CN-121974118-A - Inductance core feeding control system, method and medium based on visual detection

CN121974118ACN 121974118 ACN121974118 ACN 121974118ACN-121974118-A

Abstract

The application relates to an inductance core feeding control system, method and medium based on visual detection, and relates to the technical field of battery management; the inductor core feeding control system comprises a material quantity detection module, a vibration driving module, a visual identification module, a material taking and discharging module, a positioning correction module, a mold control module and a core control module, wherein the light and reflection components of a core image are decomposed through a convolutional neural network, enhancement parameters are extracted through a multi-branch prediction network, self-adaptive channel adjustment is performed in an HSV space, finally, light balance images are generated through weighting loss function fusion, the robustness and accuracy of visual detection in a complex industrial light environment are remarkably improved, the front side and the back side of the inductor core are automatically identified through visual detection, and the coordinate positions are marked, so that the automatic feeding, the positioning automatic correction and the automatic discharge of unqualified products of the core are realized, and the production efficiency of the inductor component is improved.

Inventors

Shan Zhongbo
Liang Shiduo

Assignees

苏州鸿华芯创科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260114

Claims (9)

1. Inductance core material loading control system based on visual detection, its characterized in that includes: The output end of the material quantity detection module is connected with the first input end of the core control module and is used for weighing the inductance core body to obtain the weight of the core body; The input end of the vibration driving module is connected with the first output end of the core control module and is used for realizing direct vibration feeding and vibration turning of the inductance core according to the vibration instruction; the output end of the visual identification module is connected with the second input end of the core control module and is used for collecting a core image of the inductance core, identifying the placement state of the core and determining the coordinates of the core; The input end of the material taking and discharging module is connected with the second output end of the core control module, and is used for picking up the inductance core according to the core coordinates, moving to the positioning correction station according to the core transfer path to finish position correction, and moving to the material discharging station to finish target transfer; the input end of the positioning correction module is connected with the third output end of the core control module, and the positioning correction module is used for finishing the position correction of the inductance core body at the positioning correction station according to the abnormal positioning correction quantity; the input end of the die control module is connected with the fourth output end of the core control module and is used for controlling the hot-pressing die cavity to topple the inferior core body according to the abnormal state correction instruction; The core control module is used for receiving the weight of the core body, generating the vibration instruction, receiving the core body placement state and the core body coordinates, and combining a positioning correction station and a discharging station to generate the abnormal positioning correction quantity and the core body transfer path.
2. The vision-detection-based inductor core feeding control system according to claim 1, wherein the core control module comprises: The mass judgment unit is used for judging according to a preset bearing threshold value and the weight of the core body, and generating a vibration instruction if the weight of the core body is smaller than the bearing threshold value; the state detection unit is used for cleaning and dividing the core image to obtain a single-core view, judging the core placement state by combining a preset core front template, screening the abnormal core state, generating an abnormal positioning correction amount and calculating the core coordinates; and the path planning unit is used for combining the positioning correction station and the discharging station according to the core coordinates to generate a core transfer path.
3. The inductance core feeding control system based on visual inspection according to claim 2, wherein the status detection unit includes: the illumination balancing layer is used for carrying out illumination detection on the core image and calculating illumination uniformity; If the illumination uniformity is greater than a preset uniformity threshold, judging that the current core image is an equilibrium core image; If not, carrying out illumination correction on the current core image by combining the self-adaptive illumination adjustment model to obtain the balanced core image; The denoising enhancement layer is used for filtering, denoising and contrast stretching the balanced core diagram to obtain an enhanced core diagram; The edge sharpening layer is used for carrying out edge marking on the reinforced core body diagram to obtain a core body outline diagram; the target conversion layer is used for carrying out difference degree calculation on the core body and the background of the core body profile graph to obtain a gray level difference value; If the gray level difference value is larger than a preset background difference threshold value, calculating an optimal segmentation threshold value and converting the current core contour map into a binary core map; if not, carrying out watershed treatment on the current core profile to obtain the binary core profile; The target screening layer is used for pixelating a preset core physical size interval, calculating a core pixel size interval, and filtering the binary core map to obtain a target core profile map; and the single segmentation layer is used for cutting and segmenting the target core contour map to obtain a single-core view.
