CN117974666A - Quality anomaly detection method for non-circular planetary gear - Google Patents
Quality anomaly detection method for non-circular planetary gear Download PDFInfo
- Publication number
- CN117974666A CN117974666A CN202410385905.9A CN202410385905A CN117974666A CN 117974666 A CN117974666 A CN 117974666A CN 202410385905 A CN202410385905 A CN 202410385905A CN 117974666 A CN117974666 A CN 117974666A
- Authority
- CN
- China
- Prior art keywords
- suspected
- pitting
- area
- characteristic value
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title abstract description 15
- 230000002159 abnormal effect Effects 0.000 claims abstract description 53
- 230000008859 change Effects 0.000 claims abstract description 44
- 238000000034 method Methods 0.000 claims abstract description 24
- 238000009826 distribution Methods 0.000 claims abstract description 19
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 9
- 230000000877 morphologic effect Effects 0.000 claims abstract description 8
- 230000005856 abnormality Effects 0.000 claims abstract description 7
- 238000013507 mapping Methods 0.000 claims description 24
- 238000012545 processing Methods 0.000 abstract description 4
- 239000000428 dust Substances 0.000 description 20
- 230000001788 irregular Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000007797 corrosion Effects 0.000 description 3
- 238000005260 corrosion Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000003708 edge detection Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000009776 industrial production Methods 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000010419 fine particle Substances 0.000 description 1
- 239000004519 grease Substances 0.000 description 1
- 239000000314 lubricant Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image data processing, in particular to a quality abnormality detection method of a non-circular planetary gear; acquiring a suspected abnormal region of the appearance image according to the convolutional neural network; obtaining a suspected pitting area according to the appearance image and the gray level difference characteristic of the suspected abnormal area, and obtaining a shape characteristic value of the suspected pitting area according to the morphological characteristic of the suspected pitting area and the change characteristic of the edge contour; and obtaining the depth characteristic value of the suspected pitting area according to the gray level distribution characteristic and the gradient distribution characteristic of the suspected pitting area. The method comprises the steps of obtaining a target pitting index according to a shape characteristic value and a depth characteristic value; the target pitting area of the non-circular planetary gear is obtained according to the target pitting index of the suspected pitting area, so that the situation that all suspected abnormal areas are regarded as pitting areas by an image vision algorithm is avoided, and the detection accuracy of the pitting condition of the non-circular planetary gear is improved.
Description
Technical Field
The invention relates to the technical field of image data processing, in particular to a quality abnormality detection method of a non-circular planetary gear.
Background
The non-circular planetary gear surface quality directly affects the efficiency and performance of the drive train, and poor surface quality can increase frictional losses and reduce drive efficiency. Pitting is a corrosion form concentrated on a small range of the metal surface and penetrating into the metal, and the common gear pitting can cause irregular damage to the gear surface, increase friction and noise level, reduce transmission efficiency and cause mechanical faults, so that the surface of a non-circular planetary gear needs to be detected regularly in the industrial production process, and the normal operation of a transmission system is ensured.
In the traditional mode, defects and anomalies on the surface of gears in industrial production are generally identified by using an image vision technology, and the detection of the targets of the pitting areas is realized by acquiring and training the characteristics of the communication areas and the gray level change characteristics of a large number of images of the pitting areas. Since the non-circular planetary gear generates large static electricity and friction force during operation, dust particles in the air are attracted to the gear surface, and grease or lubricant which does not clean the gear surface increases the degree of dust adhesion. Since the feature of dust collection on the gear surface in image processing is similar to the pitting feature, the machine learning is caused to consider the area of dust collection on the gear surface as the pitting area; so that the identification of the pitting areas on the gear surface is inaccurate through the image vision technology.
