Nondestructive detection device and method for magnetic shoe defects

Abstract

The invention discloses a magnetic shoe defect nondestructive testing device and method, and belongs to the technical field of magnetic shoe detection.The magnetic shoe defect nondestructive testing device comprises a power mechanism, an oscillation mechanism connected with the power mechanism, an arrangement mechanism connected with the oscillation mechanism, a heating mechanism arranged on the arrangement mechanism and a discharging mechanism connected with the arrangement mechanism; the detection mechanism is arranged on the discharging mechanism, and the distribution mechanism is in communication connection with the detection mechanism. Wherein the arrangement mechanism is used for arranging magnetic shoes so as to enable the magnetic shoes to be uniformly heated; the detection mechanism is used for determining a magnetic shoe detection result, and the material separating mechanism is used for separating bad magnetic shoes according to the detection result of the detection mechanism. The technical problems of low manual detection efficiency and easy misjudgment are solved, and the technical effects of detecting the magnetic shoes in batches and improving the defect classification accuracy are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of magnetic tile detection, and particularly relates to a non-destructive detection device and method for magnetic tile defects. Background Art

[0002] A magnetic tile is a permanent magnet, usually in the shape of a tile, and is mainly used in permanent magnet motors. It is usually prepared from magnetic materials such as ferrite and neodymium iron boron, and is a key component for generating a magnetic field in a permanent magnet motor.

[0003] During the production process of magnetic tiles, defects may inevitably occur. Compared with qualified magnetic tiles, defective magnetic tiles will seriously reduce the magnetic properties and mechanical properties of the magnetic tiles, thus greatly affecting the service life and operating efficiency of permanent magnet motors. Therefore, during the production process, detecting and removing defective magnetic tiles is the primary task to ensure the quality of magnetic tile products.

[0004] In the prior art, the detection method for magnetic tile defects is usually to manually strike the magnetic tile with a metal block to excite sound, and to identify whether there are internal defects by the clarity and turbidity of the sound through artificial listening experience. However, manual detection not only has low efficiency, but also is prone to misjudgment during the manual detection process, thus affecting product quality. Summary of the Invention

[0005] To solve the above problems, a first aspect of the present invention provides a non-destructive detection device for magnetic tile defects, including:

[0006] A power mechanism, an oscillation mechanism connected to the power mechanism, an arrangement mechanism connected to the oscillation mechanism, a heating mechanism provided on the arrangement mechanism, a blanking mechanism connected to the arrangement mechanism, a detection mechanism provided on the blanking mechanism, and a sorting mechanism communicatively connected to the detection mechanism;

[0007] Wherein, the arrangement mechanism is used for arranging magnetic tiles so that the magnetic tiles are uniformly heated;

[0008] The detection mechanism is used for determining the magnetic tile detection result, and the sorting mechanism is used for separating defective magnetic tiles according to the detection result of the detection mechanism.

[0009] In some embodiments, the oscillation mechanism includes a driving wheel, a driven wheel drivingly connected to the driving wheel, a rotating shaft connected to the driven wheel, a cam connected to the rotating shaft, and an oscillating hopper adapted to the cam;

[0010] The cam is adapted to the oscillating hopper for oscillating the magnetic tiles in the oscillating hopper;

[0011] The oscillating hopper includes a storage bin, a sliding plate provided on the storage bin, a sliding seat slidably connected to the sliding plate, and an oscillating plate connected to the sliding plate,

[0012] The oscillating plate has a bevel structure, and the oscillating plate is adapted to the cam for lifting the storage bin to oscillate the magnetic tiles in the storage bin.

[0013] In some embodiments, the arranging mechanism includes: an arranging cylinder, a material distributing head disposed in the arranging cylinder, a material distributing column connected to the material distributing head, and a quartz sleeve sleeved on the outer periphery of the arranging cylinder;

[0014] The material distributing head is disposed in the hollow structure of the arranging cylinder. The gap between the material distributing head and the arranging cylinder is annular and is adapted to the magnetic tiles;

[0015] The material distributing column is connected to the material distributing head and is used for receiving the magnetic tiles arranged by the material distributing head and the arranging cylinder and performing secondary arrangement;

[0016] The material distributing column is columnar and has an arc-shaped tile-shaped groove on its outer periphery; the material distributing column is also connected to a power mechanism.

[0017] In some embodiments, the heating mechanism includes a heating wire wound around the outer periphery of the arranging cylinder, and the heating wire is wound on the quartz sleeve;

[0018] The blanking mechanism includes: a blanking electromagnet, a conductive slip ring connected to the blanking electromagnet, and a blanking plate disposed below the blanking electromagnet;

[0019] The blanking electromagnet is disposed in the material distributing column, and the blanking electromagnet is used for controlling the blanking speed of the magnetic tiles in the arc-shaped tile-shaped groove of the material distributing column.

[0020] In some embodiments, the detection mechanism includes a camera and an infrared thermal imager,

[0021] The camera is used for acquiring the geometric distribution characteristics of the magnetic tiles, and the infrared thermal imager is used for acquiring the temperature distribution characteristics of the magnetic tiles;

[0022] The material distributing mechanism includes a material distributing groove, a material distributing plate disposed in the material distributing groove, a connecting arm disposed on the material distributing groove and connected to the material distributing plate, a metal block disposed on the connecting arm, a material distributing electromagnet disposed on the material distributing groove and adapted to the metal block, and a tension spring disposed on the connecting arm; wherein, the material distributing groove is an inclined guiding structure;

[0023] The connecting arm, the metal block, the material distributing electromagnet and the tension spring cooperate to adjust the rotation state of the material distributing plate;

[0024] One end of the connecting arm is connected to the material distributing plate and is provided with a metal block, and one end of the tension spring is connected to the middle of the connecting arm.

