): 4Hz.

[0105] Normal vibration amplitude (ZF): 0.3mm.

[0106] 2. Calculation of fault assessment coefficient;

[0107] Assume proportional coefficients: A=0.5, B=0.2, C=0.3.

[0108] The fault assessment coefficient is calculated by the formula: P = A × ln (T) × tanh (T) + B × (L-ZL)^2 + C × (F-ZF)^2;

[0109] P=0.5×ln(60)×tanh(60)+0.2×(5-4)^2+0.3×(0.5-0.3)^2≈2.26

[0110] 3. Early warning judgment;

[0111] Assuming that the preset threshold is set to 2, compare the calculated fault assessment coefficient P with the preset threshold:

[0112] Since the fault assessment coefficient (2.26) is greater than the preset threshold (2), this means that the equipment is currently in an abnormal state and immediate action is required.

[0113] 4. Send warning information;

[0114] The system will automatically trigger the early warning mechanism and send early warning information to the user end (for example, the maintenance engineer’s mobile phone or computer), including the current equipment status, fault assessment coefficient, recommended inspection and maintenance measures, etc.

[0115] The present invention monitors the operating parameters of the flexible loading tray equipment. When the operating parameters of the flexible loading tray equipment deviate from the normal range, the fault detection module can timely calculate the fault assessment coefficient and determine whether to issue an early warning according to the preset threshold, thereby effectively preventing potential equipment failures and ensuring the stable operation of the production line. This real-time monitoring and early warning mechanism is crucial to improving production efficiency and reducing maintenance costs.

[0116] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0117] Working principle of the present invention:

[0118] The present invention obtains image data of materials in a target area and operating data of a flexible loading tray; obtains material defect data by processing historical image data; trains an artificial intelligence model based on the historical image data and material defect data to obtain a material defect recognition model, recognizes the image data based on the material defect recognition model to obtain a recognition result, and processes the material based on the recognition result; and, calculates a fault assessment coefficient based on the operating data; determines whether to issue an early warning to the flexible loading tray equipment based on the fault assessment coefficient and a preset threshold, thereby solving the technical problem that the prior art does not take into account the existence of defects in the material before loading or the defects in the transportation process, thereby affecting product quality, production efficiency and production cost.

[0119] 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 skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A monitoring system for a flexible loading tray based on visual recognition, characterized in that: include: Data acquisition module, data analysis module, defect identification module and fault detection module; Data acquisition module: used to obtain image data of materials in the target area and operation data of the flexible loading tray; Data analysis module: obtains material defect data by processing historical image data; Defect recognition module: trains the artificial intelligence model based on historical image data and material defect data to obtain a material defect recognition model; as well as, Identify the image data based on the material defect recognition model, obtain the recognition result, and process the material based on the recognition result; Fault detection module: calculates fault assessment coefficient based on operating data; Determine whether to issue an early warning to the flexible loading tray equipment based on the fault assessment coefficient and the preset threshold; If yes, a warning message is sent to the user end; If not, continue to detect and judge; the user end includes: mobile phone or computer.

2. According to the monitoring system of a flexible loading tray based on visual recognition according to claim 1, it is characterized in that: The data analysis module communicates with and / or is electrically connected to the data acquisition module and the defect identification module respectively; the fault detection module communicates with and / or is electrically connected to the data acquisition module.

3. The monitoring system of a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The step of obtaining image data of a material in a target area includes: The image data of the target area is collected in real time by an image acquisition device; wherein the image acquisition device includes: a camera or a video camera; the target area refers to the disk surface area of ​​the flexible loading disk; and the image data refers to the image of the material on the disk surface.

4. The monitoring system for a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The operation data of the flexible loading tray includes: The operation data of the flexible loading tray is collected by a data collection device; wherein the operation data includes: the motor temperature of the flexible loading tray device, the vibration frequency of the loading tray and the vibration amplitude of the loading tray.

5. The monitoring system for a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The material defect data is obtained by processing the historical image data, including: Based on the historical image data, the corresponding material defect types are matched from the database; wherein the database is an existing data set storing various types of material defects; the material defect types include: shape mismatch, size mismatch or material surface damage; The historical image data is classified according to the types of material defects to obtain material defect data; wherein the historical image data is the historical data of the material image data of the target area.

6. The monitoring system for a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The training of the artificial intelligence model based on historical image data and material defect data includes: Integrate historical image data into standard input data, and integrate material defect data corresponding to the historical image data into standard output data; The artificial intelligence model is trained based on standard input data and standard output data to obtain a material defect recognition model.

7. The monitoring system of a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The method of identifying the image data based on the material defect recognition model to obtain the recognition result includes: By inputting the image data of the target area collected in real time into the material defect recognition model, the material defect type corresponding to the image data recognized by the material defect recognition model is obtained.

8. The monitoring system of a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The processing of the material based on the identification result includes: S1: converting the image data corresponding to the material defect type in the recognition result into a black-and-white binary image by a binarization method; wherein the white in the black-and-white binary image represents the defect area and the black represents the background; the binarization method includes: a global threshold or an adaptive threshold; S2: Based on OpenCV, connected component analysis is performed on all black and white binary images to obtain the connected components of the independent white areas of each black and white binary image; S3: Calculate the actual area of ​​the connected component based on the number of pixels of each connected component; the specific calculation formula is as follows: Aact=(Apix×Dact×Pmm^2) / (W×H); Where Aact is the actual area of ​​the connected component, Apix is ​​the pixel area; Dact is the known actual size reference; Pmm is the pixel resolution, and W×H is the total number of pixels in the image; S4: Determine whether the actual area of ​​the connected component is greater than a preset area threshold; if yes, determine that the material corresponding to the connected component is a serious defect; if no, determine that the material corresponding to the connected component is a slight defect; wherein the preset area threshold is set according to the qualification requirements of the actual material.

9. The monitoring system of a flexible loading tray based on visual recognition according to claim 1 is characterized in that: The calculating of the fault assessment coefficient based on the operation data includes: The motor temperature is marked as T, the vibration frequency of the loading tray is marked as L, and the vibration amplitude of the loading tray is marked as F; The fault assessment coefficient is calculated by the formula: P = A×ln(T)×tanh(T)+B×(L-ZL)^2+C×(F-ZF)^2; wherein A, B, and C are proportional coefficients, ZL is the vibration frequency of the flexible loading tray equipment under normal operation, ZF is the vibration amplitude of the flexible loading tray equipment under normal operation, ln(*) is a logarithmic function with the natural number e as the base, and tanh(*) is a hyperbolic tangent function.

10. The monitoring system of a flexible loading tray based on visual recognition according to claim 1, characterized in that: The method of judging whether to issue an early warning to the flexible loading tray device based on the fault assessment coefficient and the preset threshold value includes: Determine whether the fault assessment coefficient is greater than the preset threshold; if yes, generate a warning message; if not, continue to detect and judge.

Citation Information

Cited By

  • Crystal defect detection method based on machine vision

    CN120161068A

  • Multi-dimensional visual detection system and detection method

    CN121253559A