Disclosure of Invention
In order to solve the technical problem that the image vision technology is inaccurate in identifying the pitting area of the gear surface due to the aggregation of dust on the gear surface, the invention aims to provide a quality anomaly detection method of a non-circular planetary gear, and the adopted technical scheme is as follows:
obtaining an appearance image for detecting the surface quality of a non-circular planetary gear, and obtaining a suspected abnormal region of the appearance image according to a convolutional neural network;
Obtaining a suspected pitting area according to the appearance image and the gray level difference characteristic of the suspected abnormal area; obtaining a first rule according to morphological characteristics of the suspected pitting area; obtaining a second regularity according to the change characteristics of the edge profile of the suspected pitting area; obtaining a shape characteristic value of the suspected pitting area according to the first rule degree and the second rule degree;
Obtaining a depth difference characteristic value according to the gray level distribution characteristics of the suspected pitting area; obtaining a depth change characteristic value according to gradient distribution characteristics of the suspected pitting area; obtaining a depth characteristic value of the suspected pitting area according to the depth difference characteristic value and the depth change characteristic value;
Obtaining a target pitting index according to the shape characteristic value and the depth characteristic value; and obtaining a target pitting area of the non-circular planetary gear according to the target pitting index of the suspected pitting area.
Further, the step of obtaining the suspected pitting corrosion area according to the appearance image and the gray level difference characteristic of the suspected abnormal area comprises the following steps:
Calculating the average value of gray values of any suspected abnormal region, and obtaining the region gray average value of the any suspected abnormal region; calculating the average value of gray values of the appearance image to obtain the overall appearance gray average value; calculating and normalizing the absolute value of the difference value of the regional gray average value and the overall appearance gray average value to obtain an abnormal characteristic value of the arbitrary suspected abnormal region; and when the abnormal characteristic value exceeds a preset abnormal threshold value, the any suspected abnormal region is a suspected pitting region.
Further, the step of obtaining the first regularity according to the morphological characteristics of the suspected pitting area includes:
and calculating the fractal dimension value of the suspected pitting area and carrying out negative correlation mapping to obtain a first regularity of the suspected pitting area.
Further, the step of obtaining the second regularity according to the change characteristics of the edge profile of the suspected pitting area includes:
Calculating the variance of the curvature of the edge pixel points of the suspected pitting area and carrying out negative correlation mapping to obtain a first contour change characteristic value of the suspected pitting area; counting the number value of the curvature maximum value of the edge pixel point and carrying out negative correlation mapping to obtain a second contour change characteristic value; and calculating the average value of the first contour change characteristic value and the second contour change characteristic value to obtain the second regularity of the suspected pitting area.
Further, the step of obtaining the shape characteristic value of the suspected pitting area according to the first rule degree and the second rule degree comprises the following steps:
calculating the product of a preset first weight and the first regularity to obtain a first shape characteristic weight; calculating the product of a preset second weight and the second regularity to obtain a second shape characteristic weight; and calculating the sum value of the first shape feature weight and the second shape feature weight to obtain the shape feature value of the suspected pitting area.
Further, the step of obtaining the depth difference characteristic value according to the gray level distribution characteristic of the suspected pitting area comprises the following steps:
And acquiring a gray level histogram of the suspected pitting area, calculating the ratio of the vertical axis extremum difference and the horizontal axis extremum difference of the gray level histogram, and normalizing to obtain the depth difference characteristic value of the suspected pitting area.
Further, the step of obtaining the depth variation characteristic value according to the gradient distribution characteristic of the suspected pitting area comprises the following steps:
and calculating the average value of the gradients of the pixel points in the suspected pitting area and carrying out negative correlation mapping to obtain the depth change characteristic value of the suspected pitting area.
Further, the step of obtaining the depth characteristic value of the suspected pitting area according to the depth difference characteristic value and the depth variation characteristic value comprises the following steps:
Calculating the product of a preset third weight and a depth difference characteristic value to obtain a first depth characteristic weight; calculating the product of a preset fourth weight and a depth change characteristic value to obtain a second depth characteristic weight; and calculating the sum value of the first depth feature weight and the second depth feature weight to obtain the depth feature value of the suspected pitting area.