[0025] The second aspect of the present invention provides a non-destructive detection method for magnetic tile defects, based on a non-destructive detection device for magnetic tile defects, including the following steps:

[0026] Obtain the temperature distribution characteristics and geometric characteristics of the magnetic tile;

[0027] The temperature distribution characteristics include: mean temperature, temperature variance, and maximum temperature gradient amplitude;

[0028] The geometric characteristics include: defect area, defect length-width ratio, and defect compactness;

[0029] Based on the temperature distribution characteristics, temperature distribution threshold set, geometric characteristics, and geometric threshold set of the magnetic tile, determine the defect state of the magnetic tile;

[0030] In response to the magnetic tile having a defect, determine the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics;

[0031] In response to the magnetic tile having a defect, determine the second defect category and classification probability of the magnetic tile based on the geometric characteristics;

[0032] Based on the first defect category and classification probability of the magnetic tile, the second defect category and classification probability of the magnetic tile, and the dynamic weight, determine the final classification probability of each defect category;

[0033] Judge the defect category based on the final classification probability of each defect category.

[0034] In some embodiments, the determining the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics includes:

[0035] Standardize the temperature distribution characteristics,

[0036] Calculate the decision value of each defect category based on the standardized temperature distribution characteristics,

[0037] The defect categories include: crack, air hole, and delamination;

[0038] Determine the classification probability of each defect category based on the decision value of each defect category,

[0039] Normalize the classification probability of each defect category.

[0040] In some embodiments, the determining the second defect category and classification probability of the magnetic tile based on the geometric characteristics includes:

[0041] Standardize the geometric characteristics,

[0042] Based on the standardized geometric characteristics and the weight matrix, determine the linear score of each defect category,

[0043] Determine the probability of each defect category based on the linear score of each defect category.

[0044] In some embodiments, the decision values for each defect category are calculated as follows:

[0045]

[0046] where f k (x new ) is the decision value, SV k is the index set of support vectors, α i,k is the Lagrange multiplier, y i,k is the class label of the support vector, is the kernel function, calculating the similarity between the support vector x i and the standardized sample , and b k is the bias term;

[0047] The method for determining the classification probability of each defect category based on the decision value of each defect category is as follows:

[0048]

[0049] where P(y = k|x new ) is the classification probability of each defect category, α k and β k are the fitting parameters of each defect category, used to adjust the shape of the Sigmoid function;

[0050] The calculation method of the linear score Z k for each defect category is as follows:

[0051] Z k = w k1 A + w k2 r + w k3 C + b k :

[0052] where [w k1 , w k2 , w k3 is the weight vector, and b k is the bias term;

[0053] The calculation method of the probability of each defect category is as follows:

[0054]

[0055] where P(y = k|x) represents the probability that the sample x belongs to the category k, is the exponential sum of all class scores.

[0056] In some embodiments, the dynamic weight is a parameter used to describe the determination of the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics, as well as the credibility of the determination of the second defect category and classification probability of the magnetic tile based on the geometric characteristics. The dynamic weight is set based on the defect category.

[0057] The calculation method for the final classification probability of each defect category is as follows:

[0058] P final = βP F (y = k|x new )+(1 - β)P(y = k|x)

[0059] where P final is the final classification probability, and β is the weight.

[0060] By adopting the above technical solutions, the present invention mainly has the following technical effects:

[0061] 1. The magnetic tiles are arranged by the arranging mechanism so that the magnetic tiles can be uniformly heated in a vertical form. The detecting mechanism determines the detection result of the magnetic tiles according to the geometric distribution characteristics and temperature distribution characteristics of the magnetic tiles, and the separating mechanism separates the defective magnetic tiles according to the detection result of the detecting mechanism, solving the technical problem of low efficiency of manual detection and realizing the technical effect of batch detection of magnetic tiles.

[0062] 2. The first defect category and classification probability of the magnetic tile are determined through the temperature distribution characteristics, and the second defect category and classification probability of the magnetic tile are determined through the geometric characteristics; according to the first defect category and classification probability of the magnetic tile, the second defect category and classification probability of the magnetic tile, and the dynamic weight, the final classification probability of each defect category is determined, thereby determining the defect category, solving the technical problem of easy misjudgment in the process of manual detection and realizing the improvement of classification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 is a schematic structural diagram of a non-destructive magnetic tile defect detection device of the present invention;

[0064] Figure 2 is a schematic structural diagram of a non-destructive magnetic tile defect detection device of the present invention (from another perspective);

[0065] Figure 3 is a schematic sectional view of a non-destructive magnetic tile defect detection device of the present invention;

[0066] Figure 4 is an exploded structural diagram of the arranging structure in a non-destructive magnetic tile defect detection device of the present invention;

[0067] Figure 5 is a schematic top view of a non-destructive magnetic tile defect detection device of the present invention;

[0068] Figure 6 This is a schematic structural diagram of the material distribution mechanism in a non-destructive detection device for magnetic tile defects according to the present invention.

[0069] Among them, the meanings of the reference numerals are as follows:

[0070] 1. Power mechanism; 11. Motor;

[0071] 2. Oscillation mechanism; 21. Driving wheel; 22. Driven wheel; 23. Rotating shaft; 24. Cam; 25. Oscillating hopper; 251. Storage bin; 252. Sliding plate; 253. Sliding seat; 254. Oscillating plate;

[0072] 3. Arrangement mechanism; 31. Arrangement cylinder; 32. Material distribution head; 33. Material distribution column; 34. Quartz sleeve;

[0073] 4. Heating mechanism; 41. Heating wire;

[0074] 5. Feeding mechanism; 51. Feeding electromagnet; 52. Conductive slip ring; 53. Feeding plate; 531. Discharge port; 532. Baffle;

[0075] 6. Detection mechanism;

[0076] 7. Material distribution mechanism; 71. Material distribution groove; 72. Material distribution plate; 73. Connecting arm; 74. Metal block; 75. Material distribution electromagnet; 76. Tension spring; 77. Limiting part; 78. Limiting plate. Detailed implementation manners

[0077] In order to enable those skilled in the art of this technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the specification drawings in the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0078] Mentioning "embodiment" in this article means that the specific features, structures or characteristics described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art understand explicitly and implicitly that the embodiments described herein can be combined with other embodiments.