4. The vision-detection-based inductor core feeding control system according to claim 3, wherein the state detection unit further comprises: the feature extraction layer is used for carrying out feature detection on a preset core front template and extracting the front features of the template; the state screening layer is used for matching the single-core view according to the front characteristics of the template and calculating the matching quantity of key points and the angle deviation value; If the key point matching number is larger than a preset matching number threshold value and the angle deviation value is located in a preset deviation threshold value interval, judging that the core placement state is the front; if the key point matching number is larger than a preset matching number threshold value, but the angle deviation value is not located in the deviation threshold value interval, judging that the core placement state is positive, and the abnormal core state is an angle out-of-tolerance core or a position deviation core; If not, judging that the core body is placed in a reverse side; The positioning correction layer is used for carrying out angle identification on the angle out-of-tolerance core body to obtain a single-core angle and calculating abnormal positioning correction by combining a target angle; positioning the position deviation core body to obtain single-core center point coordinates, and performing deviation operation with a theoretical placement center point to obtain the abnormal positioning correction quantity; and carrying out coordinate transformation on the coordinates of the single core central point according to the target core contour diagram, and calculating the core coordinates of each inductance core.
5. The vision-detection-based inductor core feeding control system according to claim 3, wherein the adaptive illumination adjustment model comprises: The image decomposer is used for decomposing the core image to generate a reflection component diagram and an illumination component diagram; the parameter predictor is used for extracting main features and residual features of the illumination component diagram through a first prediction branch, and obtaining illumination enhancement feature parameters after fusion operation; performing feature mapping and smooth difference on the reflection component diagram through a second prediction branch to obtain reflection enhancement feature parameters; the self-adaptive regulator is used for carrying out space decomposition on the core image through a channel decomposition branch to obtain a brightness channel V, a saturation channel S and a hue channel H; Carrying out nonlinear transformation on the brightness channel through a first adjusting branch and combining with the illumination enhancement characteristic parameters, carrying out multi-scale noise suppression on the saturation channel, carrying out contrast enhancement on the hue channel, and carrying out fusion operation on three channels to obtain an illumination adjustment chart; performing linear projection and feature segmentation on the core image through a second adjusting branch to obtain triples (R, G, B), and performing scaling integration by combining the reflection enhancing feature parameters to obtain a reflection adjusting diagram; and fusing the illumination regulation map and the reflection regulation map by combining the self-adaptive fusion branch with a cross fusion loss function to obtain the balanced core map.
6. The vision-based inductor core loading control system of claim 5, wherein the cross-fusion loss function The method comprises the following steps: 。
7. the vision-detection-based inductor core feeding control system according to claim 2, wherein the path planning unit comprises: The parameter integration layer is used for positioning and sequencing the positioning correction stations and the discharging stations according to the station layout to obtain correction station coordinates and discharging station coordinates; The state decision layer is used for selecting the correction station coordinates and the discharging station coordinates according to the core placement state to generate a core transfer strategy; the node planning layer is used for generating a core transfer node according to the core coordinates and the discharging station coordinates in combination with environment constraint parameters if the core placement state is positive; If the core placement state is an angle out-of-tolerance core or a position deviation core, the core transfer strategy is to generate the core transfer node according to the core coordinates, the correction station coordinates and the discharge station coordinates in combination with the environment constraint parameters; And the path optimization layer is used for carrying out global smoothing on the core transfer nodes according to the action execution time to construct a core transfer path.
8. An inductor core feeding control method based on visual detection, which is applied to the system of any one of claims 1-7, and is characterized by comprising the following steps: Weighing the inductance core, and finishing feeding according to a preset bearing threshold value to obtain a quantitative core; collecting and detecting a core image of the quantitative core, and correcting and dividing by combining with a self-adaptive illumination adjustment model to obtain a single-core view; identifying the single-core view according to a preset core front template, judging a core placement state, screening an abnormal core state, generating an abnormal positioning correction amount, and calculating a core coordinate; And according to the core coordinates, combining a positioning correction station and a discharging station to generate a core transfer path.
9. A storage medium having stored therein at least one instruction, at least one program, a set of codes, or a set of instructions, the at least one instruction, the at least one program, the set of codes, or the set of instructions being loaded and executed by a processor to implement the inductor core loading control method of claim 8.