Further, the step of obtaining a target pitting index from the shape feature value and the depth feature value includes:
And mapping the shape characteristic value in a negative correlation way to obtain a shape characteristic mapping value, and calculating the average value of the shape characteristic mapping value and the depth characteristic value to obtain the target pitting index of the suspected pitting area.
Further, the step of obtaining the target pitting area of the non-circular planetary gear according to the target pitting index of the suspected pitting area comprises the following steps:
and when the target pitting index of the suspected pitting area exceeds a preset index threshold value, the suspected pitting area is the target pitting area.
The invention has the following beneficial effects:
According to the method, the suspected abnormal region can be obtained, the region which is likely to appear pitting can be marked out efficiently and rapidly through an image visual algorithm, and the detection efficiency is improved; the suspected pitting area can be obtained, the suspected pitting area can be initially determined according to the gray level difference characteristics, the error of the image visual algorithm mark is reduced, and the accuracy of obtaining the pitting area is improved. The shape characteristic values of the suspected pitting areas are obtained according to the first regularity and the second regularity, the distinction can be made based on the difference of the area shape characteristic of the real pitting areas and the dust collecting areas, and the accuracy of obtaining the pitting areas is improved. The depth characteristic value of the suspected pitting area is obtained according to the depth difference characteristic value and the depth change characteristic value, the difference of the internal depth difference characteristics of the real pitting area and the dust collecting area can be distinguished, and the accuracy of obtaining the pitting area is improved. And finally, acquiring a target pitting area of the non-circular planetary gear according to the target pitting index, so that the situation that all suspected abnormal areas are regarded as pitting areas by an image vision algorithm can be avoided, and the detection accuracy of the pitting condition of the non-circular planetary gear is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a quality anomaly detection method for a non-circular planetary gear according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description refers to the specific implementation, structure, characteristics and effects of a method for detecting quality anomalies of a non-circular planetary gear according to the present invention, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the quality abnormality detection method for a non-circular planetary gear provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of a method for detecting quality anomalies of a non-circular planetary gear according to an embodiment of the invention is shown, the method includes the following steps:
Step S1, obtaining an appearance image for detecting the surface quality of the non-circular planetary gear, and obtaining a suspected abnormal area of the appearance image according to the convolutional neural network.
In the embodiment of the invention, the implementation scene is to detect the surface quality of the non-circular planetary gear; firstly, obtaining an appearance image of the surface of the non-circular planetary gear, shooting the image of the surface of the non-circular planetary gear at a proper position by an industrial camera, and carrying out denoising and enhancing pretreatment on the image to obtain the appearance image. Extracting the characteristic of the suspected pitting abnormal region on the gear surface from the appearance image by using an edge detection algorithm, and training and learning an identification model of the characteristic of the suspected pitting abnormal region by using a CNN convolutional neural network algorithm, so that the identification model can rapidly identify the suspected abnormal region on the gear surface; the suspected abnormal region characterizes a region of the non-circular planetary gear surface that may be abnormal. It should be noted that, the edge detection algorithm and the CNN convolutional neural network algorithm belong to the prior art, and specific calculation steps are not repeated.
S2, obtaining a suspected pitting area according to the appearance image and the gray level difference characteristics of the suspected abnormal area; obtaining a first rule according to morphological characteristics of the suspected pitting area; obtaining a second regularity according to the change characteristics of the edge profile of the suspected pitting area; and obtaining the shape characteristic value of the suspected pitting area according to the first rule degree and the second rule degree.