[0079] Please refer to Figure 1-6, in the first aspect of the present invention, a non-destructive detection device for magnetic tile defects is provided, including: a power mechanism 1, a vibration mechanism 2 connected to the power mechanism 1, an arrangement mechanism 3 connected to the vibration mechanism 2, a heating mechanism 4 provided on the arrangement mechanism 3, a feeding mechanism 5 connected to the arrangement mechanism 3, a detection mechanism 6 provided on the feeding mechanism 5, and a material sorting mechanism 7 communicatively connected to the detection mechanism 6.

[0080] In some embodiments, the power mechanism 1 is the part of the non-destructive detection device for magnetic tile defects that generates power and transmits it to subsequent components or actuators. An exemplary first power mechanism 1 may include a motor 11, and the motor 11 can convert electrical energy into mechanical energy based on the principle of electromagnetic induction.

[0081] In some embodiments, the vibration mechanism 2 is the part of the non-destructive detection device for magnetic tile defects that vibrates the magnetic tiles, enabling the magnetic tiles to enter the arrangement mechanism 3 in a certain state. The specific vibration process will be further described below.

[0082] Furthermore, the vibration mechanism 2 includes a driving wheel 21, a driven wheel 22 in transmission connection with the driving wheel 21, a rotating shaft 23 connected to the driven wheel 22, a cam 24 connected to the rotating shaft 23, and a vibration hopper 25 adapted to the cam 24. In some embodiments, the driving wheel 21 is the wheel that actively provides power in the mechanical transmission system. The driving wheel 21 can be connected to the output shaft of the motor 11, and then by driving the motor 11, the power is transmitted to the driven wheel 22 through a belt. The driven wheel 22 is the wheel that passively receives power in the mechanical transmission system, and it is driven to rotate by the friction of the belt.

[0083] In some embodiments, one end of the rotating shaft 23 is connected to the driven wheel 22, and the other end is connected to the cam 24. When the driven wheel 22 is driven to rotate, it will drive the rotating shaft 23 and the cam 24 to rotate synchronously. In some embodiments, the cam 24 is a disk-shaped member that rotates around a fixed axis and has a variable diameter, and the cam 24 is adapted to the vibration hopper 25 for vibrating the magnetic tiles in the vibration hopper 25.

[0084] In some embodiments, the vibrating hopper 25 is used to vibrate the magnetic tiles to adjust the state of the magnetic tiles. Further, the vibrating hopper 25 includes a storage bin 251, a sliding plate 252 provided on the storage bin 251, a sliding seat 253 slidably connected to the sliding plate 252, and a vibrating plate 254 connected to the sliding plate 252. Among them, the storage bin 251 is a funnel-shaped structure that is narrow at the top and wide at the bottom, and is used to accommodate magnetic tiles to enter the magnetic tile defect non-destructive detection device for detection. The sliding plate 252 is provided at the bottom of the storage bin 251. In some embodiments, the magnetic tiles in the storage bin 251 can be vibrated by lifting and lowering the sliding plate 252 to adjust the shape of the magnetic tiles. For example, the magnetic tiles enter the arranging mechanism 3 in a vertical state.

[0085] Further, through holes are provided on the sliding plate 252, and the sliding seat 253 is a structure provided with a columnar connecting member. The columnar connecting member on the sliding seat 253 can be slidably connected to the through holes of the sliding plate 252, so that the sliding plate 252 can only be lifted and lowered in the direction set by the columnar connecting member of the sliding seat 253. In some embodiments, the vibrating plate 254 has a hypotenuse structure, and the vibrating plate 254 is adapted to the cam 24 to lift the storage bin 251 to vibrate the magnetic tiles in the storage bin 251. Specifically, after the hypotenuse structure of the vibrating plate 254 abuts against the cam 24, first, the higher end of the hypotenuse of the vibrating plate 254 abuts against the cam 24, and then as the cam 24 rotates, the cam 24 abuts against the higher end to the lower end of the hypotenuse of the vibrating plate 254 in sequence. During the abutting process, under the action of the cam 24, the vibrating plate 254 will continuously rise. When the cam 24 is separated from the vibrating plate 254, the sliding plate 252 will slide down along the columnar connecting member of the sliding seat 253 to the initial position until it is lifted again after abutting against the cam 24 next time. Thus, by rotating the cam 24, the sliding plate 252 is lifted and lowered to vibrate the magnetic tiles in the storage bin 251 to adjust the shape of the magnetic tiles.

[0086] In some embodiments, the arranging mechanism 3 is the part of the magnetic tile defect non-destructive detection device for arranging magnetic tiles, so that the magnetic tiles can be uniformly heated in a vertical state. Those skilled in the art can understand that in this embodiment, the magnetic tiles are in a sheet-like arc structure.

[0087] Further, the arranging mechanism 3 includes: an arranging cylinder 31, a distributing head 32 provided in the arranging cylinder 31, a distributing column 33 connected to the distributing head 32, and a quartz sleeve 34 sleeved on the outer periphery of the arranging cylinder 31. In some embodiments, the arranging cylinder 31 is a hollow structure, one end of the distributing head 32 is conical, and the arranging cylinder 31 cooperates with the distributing head 32 to arrange magnetic tiles. The specific arranging process will be further described below.

[0088] In some embodiments, the material distributing head 32 is disposed in the hollow structure of the arranging cylinder 31. The gap between the material distributing head 32 and the arranging cylinder 31 is annular, and this gap is adapted to the structure of the magnetic tile, such that the magnetic tile can only enter the gap between the material distributing head 32 and the arranging cylinder 31 when in a vertical state and with the arc-shaped tile fitting the annular gap, thereby arranging the magnetic tiles vertically in the gap between the material distributing head 32 and the arranging cylinder 31. On the other hand, the conical design of the material distributing head 32 can cause the magnetic tiles to continuously change direction during the oscillation process until they enter the gap between the material distributing head 32 and the arranging cylinder 31 in a vertical state.

[0089] In some embodiments, the material distributing column 33 is connected to the material distributing head 32 and is used to receive the magnetic tiles arranged by the material distributing head 32 and the arranging cylinder 31 and perform secondary arrangement. Further, the material distributing column 33 has a columnar structure with arc-shaped tile-shaped grooves provided on its outer periphery. Among them, the arc-shaped tile-shaped grooves on the material distributing column 33 are adapted to the magnetic tiles, so that the magnetic tiles arranged by the material distributing head 32 and the arranging cylinder 31 can only pass through the arc-shaped tile-shaped grooves on the material distributing column 33.

