CN-116183495-B - Commutator hook type visual detection system and method based on optical-mechanical mixing

CN116183495BCN 116183495 BCN116183495 BCN 116183495BCN-116183495-B

Abstract

The invention relates to the technical field of visual detection of product quality, in particular to a reverser hook-type visual detection system and method based on optical-mechanical mixing, wherein the system comprises a hook-type rotary oblique angle front and back side imaging module and a hook-type dynamic prediction correction measurement module; based on a hook-type rotation oblique angle front and back side imaging module, an unhatched hook-type image to be detected is obtained, a hook-type rotation error compensation value in the hook-type image to be detected is calculated by using a hook-type rotation error prediction model, then hook-type parameters in the image are calculated, and the hook-type rotation error compensation value is used for correcting the current hook-type parameter measurement value to obtain the actual hook-type parameter value. And finally, judging whether the product to be detected is a qualified product or a defective product based on a detection threshold value set by the product standard. The system can rapidly detect the hook shape with high precision, and solves the detection problems caused by the fact that the shape of the commutator hook shape is complex, the size is small, the object rapidly rotates for collecting hook images, and the like.

Inventors

XU LIANG
QIAO JUNJIE
ZHENG BOYUAN

Assignees

广东工业大学

Dates

Publication Date: 20260512
Application Date: 20221230

Claims (10)