In order to improve the detection accuracy of the pitting condition of the surface of the non-circular planetary gear, firstly, the suspected abnormal region needs to be analyzed to judge whether the suspected abnormal region accords with the gray level change characteristics of the pitting region or the dust collecting region; because the gray level difference between the pitting area or the dust collecting area and other normal areas on the surface of the gear is obvious and the light-dark contrast ratio is high, the suspected pitting area is obtained according to the gray level difference characteristics of the appearance image and the suspected abnormal area, preferably, in one embodiment of the invention, the obtaining the suspected pitting area comprises: calculating the average value of the gray values of any suspected abnormal region, and obtaining the region gray average value of the any suspected abnormal region; calculating the average value of gray values of the appearance image to obtain the overall appearance gray average value; calculating and normalizing the absolute value of the difference value of the regional gray average value and the overall appearance gray average value to obtain an abnormal characteristic value of any suspected abnormal region; when the abnormal characteristic value is larger, the gray level difference between the random suspected abnormal region and the whole appearance image is larger, the light-dark contrast ratio is more obvious, and the random suspected abnormal region is more likely to be a pitting region or a dust gathering region; conversely, a smaller abnormal feature value means that the smaller the difference in gradation between the arbitrary suspected abnormal region and the entire appearance image, the more likely the arbitrary suspected abnormal region is a normal region. When the abnormal characteristic value exceeds a preset abnormal threshold value, any suspected abnormal region is a suspected pitting region; in the embodiment of the invention, the preset abnormal threshold is 0.5, and an implementer can determine according to implementation scenes. The suspected pitting area is obtained through screening by analyzing the suspected abnormal area, so that the detection accuracy of the pitting condition of the gear surface is preliminarily improved.
Further, because the suspected pitting area comprises a real pitting area and a dust collecting area, the pitting area usually presents small holes or pits on the surface of the gear, the local brightness of the surface can be changed, and the dust collecting area also has the characteristic of similar brightness change; however, the difference is that the degree of regularity of the communicating regions between the two is different to a certain extent, and the theoretical characteristics show that the shape of the pitting area is irregular, and the dust gathering area is uniformly distributed with fine particles or clustered particles, so that the overall shape is regular. Secondly, the depth of the pitting area and the pitting area are different to a certain extent, the pitting area is usually a tiny change of the surface caused by abrasion or chemical action, the tiny pits or bulges can be left on the surface by the punctiform defects, but the depth difference is not too large on the whole, so that the depth change of the pitting area is shallow, otherwise, the depth difference of the dust collecting area is larger than that of the pitting area, obvious height change is usually formed due to accumulation of particles, and the depth difference can be observed in an image more clearly by the dust collecting area. Therefore, the real pitting areas in the suspected pitting areas can be obtained by distinguishing the rule degree and the depth change characteristics of the suspected pitting areas.
First, obtaining a first regularity according to morphological characteristics of a suspected pitting area, wherein the method specifically comprises the following steps: calculating fractal dimension values of suspected pitting areas and performing negative correlation mapping to obtain first regularity of the suspected pitting areas, wherein the first regularity is obtained by the method in the embodiment of the inventionThe function performs a negative correlation mapping, wherein/>Numerical value representing the need for a negative correlation map,/>The method is the same when the following steps of the negative correlation mapping are not specifically described, and the exponential function based on the natural constant is represented. It should be noted that, the fractal dimension belongs to the prior art, and specific calculation steps are not repeated; fractal dimension can be used to describe the degree of irregularity of an image, and common box counting methods are used in embodiments of the present invention to calculate the fractal dimension. When the fractal dimension is lower, the first regularity is higher, which means that the edge regularity of the suspected pitting area is higher, and the suspected pitting area is more likely to be a dust collecting area; conversely, when the fractal dimension is higher, the first regularity is smaller, which means that the edge regularity of the suspected pitting area is lower and more likely to be a pitting area.