[0090] Further, the material distributing column 33 is also connected to the power mechanism 1, so as to drive the material distributing column 33 and the material distributing head 32 to rotate by using the motor 11, so that the magnetic tiles in the gap between the material distributing head 32 and the arranging cylinder 31 can enter the arc-shaped tile-shaped grooves on the material distributing column 33 for secondary arrangement through the rotation of the material distributing column 33, and through the rotation of the material distributing head 32, the magnetic tiles continuously change direction during the oscillation process.

[0091] In some embodiments, the heating mechanism 4 is the part of the magnetic tile defect non-destructive detection device for heating the magnetic tiles, so as to detect the internal defects of the magnetic tiles through the temperature distribution characteristics of the magnetic tiles. The process of detecting the internal defects of the magnetic tiles will be further described below.

[0092] Further, the heating mechanism 4 includes a heating wire 41 wound around the outer periphery of the arranging cylinder 31. The heating wire 41 is wound around the quartz sleeve 34, and the position of the quartz sleeve 34 is adapted to the position of the material distributing column 33. Thus, by using the heating effect of the current of the heating wire 41, electrical energy is converted into heat energy to heat the magnetic tiles arranged in the arranging mechanism 3 for the second time. On the other hand, the quartz sleeve 34 is used to conduct heat evenly, so that the magnetic tiles are heated more evenly, improving the heating effect for subsequent internal defect detection.

[0093] In some embodiments, the blanking mechanism 5 is the part of the magnetic tile defect non-destructive testing device for conveying magnetic tiles to the testing mechanism 6. In some embodiments, the blanking mechanism 5 includes: a blanking electromagnet 51, a conductive slip ring 52 connected to the blanking electromagnet 51, and a blanking disk 53 disposed below the blanking electromagnet 51. In some embodiments, the blanking electromagnet 51 is disposed in the material distribution column 33, and the blanking electromagnet 51 is used to control the blanking speed of the magnetic tiles in the arc-shaped tile grooves of the material distribution column 33. For example, the blanking can be stopped by the blanking electromagnet 51 adsorbing the magnetic tiles in the arc-shaped tile grooves of the material distribution column 33, and the blanking can be carried out by the blanking electromagnet 51 stopping adsorbing the magnetic tiles in the arc-shaped tile grooves of the material distribution column 33, so as to adjust the blanking speed of the magnetic tiles.

[0094] In some embodiments, the conductive slip ring 52 is a precision electromechanical component capable of transmitting electric power, electrical signals, and data signals between rotating parts and stationary parts, so as to supply power to the blanking electromagnet 51 located in the material distribution column 33 through the conductive slip ring 52.

[0095] In some embodiments, the blanking disk 53 is used to receive the magnetic tiles after blanking from the arranging mechanism 3 and convey them to the lower part of the testing mechanism 6 to detect the internal defects of the magnetic tiles. Further, the blanking disk 53 is connected to the power mechanism 1. Driven by the motor 11, the blanking disk 53 can rotate. The blanking disk 53 is provided with a discharge port 531 and a baffle 532. Under the action of the baffle 532, the rotating blanking disk 53 will drive the magnetic tiles to the discharge port 531 for detection.

[0096] In some embodiments, the testing mechanism 6 is the part of the magnetic tile defect non-destructive testing device for testing magnetic tiles. In some embodiments, the testing mechanism 6 includes a camera and an infrared thermal imager. Among them, the camera is used to obtain the geometric distribution characteristics of the magnetic tiles, and the infrared thermal imager is used to obtain the temperature distribution characteristics of the magnetic tiles. The testing mechanism 6 is used to determine the magnetic tile testing result according to the geometric distribution characteristics and temperature distribution characteristics of the magnetic tiles. In some embodiments, the magnetic tile testing result can include good products and defective magnetic tiles. Exemplary defective magnetic tiles include cracks and pores, etc.

[0097] In some embodiments, the material distribution mechanism 7 is a part of the magnetic tile defect non-destructive detection device for separating defective magnetic tiles according to the detection results of the detection mechanism 6. The material distribution mechanism 7 includes a material distribution groove 71, a material distribution plate 72 disposed in the material distribution groove 71, a connecting arm 73 disposed on the material distribution groove 71 and connected to the material distribution plate 72, a metal block 74 disposed on the connecting arm 73, a material distribution electromagnet 75 disposed on the material distribution groove 71 and adapted to the metal block 74, and a tension spring 76 disposed on the connecting arm 73. Among them, the material distribution groove 71 is an inclined guiding structure for guiding the magnetic tiles after detection into a container.

[0098] In some embodiments, the material distribution plate 72 is disposed on the bottom plate of the material distribution groove 71 and hinged to the bottom plate of the material distribution groove 71, and is used to separate defective magnetic tiles according to the detection results of the detection mechanism 6. For example, when the detection mechanism 6 detects that the magnetic tile is a good product, the material distribution plate 72 and the material distribution groove 71 are in a closed state, and the magnetic tile is guided into the storage container along the setting direction of the material distribution groove 71. When the detection mechanism 6 detects that the magnetic tile is a defective product, the material distribution plate 72 can rotate, and the defective magnetic tile is guided into the recycling container along the direction after rotation by using the material distribution plate 72, so as to realize batch detection of magnetic tiles and improve the detection effect.

[0099] In some embodiments, the connecting arm 73, the metal block 74, the material distribution electromagnet 75 and the tension spring 76 cooperate to adjust the rotation state of the material distribution plate 72. Specifically, one end of the connecting arm 73 is connected to the material distribution plate 72 and is provided with a metal block 74, one end of the tension spring 76 is connected to the middle of the connecting arm 73. When the material distribution electromagnet 75 is not working, due to the limiting force of the tension spring 76, the material distribution plate 72 and the material distribution groove 71 are in a closed state. When the material distribution electromagnet 75 works, the metal block 74 will approach the material distribution electromagnet 75 against the action of the tension spring 76, thereby driving one end of the connecting arm 73 connected to the material distribution plate 72 and the material distribution plate 72 to rotate. Then, the defective magnetic tile is guided into the recycling container through the material distribution plate 72. Finally, as the material distribution electromagnet 75 stops working, the restoring force of the tension spring 76 will drive one end of the connecting arm 73 connected to the material distribution plate 72 to rotate, so that the material distribution plate 72 rotates to a closed state with the material distribution groove 71.