1. Commutator hook type visual detection system based on ray machine is mixed, characterized by comprising: The hook type rotary bevel angle front and back side imaging module and the hook type dynamic prediction correction measurement module are arranged; The hook type rotary oblique angle front and back side imaging module comprises a laser sensor, a camera and a rotary front and back side light device, wherein the laser sensor is electrically connected with the camera, the camera is electrically connected with the hook type dynamic prediction correction measuring module, the commutator is arranged in the rotary front and back side light device, the laser sensor is used for sensing the change of the distance between the laser sensor and the commutator to generate an electric signal to trigger the camera to shoot, specifically, laser of the laser sensor irradiates to a hook on the commutator to sense the change of the distance between a copper sheet or the hook on the commutator and the laser sensor, when one hook edge of the commutator is identified through the change of the distance, the laser sensor emits the electric signal to trigger the camera to shoot, the camera is used for collecting an image of the hook to be detected in a specific direction, and sending the image of the hook to be detected to the hook type dynamic prediction correction measuring module, wherein the specific direction is the direction in which the hooks of the commutator in the image are not blocked by each other, and the rotary front and back side light device is used for enabling the laser sensor and the camera to rotate relatively to the axis of the commutator respectively, and providing a multi-angle integrated light beam required for imaging for the commutator; the hook type dynamic prediction correction measurement module is connected with the camera and used for acquiring the width of a copper sheet in a hook image to be detected in the camera, inputting the width of the copper sheet into the hook type rotation error prediction model to calculate a hook type rotation error compensation value, wherein the hook type rotation error prediction model is obtained by training according to the width of the copper sheet under different shooting angles and the data corresponding to the hook rotation error, calculating the hook type parameters in the hook image to be detected, correcting the current hook type parameters by using the hook type rotation error compensation value to obtain a hook type detection actual value, and judging whether the hook is qualified or not according to a preset hook type standard range.
2. The optical-mechanical hybrid-based commutator hook-type visual detection system according to claim 1, wherein in the hook-type rotary oblique angle front-back side imaging module, the specific direction of the camera is specifically: The shooting direction of the camera is parallel to the surface of a copper sheet, and the included angle between the shooting direction and the side surface of the hook on the copper sheet is not larger than the included angle of a gap, wherein the included angle of the gap is the included angle between the connecting line of the highest point of the hook on the copper sheet and the lowest point of the adjacent hook and the side surface of the hook on the copper sheet.
3. The optical-mechanical-hybrid-based visual inspection system for a commutator hook of claim 1, wherein in the hook-type rotary oblique angle front and back imaging module, the specific direction of the camera is that the camera and the commutator hook are in the same horizontal plane, and the shooting direction of the camera is perpendicular to a copper sheet on the commutator and faces the side face of the adjacent hook of the copper sheet.
4. The optical-mechanical-hybrid-based hook-type visual inspection system for a commutator of claim 1, wherein the hook-type dynamic prediction correction measurement module is further configured to segment an image of a hook to be inspected, and identify copper sheets on the commutator in the segmented image.
5. The optical-mechanical-hybrid-based commutator hook visual detection system according to claim 1, wherein the hook dynamic prediction correction measurement module is further configured to establish a coordinate system on an image of a hook to be detected, identify a front end region of an unobstructed hook in the image of the hook to be detected, and calculate a center point coordinate of the front end region, where the center point coordinate is used to calculate the hook parameters.
6. The optical-mechanical-hybrid-based commutator hook visual detection system according to claim 1, wherein the hook dynamic prediction correction measurement module is further configured to determine whether the actual hook detection value falls within a preset hook standard range, and if so, determine that the hook is qualified.
7. The optical-mechanical-mixing-based visual detection method for the hook type of the commutator, which is based on the optical-mechanical-mixing-based visual detection system for the hook type of the commutator, disclosed by any one of claims 1 to 6, is characterized by comprising the following steps: the hook image to be detected is segmented by a hook dynamic prediction correction measurement module, adjacent copper sheets of a target hook in the hook image to be detected are identified, and the width of the copper sheets of the adjacent copper sheets is obtained; inputting the width of the copper sheet into a hook-type rotation error prediction model to obtain a rotation error compensation value required by a target hook-type parameter; Recognizing the front end area of the hook in the hook image to be detected, obtaining the center point coordinate of the front end area, and calculating the hook type parameter according to the center point coordinate; and combining the hook type parameter with the rotation error compensation value to obtain a hook type detection actual value, and judging whether the hook is qualified or not.
8. The optical-mechanical-hybrid-based visual detection method for hook type of a commutator of claim 7, wherein the identifying the front end region of the hook in the image of the hook to be detected, after obtaining the center point coordinates of the front end region, calculates the hook type parameters according to the center point coordinates, specifically comprises: Setting front end area F of commutator hook in image ) And calculating the center point coordinate P of the front end area of the commutator hook obtained from the area center The edge of the area, close to the front end, of the adjacent copper sheet is L 1 % ) The formula for calculating the hook type parameter D 1 is specifically as follows: ( ) Wherein, the As a reference point on the edge of an adjacent copper sheet, Is on-line with the inner side edge of the adjacent copper sheet, Is at point P Shortest distance, i.e. point P at Projecting upwards, calculating to obtain front end area F of commutator hook ) Inner side edge The distance between the two is the hook type parameter D 1 .
9. The optical-mechanical-hybrid-based visual detection method for hook type of commutator of claim 7, wherein inputting the width of the copper sheet into a hook type rotation error prediction model to obtain a rotation error compensation value required by a target hook type parameter specifically comprises: Acquiring the width of a copper sheet in an obtained image, wherein the shooting direction of a camera forms different included angles with the side surface of a hook, and the copper sheet is an adjacent copper sheet of the copper sheet where the non-shielding hook is positioned; Calculating error values between hook-type parameter measured values and hook-type actual values in images under different included angles; Generating a training set and a verification set of a hook-type rotation error prediction model according to the width of the copper sheet and the corresponding hook-type error value; And establishing a hook-type rotation error prediction model by using a training set based on a depth support vector regression method, checking the model by using a verification set, and obtaining a corresponding model after the checking precision reaches a preset threshold value.
10. The optical-mechanical hybrid-based commutator hook visual detection method according to claim 7, wherein before the hook image to be detected is segmented by the hook dynamic prediction correction measurement module, the method further comprises: And starting a rotary front-back side light device in the hook-type rotary oblique angle front-back side imaging module to enable the camera, the laser sensor and the rotary front-back side light device to rotate relatively with the commutator around the axis of the commutator, detecting the distance between the commutators by the laser sensor, triggering the camera to shoot when the distance changes, and acquiring an image of a hook to be detected of a hook which is not blocked by an adjacent hook.