Further, the curve change of the edge pixel points of the suspected pitting area can reflect the rule degree, so that a second rule degree is obtained according to the change characteristics of the edge profile of the suspected pitting area; preferably, in one embodiment of the present invention, obtaining the second rule degree includes: calculating the variance of the curvature of the edge pixel points of the suspected pitting area and carrying out negative correlation mapping to obtain a first contour change characteristic value of the suspected pitting area; it should be noted that, the obtaining of the edge pixel points and the corresponding curvatures belongs to the prior art, and specific calculation steps are not repeated; more regular edges generally exhibit a relatively constant curvature, while irregular edges may vary in curvature substantially. The smaller the variance of the curvature, the larger the first profile variation feature value, meaning that the edge profile of the suspected pitting area is more regular. Counting the number value of the curvature maximum value of the edge pixel point and carrying out negative correlation mapping to obtain a second contour change characteristic value; when the curvature of the edge pixel point is larger than the curvature of two other adjacent edge pixel points on the edge contour, the curvature of the edge pixel point is the maximum value; the smaller the number of maxima, the larger the second profile variation characteristic value, meaning that the smaller the fluctuation of curvature variation, the more regular the edge profile of the suspected pitting area. Calculating the average value of the first contour change characteristic value and the second contour change characteristic value to obtain a second regularity of the suspected pitting area; when the second regularity is larger, this means that the edge profile of the suspected pitting area is more regular, and more likely to be a dust collection area.
And further obtaining the shape characteristic value of the suspected pitting area according to the first rule degree and the second rule degree, wherein the method specifically comprises the following steps: calculating the product of a preset first weight and a first rule degree to obtain a first shape characteristic weight; calculating the product of a preset second weight and a second regularity to obtain a second shape characteristic weight; and calculating the sum value of the first shape feature weight and the second shape feature weight to obtain the shape feature value of the suspected pitting area. In the embodiment of the invention, the preset first weight and the preset second weight are respectively 0.5, and an implementer can determine according to implementation scenes; when the shape feature value is smaller, this means that the shape of the suspected pitting area is more irregular, and the more likely it is a true pitting area.
Step S3, obtaining a depth difference characteristic value according to the gray level distribution characteristics of the suspected pitting area; obtaining a depth change characteristic value according to gradient distribution characteristics of the suspected pitting area; and obtaining the depth characteristic value of the suspected pitting area according to the depth difference characteristic value and the depth change characteristic value.
The depth difference of the real pitting area and the dust collecting area is obviously different, so that the difference of the gray level distribution is caused by the depth difference reflected in the image, the more obvious the depth difference is, the wider the gray level distribution is, and conversely, the more uniform the depth is, the more concentrated the gray level is; at the same time, the depth difference also causes the gradient characteristics to be different between pixel points. And obtaining a depth difference characteristic value according to the gray level distribution characteristic of the suspected pitting area.
Preferably, in one embodiment of the present invention, acquiring the depth difference feature value includes: and acquiring a gray level histogram of the suspected pitting area, calculating the ratio of the vertical axis extremum difference and the horizontal axis extremum difference of the gray level histogram, and normalizing to obtain the depth difference characteristic value of the suspected pitting area. The horizontal axis of the gray level histogram represents different gray levels, and the vertical axis represents the number of pixels corresponding to the gray levels; the vertical axis extremum difference of the gray level histogram represents the difference of the number of the most and the least pixel points in the gray level, when the vertical axis extremum difference is larger, the pixel points in the suspected pitting area are mostly in individual gray level, the gray level distribution of the pixel points in the suspected pitting area is more uniform, and the depth change is less obvious. The smaller the horizontal axis extremum difference, the smaller the number of gray levels in the suspected pitting area, the less the gray values of the pixel points are similar, and the less obvious the depth change is. Therefore, when the vertical axis extremum difference of the gray level histogram is larger, the horizontal axis extremum difference is smaller, and the depth difference characteristic value is larger, which means that the less obvious the depth difference of the suspected pitting area is, the more likely the suspected pitting area is a real pitting area; conversely, when the vertical axis extremum difference is smaller, the number of pixel points in each gray level is more similar, the horizontal axis extremum difference is larger, the number of gray levels is more, the obtained depth difference characteristic value is smaller, the depth difference in the suspected pitting area is more obvious, and the suspected pitting area is more likely to be a dust collecting area.