[0100] In some embodiments, a limiting member 77 is further provided at one end of the connecting arm 73 away from the metal block 74, and a limiting plate 78 adapted to the limiting member 77 is provided on the material distribution groove 71. The limiting plate 78 and the limiting member 77 cooperate to limit the rotation angle of the material distribution plate 72.

[0101] The second aspect of the present invention provides a method for non-destructively detecting magnetic tile defects, including the following steps:

[0102] (a) Obtain the temperature distribution characteristics and geometric characteristics of the magnetic tile;

[0103] In some embodiments, the temperature distribution characteristics are used to reflect the abnormal heat conduction inside the magnetic tile, and the geometric characteristics are used to describe the shape and position of the defect.

[0104] In some embodiments, the temperature distribution characteristics can be obtained from a temperature distribution image. An exemplary temperature distribution image can be acquired by an infrared thermal imager, which is usually a two-dimensional grayscale image. Each pixel value represents the temperature at that location, and then the temperature distribution characteristics are determined based on the temperature distribution image.

[0105] In some embodiments, the temperature distribution characteristics include:

[0106] Mean temperature,

[0107] In some embodiments, the mean temperature is used to reflect the overall thermal state of the magnetic tile and to detect whether there are pore defects inside the magnetic tile. Pore defects often cause a local temperature increase in the magnetic tile.

[0108] In some embodiments, the mean temperature of the magnetic tile is calculated as follows:

[0109]

[0110] where μ T is the mean temperature, N is the total number of temperature measurement points, and T i is the temperature value at the i-th temperature measurement point.

[0111] Temperature variance,

[0112] In some embodiments, the temperature variance is used to reflect the uniformity of the temperature distribution. The temperature variance is used to detect whether there are crack regions inside the magnetic tile. In crack regions, due to the hindrance of heat conduction, the temperature gradient is large and the variance is relatively high.

[0113] In some embodiments, the temperature variance of the magnetic tile is calculated as follows:

[0114]

[0115] where is the temperature variance, μ T is the mean temperature, N is the total number of temperature measurement points, and T i is the temperature value at the i-th temperature measurement point.

[0116] Magnitude of the maximum temperature gradient,

[0117] In some embodiments, the maximum temperature gradient amplitude is used to reflect the position where the temperature of the magnetic tile changes most violently. Specifically, the heat conduction in the crack region is blocked, resulting in a violent local temperature change. Therefore, the crack edge usually has a large temperature gradient, and the temperature change in the center region of the pore is relatively gentle, so the temperature gradient at the center of the pore is small.

[0118] In some embodiments, the calculation method of the maximum temperature gradient amplitude is as follows:

[0119]

[0120] where is the maximum temperature gradient amplitude, is the partial derivative of temperature in the x direction, is the partial derivative of temperature in the y direction.

[0121] In some embodiments, the geometric features can be obtained from visible light images, and exemplary visible light images can be acquired by an industrial camera.

[0122] In some embodiments, the geometric features include:

[0123] defect area,

[0124] In some embodiments, the defect area refers to the number of pixels in the defect region and is used to reflect the size of the defect on the surface of the magnetic tile.

[0125] In some embodiments, the defect boundary of the magnetic tile can be extracted by the Canny operator, and then the geometric features can be determined. The exemplary code for implementing the Canny operator to extract the defect boundary in Python can be:

[0126] import cv2

[0127] import numpy as np

[0128] import matplotlib.pyplot as plt

[0129] # Read the image

[0130] image=cv2.imread('magnet_wafer.jpg',0)

[0131] # Gaussian filtering for denoising

[0132] blurred=cv2.GaussianBlur(image,(5,5),0)

[0133] # Canny edge detection

[0134] edges = cv2.Canny(blurred, 100, 200)

[0135] # Display the result

[0136] plt.figure(figsize=(10, 5))

[0137] plt.subplot(1, 2, 1)

[0138] plt.imshow(image, cmap='gray')

[0139] plt.title('Original Image')

[0140] plt.subplot(1, 2, 2)

[0141] plt.imshow(edges, cmap='gray')

[0142] plt.title('Canny Edge Detection')

[0143] plt.show()

[0144] In some embodiments, the calculation method of the defect size is as follows:

[0145] A = Σ (x，y) ∈R 1

[0146] where A is the defect area, (x, y) is the coordinate of a pixel point in the image, R is the defect region, and 1 represents the counting unit of each pixel point.

[0147] Defect aspect ratio,

[0148] In some embodiments, the defect aspect ratio refers to comparing the width and height of the bounding box of the defect region to describe the shape characteristics of the defect. Exemplary crack defects are slender, with their width much smaller than their height, so the aspect ratio is much greater than 1. Exemplary pore defects are close to circular, with their width and height being similar, so the aspect ratio is close to 1.

[0149] In some embodiments, the calculation method of the defect aspect ratio is as follows:

[0150]

[0151] where r is the defect aspect ratio, w is the width of the bounding box, and h is the height of the bounding box.

[0152] Defect perimeter,

[0153] In some embodiments, the defect perimeter refers to the total length of the defect boundary and is used to describe the complexity of the shape of the magnetic tile defect.

[0154] In some embodiments, the defect perimeter is calculated as follows:

[0155]

[0156] where n is the number of pixel points on the boundary, (x i , y i ) is the coordinate of the i-th pixel point on the boundary, and (x i+1 , y i+1 ) is the coordinate of the (i + 1)-th pixel point on the boundary;

[0157] Defect compactness,

[0158] In some embodiments, the defect compactness refers to comparing the area and perimeter of the defect region and is used to describe the compactness of the defect shape. Exemplary pore defects are close to circular, so their compactness values are relatively high. Exemplary crack defects are long and irregular, so their compactness values are relatively low.