Description

Commutator hook type visual detection system and method based on optical-mechanical mixing Technical Field The invention relates to the technical field of visual detection of product quality, in particular to a commutator hook type visual detection system and method based on optical-mechanical mixing. Background The motor is an essential basic device which is indispensable in industry, traffic, national defense and daily life, and the commutator is used as one of core devices of the motor, and the quality of the commutator directly influences the quality of the motor. The appearance quality detection of the commutator, in particular to the hook-type defect detection of the commutator, is an important index for evaluating the quality of the commutator, and is also an important procedure in the production of the commutator. At present, the hook-type detection of the commutator still adopts a manual mode, and the problems of low detection efficiency, high misjudgment rate and high omission factor, serious measurement error, large labor force required for manual detection, high cost, large error and poor consistency of detection quality are caused by the complex shape and small size of the hook-type detection of the commutator. Disclosure of Invention The application provides a commutator hook type inspection and detection system and method based on optical-mechanical mixing, which solve the problems of high manual detection cost, low precision and poor detection quality consistency in the prior art. The first aspect of the application provides a commutator hook type visual detection system based on optical-mechanical mixing, which comprises the following components: The hook type rotary bevel angle front and back side imaging module and the hook type dynamic prediction correction measurement module are arranged; The hook-type rotation oblique angle front and back side imaging module comprises a laser sensor, a camera and a rotation front and back side light device, wherein the laser sensor is electrically connected with the camera, the camera is electrically connected with the hook-type dynamic prediction correction measuring module, the commutator is arranged in the rotation front and back side light device, the laser sensor is used for sensing the change of the distance between the laser sensor and the commutator to generate an electric signal to trigger the camera to shoot, the camera is used for collecting a hook-type image to be detected in a specific direction, and sending the hook-type image to the hook-type dynamic prediction correction measuring module, the specific direction is the direction in which the commutator hooks in the image are not blocked, and the rotation front and back side light device is used for enabling the laser sensor and the camera to rotate relatively to the commutator around the axis of the commutator respectively and providing multi-angle integrated light beams required by imaging for the commutator; The hook type dynamic prediction correction measurement module is connected with the camera and used for acquiring the width of a copper sheet in a hook type image to be detected in the camera, inputting the width of the copper sheet into the hook type rotation error prediction model to calculate a hook type rotation error compensation value, wherein the hook type rotation error prediction model is obtained by training the copper sheet width under different shooting angles and the data corresponding to the hook type rotation error, calculating the hook type parameters in the hook type image to be detected, correcting the current hook type measured value by using the hook type rotation error compensation value to obtain a hook type detection actual value, and judging whether the hook type is qualified or not according to the hook type detection actual value. Optionally, in the hook-type rotary oblique angle front and back side imaging module, the specific direction of the camera is specifically: The shooting direction of the camera is parallel to the surface of a copper sheet, and the included angle between the shooting direction and the side surface of the hook on the copper sheet is not larger than the included angle of a gap, wherein the included angle of the gap is the included angle between the connecting line of the highest point of the hook on the copper sheet and the lowest point of the adjacent hook and the side surface of the hook on the copper sheet. Optionally, in the hook-type rotary oblique angle front and back side imaging module, the specific direction of the camera is that the camera and the commutator hook are in the same horizontal plane, and the shooting direction of the camera is perpendicular to a copper sheet on the commutator and faces the side face of the copper sheet adjacent to the hook. Optionally, the hook-type dynamic prediction correction measurement module is further used for dividing the hook-type image to be detected, and identifying the copper shee