Further, since the gradient characteristics between the pixels are also different due to the depth difference, the depth variation characteristic value is obtained according to the gradient distribution characteristics of the suspected pitting area, which specifically includes: and calculating the average value of the gradients of the pixel points in the suspected pitting area and carrying out negative correlation mapping to obtain the depth change characteristic value of the suspected pitting area. The whole inside of the real pitting area does not have too large depth difference, so that the gradient average value of the pixel points in the suspected pitting area is smaller; on the contrary, the dust collection area has larger depth difference, so that the gradient average value of the pixel points in the suspected pitting area is larger. Therefore, the smaller the gradient mean value of the pixel points in the suspected pitting area is, the larger the depth change characteristic value is, which means that the suspected pitting area is more likely to be a real pitting area.
Further, the depth characteristic value of the suspected pitting area is obtained according to the depth difference characteristic value and the depth variation characteristic value, and preferably, in one embodiment of the present invention, the obtaining the depth characteristic value includes: calculating the product of a preset third weight and a depth difference characteristic value to obtain a first depth characteristic weight; calculating the product of a preset fourth weight and a depth change characteristic value to obtain a second depth characteristic weight; and calculating the sum value of the first depth feature weight and the second depth feature weight to obtain the depth feature value of the suspected pitting area. In the embodiment of the present invention, the third weight is preset to be 0.4, the fourth weight is preset to be 0.6, and the implementer can determine according to the implementation scenario, when the depth characteristic value is larger, the suspected pitting area is more likely to be a real pitting area.
S4, obtaining a target pitting corrosion index according to the shape characteristic value and the depth characteristic value; and obtaining a target pitting area of the non-circular planetary gear according to the target pitting index of the suspected pitting area.
After the shape characteristic value and the depth characteristic value of the suspected pitting area are obtained, a target pitting index can be obtained according to the shape characteristic value and the depth characteristic value; preferably, in one embodiment of the present invention, obtaining the target pitting index includes: and mapping the negative correlation of the shape characteristic values to obtain shape characteristic mapping values, calculating the average value of the shape characteristic mapping values and the depth characteristic values, and obtaining the target pitting index of the suspected pitting area. When the shape feature value is smaller, the depth feature value is larger, the target pitting index is larger, which means that the suspected pitting area is more likely to be a real pitting area. The formula for obtaining the target pitting index comprises the following steps:
in the method, in the process of the invention, Target pitting index representing suspected pitting area,/>Representing shape eigenvalues,/>Representing a negative correlation mapping to shape feature values, i.e. shape feature map values,/>, forRepresenting depth characteristic values.
Further, the target pitting area of the non-circular planetary gear can be obtained according to the target pitting index of the suspected pitting area, which specifically comprises: when the target pitting index of the suspected pitting area exceeds a preset index threshold value, the suspected pitting area is the target pitting area. In the embodiment of the invention, the preset index threshold is 0.7, and the implementer can determine according to the implementation scene. And finally, the target pitting index and the target pitting area are obtained by analyzing the shape rule features and the depth difference features of the suspected pitting area, so that the situation that all suspected abnormal areas are regarded as pitting areas on the surface of the non-circular planetary gear by directly using an image vision algorithm is avoided, and the detection accuracy of the pitting condition is improved.