[0159] In some embodiments, the defect compactness is calculated as follows:

[0160]

[0161] where C is the compactness, A is the area of the defect region, and P is the perimeter of the defect region.

[0162] In some embodiments, after obtaining the temperature distribution characteristics and geometric characteristics of the magnetic tile, the parameters in the temperature distribution characteristics and geometric characteristics can be standardized by means of Z-score standardization to eliminate the dimensional differences between different characteristics.

[0163] (b) Based on the temperature distribution characteristics of the magnetic tile, the temperature distribution threshold set, the geometric characteristics, and the geometric threshold set, determine the defect state of the magnetic tile;

[0164] In some embodiments, after obtaining the temperature distribution characteristics and geometric characteristics of the magnetic tile, it can be determined whether the magnetic tile has defects according to the temperature distribution characteristics and geometric characteristics of the magnetic tile.

[0165] In some embodiments, the temperature distribution threshold set refers to the set of thresholds used to determine whether the magnetic tile has defects. In some embodiments, the temperature distribution threshold set includes: the mean temperature threshold, the temperature variance threshold, and the maximum temperature gradient amplitude threshold.

[0166] In some embodiments, the defect state of the magnetic tile can be confirmed by comparing the temperature distribution characteristics with a set of temperature distribution thresholds. For example, comparing the mean temperature with the mean temperature threshold, the temperature variance with the temperature variance threshold, and the maximum temperature gradient amplitude with the maximum temperature gradient amplitude threshold. When one or more of the data exceed the thresholds in the temperature distribution threshold set, it is determined that the magnetic tile has a defect.

[0167] In some embodiments, the mean temperature threshold, the temperature variance threshold, and the maximum temperature gradient amplitude threshold can be set based on a historical database and production standards.

[0168] In some embodiments, the geometric threshold set refers to a set of thresholds used to determine whether there are defects in the magnetic tile. In some embodiments, the geometric threshold set includes: a defect area threshold, a defect aspect ratio threshold, and a defect compactness threshold.

[0169] In some embodiments, the defect state of the magnetic tile can be confirmed by comparing geometric features with a geometric threshold set. For example, comparing the defect area with the defect area threshold, the defect aspect ratio with the defect aspect ratio threshold, and the defect compactness with the defect compactness threshold. When one or more of the data exceed the thresholds in the temperature distribution threshold set, it is determined that the magnetic tile has a defect.

[0170] In some embodiments, the defect area threshold, the defect aspect ratio threshold, and the defect compactness threshold can be set based on a historical database and production standards.

[0171] (c) In response to the magnetic tile having a defect, determining a first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics;

[0172] In some embodiments, the temperature distribution characteristics of the magnetic tile can be classified by a temperature feature classifier to determine the defect category and classification probability of the magnetic tile.

[0173] In some embodiments, the temperature features can be analyzed by a support vector machine (SVM) model, and an optimal hyperplane can be constructed to achieve defect classification.

[0174] In some embodiments, the support vector machine (SVM) model can standardize the temperature distribution characteristics and then calculate the decision value, and then determine the defect category and classification probability of the sample according to the decision value.

[0175] In some embodiments, the temperature distribution characteristics include: mean temperature, temperature variance, and maximum temperature gradient amplitude.

[0176] In some embodiments, the determining the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics includes:

[0177] Standardizing the temperature distribution characteristics,

[0178] In some embodiments, the temperature distribution feature can be expressed as:

[0179]

[0180] Normalized temperature distribution feature:

[0181] where, is the sample after normalization, μ is the mean of the training data, and σ is the standard deviation of the training data;

[0182] Calculate the decision values for each defect category based on the normalized temperature distribution feature,

[0183] In some embodiments, the defect categories may include: cracks, pores, and delamination.

[0184] In some embodiments, the calculation method of the decision values for each defect category is:

[0185]

[0186] where, f k (x new ) is the decision value, SV k is the index set of the support vectors, α i，k is the Lagrange multiplier, y i，k is the class label of the support vector, is the kernel function, calculating the similarity between the support vector x i and the normalized sample , and b k is the bias term;

[0187] Determine the classification probabilities for each defect category based on the decision values for each defect category,

[0188] In some embodiments, the method for determining the classification probabilities for each defect category based on the decision values for each defect category is:

[0189]

[0190] where, P(y = k|x new ) is the classification probability for each defect category, α k and β k are the fitting parameters for each defect category, used to adjust the shape of the Sigmoid function.

[0191] Normalize the classification probabilities for each defect category,

[0192] In some embodiments, the method for normalizing the classification probabilities for each defect category is:

[0193]

[0194] Among them, P F (y = k|x new ) is the normalized classification probability, and j is the number of defect categories.

[0195] In some embodiments, a support vector machine (SVM) model can be obtained through training. For example, the support vector machine (SVM) model can be trained based on a large number of labeled training samples. The training samples can include temperature distribution feature matrices. The labels can be the corresponding defect categories. The labels can be obtained through manual annotation. Input the temperature distribution feature matrices in the training samples into the support vector machine (SVM) model; obtain the defect categories output by the support vector machine (SVM) model. Construct a loss function based on the labeled temperature distribution feature matrices and the defect categories output by the support vector machine (SVM) model, and synchronously update the parameters of the support vector machine (SVM) model. Through parameter update, a trained support vector machine (SVM) model is obtained.

[0196] Exemplarily, let the temperature distribution feature μ T = 85°C, σ T = 12°C, The temperature distribution feature after standardization is: μ T = 0.5, σ T = 1.2,

[0197] Exemplarily, the parameters of the trained model and the decision values of each defect category after calculation are shown in the following table:

[0198]

[0199] Exemplarily, the classification probabilities of each defect category after calculation are shown in the following table:

[0200] Category <![CDATA[α k > <![CDATA[β k > <![CDATA[P(y=k|x new )]]> <![CDATA[P F (y = k|x new )]]> Crack -0.5 0.2 0.65 0.48 Porosity 0.3 -0.1 0.38 0.28 Delamination -0.2 0.5 0.33 0.24

[0201] (d) In response to the magnetic tile having a defect, determine the second defect category and classification probability of the magnetic tile based on geometric features;

[0202] In some embodiments, the defect category and probability of the magnetic tile can be determined by classifying the temperature distribution feature of the magnetic tile through a geometric feature classifier.