Description

Inductance core feeding control system, method and medium based on visual detection Technical Field The application relates to the technical field of battery management, in particular to an inductance core feeding control system, method and medium based on visual detection. Background The hot-pressing process of the inductance component is an important link in the manufacture of electronic components, and in the hot-pressing forming process, a core body (a magnetic core and a coil fixed on the magnetic core) is required to be placed in a hot-pressing die, and after magnetic-conductive alloy powder is filled in a die cavity, the die cavity is hot-pressed into an integrated structure. The prior patent discloses an electric performance and appearance detection system and method based on visual guidance, wherein the system comprises a vibration disc feeding module, a top sorting module, a manipulator module, a front detection module, an electric performance test module and a back detection module, wherein the vibration disc feeding module is used for realizing direct vibration feeding and vibration turning of components to be detected, the top sorting module is arranged above the vibration disc feeding module and is used for realizing front and back detection of the components to be detected, the manipulator module is used for sucking and moving the components to be detected, the front detection module is arranged on the manipulator module and is used for realizing front detection of the components to be detected, the electric performance test module is used for realizing electric performance detection of the components to be detected, the back detection module is used for realizing back detection of the components to be detected, and the PLC control module is used for realizing back detection of the components to be detected. Said invention is convenient for use, can automatically implement feeding, and can implement electric property detection, front detection and back detection, and can implement sorting screening by means of visual guidance, and its flow is fully-automatic, and can be substituted for manual test (it has no need of special-person operation, and can reduce operator load), and its efficiency is higher. In the hot press molding processing process of the inductor, the inductor core body is generally fed by a traditional vibration disc, the feeding efficiency is low, the detection capability of the core body in the feeding process is limited, the wrong placement position cannot be effectively identified and corrected, and the production efficiency and the product quality are reduced. Disclosure of Invention Aiming at the defects existing in the prior art, the application aims to provide an inductance core feeding control system, method and medium based on visual detection, which can automatically identify the front and back sides of the inductance core through visual detection and mark the coordinate positions so as to realize automatic feeding, automatic positioning correction and automatic discharge of unqualified products of the core and improve the production efficiency of the inductance. The method is realized by adopting the following technical scheme: In a first aspect, the present application provides an inductor core feeding control system based on visual detection, including: The output end of the material quantity detection module is connected with the first input end of the core control module and is used for weighing the inductance core body to obtain the weight of the core body; The input end of the vibration driving module is connected with the first output end of the core control module and is used for realizing direct vibration feeding and vibration turning of the inductor core according to the vibration instruction; The output end of the visual identification module is connected with the second input end of the core control module and is used for collecting a core image of the inductance core, identifying the placement state of the core and determining the coordinates of the core; the input end of the material taking and discharging module is connected with the second output end of the core control module, and is used for picking up the inductance core according to the core coordinates, moving to the positioning correction station according to the core transfer path to finish position correction, and moving to the material discharging station to finish target transfer; The input end of the positioning correction module is connected with the third output end of the core control module, and is used for finishing the position correction of the inductance core body at the positioning correction station according to the abnormal positioning correction quantity; the input end of the die control module is connected with the fourth output end of the core control module and is used for controlling the hot-pressing die cavity to topple the inferior core according to the abnormal state cor