In summary, the embodiment of the invention provides a method for detecting quality abnormality of a non-circular planetary gear; acquiring a suspected abnormal region of the appearance image according to the convolutional neural network; obtaining a suspected pitting area according to the appearance image and the gray level difference characteristic of the suspected abnormal area, and obtaining a shape characteristic value of the suspected pitting area according to the morphological characteristic of the suspected pitting area and the change characteristic of the edge contour; and obtaining the depth characteristic value of the suspected pitting area according to the gray level distribution characteristic and the gradient distribution characteristic of the suspected pitting area. The method comprises the steps of obtaining a target pitting index according to a shape characteristic value and a depth characteristic value; the target pitting area of the non-circular planetary gear is obtained according to the target pitting index of the suspected pitting area, so that all suspected abnormal areas are prevented from being regarded as pitting areas on the surface of the non-circular planetary gear by using an image vision algorithm, and the detection accuracy of the pitting condition is improved.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (8)
1. A method for detecting quality anomalies in a non-circular planetary gear, the method comprising the steps of:
obtaining an appearance image for detecting the surface quality of a non-circular planetary gear, and obtaining a suspected abnormal region of the appearance image according to a convolutional neural network;
Obtaining a suspected pitting area according to the appearance image and the gray level difference characteristic of the suspected abnormal area; obtaining a first rule according to morphological characteristics of the suspected pitting area; obtaining a second regularity according to the change characteristics of the edge profile of the suspected pitting area; obtaining a shape characteristic value of the suspected pitting area according to the first rule degree and the second rule degree;
Obtaining a depth difference characteristic value according to the gray level distribution characteristics of the suspected pitting area; obtaining a depth change characteristic value according to gradient distribution characteristics of the suspected pitting area; obtaining a depth characteristic value of the suspected pitting area according to the depth difference characteristic value and the depth change characteristic value;
Obtaining a target pitting index according to the shape characteristic value and the depth characteristic value; acquiring a target pitting area of the non-circular planetary gear according to a target pitting index of the suspected pitting area;
the step of obtaining the first regularity according to the morphological characteristics of the suspected pitting area comprises the following steps:
Calculating fractal dimension values of the suspected pitting areas and carrying out negative correlation mapping to obtain first regularity of the suspected pitting areas;
the step of obtaining the second regularity according to the change characteristics of the edge profile of the suspected pitting area comprises the following steps:
Calculating the variance of the curvature of the edge pixel points of the suspected pitting area and carrying out negative correlation mapping to obtain a first contour change characteristic value of the suspected pitting area; counting the number value of the curvature maximum value of the edge pixel point and carrying out negative correlation mapping to obtain a second contour change characteristic value; and calculating the average value of the first contour change characteristic value and the second contour change characteristic value to obtain the second regularity of the suspected pitting area.
2. The method for detecting quality anomalies of a non-circular planetary gear according to claim 1, wherein the step of obtaining a suspected pitting area based on the appearance image and the grayscale difference characteristics of the suspected anomaly area includes:
Calculating the average value of gray values of any suspected abnormal region, and obtaining the region gray average value of the any suspected abnormal region; calculating the average value of gray values of the appearance image to obtain the overall appearance gray average value; calculating and normalizing the absolute value of the difference value of the regional gray average value and the overall appearance gray average value to obtain an abnormal characteristic value of the arbitrary suspected abnormal region; and when the abnormal characteristic value exceeds a preset abnormal threshold value, the any suspected abnormal region is a suspected pitting region.
3. The method for detecting quality abnormality of a non-circular planetary gear according to claim 1, wherein the step of obtaining the shape characteristic value of the suspected pitting area according to the first rule degree and the second rule degree includes:
calculating the product of a preset first weight and the first regularity to obtain a first shape characteristic weight; calculating the product of a preset second weight and the second regularity to obtain a second shape characteristic weight; and calculating the sum value of the first shape feature weight and the second shape feature weight to obtain the shape feature value of the suspected pitting area.
4. The method for detecting quality anomalies of a non-circular planetary gear according to claim 1, wherein the step of obtaining depth difference feature values from gray distribution features of suspected pitting areas comprises:
And acquiring a gray level histogram of the suspected pitting area, calculating the ratio of the vertical axis extremum difference and the horizontal axis extremum difference of the gray level histogram, and normalizing to obtain the depth difference characteristic value of the suspected pitting area.
5. The method for detecting quality anomalies of a non-circular planetary gear according to claim 1, wherein the step of obtaining a depth variation characteristic value from a gradient distribution characteristic of a suspected pitting area includes:
and calculating the average value of the gradients of the pixel points in the suspected pitting area and carrying out negative correlation mapping to obtain the depth change characteristic value of the suspected pitting area.