[0203] In some embodiments, the relationship between features and category probabilities can be established through a logistic regression model, and the geometric features can be used to classify the defect categories of the magnetic tiles.

[0204] In some embodiments, the defect category and probability of the magnetic tile can be determined by the defect area, defect length-width ratio, and defect compactness of the magnetic tile.

[0205] In some embodiments, determining the second defect category and classification probability of the magnetic tile based on geometric features includes:

[0206] In some embodiments, the geometric features can be represented as:

[0207] x = [A, r, C]

[0208] Normalize the geometric features,

[0209] Based on the normalized geometric features and the weight matrix, determine the linear scores of each defect category.

[0210] In some embodiments, the weight matrix indicates that each defect category corresponds to a weight vector w k = [w k1 , w k2 , w k3 and a bias term b k .

[0211] In some embodiments, the linear score Z k of each defect category is calculated as follows:

[0212] Z k = w k1 A + w k2 r + w k3 C + b k

[0213] where [w k1 , w k2 , w k3 is the weight vector and b k is the bias term;

[0214] Determine the probability of each defect category based on the linear scores of each defect category.

[0215] In some embodiments, the probability of each defect category is calculated as follows:

[0216]

[0217] where P(y = k|x) represents the probability that the sample x belongs to the category k, is the sum of the exponents of all category scores, used for normalization to ensure that the sum of the probability distribution is 1.

[0218] Exemplarily, let the geometric features be A = 120mm 2 , r = 0.6, C = 0.4, and the normalized geometric features be A = 1.5, r = -0.3, C = -0.2.

[0219] Exemplarily, the trained weights and bias terms are shown in the following table:

[0220] Category <![CDATA[w k1 (Defect area)]]> <![CDATA[w k2 (Aspect ratio)]]> <![CDATA[w k3 (Compactness)]]> <![CDATA[b k (Bias term)]]> Crack 1.8 0.5 -2.0 -1.2 Porosity -0.5 -1.2 3.0 0.8 Delamination 0.7 1.0 -0.5 -0.3

[0221] Then the linear scores of each defect category are as follows:

[0222] Z(Crack) = 1.8×1.5 + 0.5×(-0.3) + (-2.0)×(-0.2) - 1.2 == 1.75;

[0223] Z(Pore) = -0.5×1.5 + (-1.2)×(-0.3) + 3.0×(-0.2) + 0.8 == -0.19;

[0224] Z(Delamination) = 0.7×1.5 + 1.0×(-0.3) + (-0.5)×(-0.2) - 0.3 = 0.55;

[0225] Calculate the probabilities based on the linear scores of each defect category:

[0226] P(Crack) = 5.75 / 8.321 ≈ 0.692;

[0227] P(Pore) = 0.85 / 8.321 ≈ 0.100;

[0228] P(Delamination) = 1.73 / 8.321 ≈ 0.208.

[0229] (e) Based on the first defect category and classification probability of the magnetic tile, the second defect category and classification probability of the magnetic tile, and the dynamic weight, determine the final classification probability of each defect category;

[0230] In some embodiments, the third defect category and classification probability of the magnetic tile can be determined by means of weighted fusion through the first defect category and classification probability of the magnetic tile and the second defect category of the magnetic tile.

[0231] In some embodiments, the dynamic weight is a parameter used to describe the credibility of determining the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics and the second defect category and classification probability of the magnetic tile based on the geometric characteristics. In some embodiments, the dynamic weight is set based on the defect category. Exemplarily, the temperature feature classifier is sensitive to cracks but prone to misjudging pores, and the geometric classifier is sensitive to pores but prone to missing small cracks. Therefore, when the defect is classified as a crack, the weight of the classification result of the temperature feature classifier can be increased and the weight of the classification result of the geometric classifier can be reduced. When the defect is classified as a pore, the weight of the classification result of the temperature feature classifier can be reduced and the weight of the classification result of the geometric classifier can be increased to improve the defect classification accuracy.

[0232] In some embodiments, the final classification probability of each defect category is calculated as follows:

[0233] P final = βP F (y = k|x new )+(1 - β)P(y = k|x)

[0234] where P final is the final classification probability, and β is the weight.

[0235] Exemplarily, the final classification probability of the crack defect is: 0.48×0.8 + 0.692×0.2 = 0.5224;

[0236] The final classification probability of the pore defect is: 0.28×0.2 + 0.1×0.8 = 0.136.

[0237] (f) Determine the defect category based on the final classification probability of each defect category;

[0238] In some embodiments, after obtaining the final classification probability of each defect category, the defect category with the highest probability can be selected from the final classification probabilities of each defect category as the detection result of the magnetic tile and its defect category can be determined.

[0239] Exemplarily, since 0.5224 is greater than 0.136 and is the maximum value among the final classification probabilities of each defect category, it is determined that the detection result of the current magnetic tile is a crack defect.

[0240] Finally, it should be noted that: What is disclosed in the embodiments of the present invention is only the preferred embodiments of the present invention, which are only used to illustrate the technical solutions of the present invention and are not intended to limit them; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A non-destructive detection device for magnetic tile defects, characterized in that, include: A power mechanism, an oscillating mechanism connected to the power mechanism, an arranging mechanism connected to the oscillating mechanism, a heating mechanism provided on the arranging mechanism, a material discharging mechanism connected to the arranging mechanism, a detection mechanism provided on the material discharging mechanism, and a material distributing mechanism communicatively connected to the detection mechanism; Wherein, the arrangement mechanism is used to arrange the magnetic tiles so that the magnetic tiles are heated evenly; The detection mechanism is used to determine the detection result of the magnetic tile, and the material separation mechanism is used to separate the defective magnetic tile according to the detection result of the detection mechanism.

2. The non-destructive detection device for magnetic tile defects according to claim 1, wherein, The oscillating mechanism includes a driving wheel, a driven wheel drivingly connected to the driving wheel, a rotating shaft connected to the driven wheel, a cam connected to the rotating shaft, and an oscillating hopper adapted to the cam; The cam is adapted to the vibration hopper and is used to vibrate the magnetic tiles in the vibration hopper; The oscillating hopper comprises a storage bin, a sliding plate arranged on the storage bin, a sliding seat slidably connected to the sliding plate, and an oscillating plate connected to the sliding plate. The oscillating plate has a bevel structure, and the oscillating plate is matched with a cam and is used for lifting and lowering the storage bin to oscillate the magnetic tiles in the storage bin.

3. The non-destructive detection device for magnetic tile defects according to claim 1, characterized in that, The arrangement mechanism comprises: an arrangement tube, a material distribution head arranged in the arrangement tube, a material distribution column connected to the material distribution head, and a quartz sleeve sleeved on the outer periphery of the arrangement tube; The material distribution head is arranged in the hollow structure of the arrangement cylinder, and the gap between the material distribution head and the arrangement cylinder is annular, and the gap is adapted to the magnetic tile; The material distribution column is connected to the material distribution head and is used to receive the magnetic tiles arranged by the material distribution head and the arrangement cylinder and perform secondary arrangement; The material distribution column is columnar and has an arc-shaped tile-shaped groove on its outer circumference; the material distribution column is also connected to a power mechanism.

4. A nondestructive testing device for magnetic tile defects according to claim 3, characterized in that: The heating mechanism comprises a heating wire wound on the outer circumference of the arrangement tube, and the heating wire is wound on a quartz sleeve; The material unloading mechanism comprises: a material unloading electromagnet, a conductive slip ring connected to the material unloading electromagnet, and a material unloading plate arranged below the material unloading electromagnet; The material discharge electromagnet is arranged in the material distribution column, and the material discharge electromagnet is used to control the material discharge speed of the magnetic tile in the arc-shaped tile-shaped groove of the material distribution column.

5. The non-destructive detection device for magnetic tile defects according to claim 1, characterized in that The detection mechanism includes a camera and an infrared thermal imager. The camera is used to obtain the geometric distribution characteristics of the magnetic tile, and the infrared thermal imager is used to obtain the temperature distribution characteristics of the magnetic tile; The material distributing mechanism comprises a material distributing trough, a material distributing plate arranged in the material distributing trough, a connecting arm arranged on the material distributing trough and connected to the material distributing plate, a metal block arranged on the connecting arm, a material distributing electromagnet arranged on the material distributing trough and matched with the metal block, and a tension spring arranged on the connecting arm; wherein the material distributing trough is a material guiding structure arranged obliquely; The connecting arm, the metal block, the material distribution electromagnet and the tension spring cooperate with each other to adjust the rotation state of the material distribution plate; One end of the connecting arm is connected to the material dividing plate and is provided with a metal block, and one end of the tension spring is connected to the middle part of the connecting arm.

6. A non-destructive detection method for magnetic tile defects, characterized in that, A nondestructive detection device for magnetic tile defects includes the following steps: Obtain temperature distribution characteristics and geometric characteristics of magnetic tiles; The temperature distribution characteristics include: mean temperature, temperature variance, and the magnitude of the maximum temperature gradient; The geometric characteristics include: defect area, defect length-width ratio, and defect compactness; Based on the temperature distribution characteristics of the magnetic tile, the temperature distribution threshold set, the geometric characteristics, and the geometric threshold set, determine the defect state of the magnetic tile; In response to the magnetic tile having a defect, determine the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics; In response to the magnetic tile having a defect, determine the second defect category and classification probability of the magnetic tile based on the geometric characteristics; Based on the first defect category and classification probability of the magnetic tile, the second defect category and classification probability of the magnetic tile, and the dynamic weight, determine the final classification probability of each defect category; Based on the final classification probability of each defect category, determine the defect category; 7. A non-destructive detection method for magnetic tile defects according to claim 6, characterized in that, The determining the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics includes: Normalize the temperature distribution characteristics; Based on the normalized temperature distribution characteristics, calculate the decision values of each defect category; The defect categories include: crack, pore, and delamination; Based on the decision values of each defect category, determine the classification probabilities of each defect category; Normalize the classification probabilities of each defect category.

8. The non-destructive detection method for magnetic tile defects according to claim 7, wherein The determining the second defect category and classification probability of the magnetic tile based on the geometric characteristics includes: Normalize the geometric characteristics; Based on the normalized geometric characteristics and the weight matrix, determine the linear scores of each defect category; Based on the linear scores of each defect category, determine the probabilities of each defect category.

9. The non-destructive detection method for magnetic tile defects according to claim 8, wherein The calculation method of the decision value of each defect category is: where f k (x new ) is the decision value, SV k is the index set of support vectors, α i,k is the Lagrange multiplier, y i,k is the class label of the support vector, is the kernel function that calculates the similarity between the support vector x i and the standardized sample , and b k is the bias term; The method of determining the classification probability of each defect category based on the decision value of each defect category is: Among them, P(y = k|x new ) is the classification probability of each defect category, α k and β k are the fitting parameters of each defect category, which are used to adjust the shape of the Sigmoid function; The linear score Z for each defect category k is calculated as follows: Z k = w k1 A + w k2 r + w k3 C + b k : Among them, [w k1 , w k2 , w k3 is the weight vector, and b k is the bias term; The calculation method of the probability of each defect category is: Among them, P(y = k|x) represents the probability that the sample x belongs to the category k, which is the exponential sum of all category scores.

10. The non-destructive detection method for magnetic tile defects according to claim 9, wherein The dynamic weight is a parameter used to describe the credibility of determining the first defect category and classification probability of the magnetic tile based on the temperature distribution characteristics and determining the second defect category and classification probability of the magnetic tile based on the geometric characteristics. The dynamic weight is set based on the defect category; The calculation method of the final classification probability of each defect category is: P final = βP F (y = k|x new )+(1 - β)P(y = k|x) Among them, P final is the final classification probability, and β is the weight.

Images