6. The method according to claim 1, wherein the step of obtaining the depth characteristic value of the suspected pitting area based on the depth difference characteristic value and the depth variation characteristic value comprises:
Calculating the product of a preset third weight and a depth difference characteristic value to obtain a first depth characteristic weight; calculating the product of a preset fourth weight and a depth change characteristic value to obtain a second depth characteristic weight; and calculating the sum value of the first depth feature weight and the second depth feature weight to obtain the depth feature value of the suspected pitting area.
7. The method for detecting quality abnormality of a non-circular planetary gear according to claim 1, wherein said step of obtaining a target pitting index from said shape characteristic value and said depth characteristic value comprises:
And mapping the shape characteristic value in a negative correlation way to obtain a shape characteristic mapping value, and calculating the average value of the shape characteristic mapping value and the depth characteristic value to obtain the target pitting index of the suspected pitting area.
8. The method for detecting quality abnormality of a non-circular planetary gear according to claim 1, wherein the step of acquiring a target pitting area of the non-circular planetary gear based on a target pitting index of the suspected pitting area includes:
and when the target pitting index of the suspected pitting area exceeds a preset index threshold value, the suspected pitting area is the target pitting area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410385905.9A CN117974666B (en) | 2024-04-01 | 2024-04-01 | Quality anomaly detection method for non-circular planetary gear |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410385905.9A CN117974666B (en) | 2024-04-01 | 2024-04-01 | Quality anomaly detection method for non-circular planetary gear |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117974666A true CN117974666A (en) | 2024-05-03 |
| CN117974666B CN117974666B (en) | 2024-06-25 |
Family
ID=90861359
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410385905.9A Active CN117974666B (en) | 2024-04-01 | 2024-04-01 | Quality anomaly detection method for non-circular planetary gear |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117974666B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104751137A (en) * | 2015-03-19 | 2015-07-01 | 江苏大学 | Automatic measuring and classifying method and system for depth uniformity of steel dimple image |
| US20160142700A1 (en) * | 2014-11-19 | 2016-05-19 | Ginni Grover | Measuring Accuracy of Image Based Depth Sensing Systems |
| CN108830822A (en) * | 2018-03-12 | 2018-11-16 | 北京交通大学 | Recognition methods is worn away based on the pantograph carbon slide for improving Canny operator |
| CN114486732A (en) * | 2021-12-30 | 2022-05-13 | 武汉光谷卓越科技股份有限公司 | Ceramic tile defect online detection method based on line scanning three-dimension |
| CN116758057A (en) * | 2023-08-10 | 2023-09-15 | 山东贺铭电气有限公司 | Communication equipment defect detection method based on artificial intelligence |
| US20230326006A1 (en) * | 2020-08-28 | 2023-10-12 | Goertek Inc. | Defect detection method and device for an lcd screen |
| CN117522873A (en) * | 2024-01-05 | 2024-02-06 | 惠汕绿创(江苏)科技有限公司 | Solar photovoltaic module production quality detection system |
-
2024
- 2024-04-01 CN CN202410385905.9A patent/CN117974666B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160142700A1 (en) * | 2014-11-19 | 2016-05-19 | Ginni Grover | Measuring Accuracy of Image Based Depth Sensing Systems |
| CN104751137A (en) * | 2015-03-19 | 2015-07-01 | 江苏大学 | Automatic measuring and classifying method and system for depth uniformity of steel dimple image |
| CN108830822A (en) * | 2018-03-12 | 2018-11-16 | 北京交通大学 | Recognition methods is worn away based on the pantograph carbon slide for improving Canny operator |
| US20230326006A1 (en) * | 2020-08-28 | 2023-10-12 | Goertek Inc. | Defect detection method and device for an lcd screen |
| CN114486732A (en) * | 2021-12-30 | 2022-05-13 | 武汉光谷卓越科技股份有限公司 | Ceramic tile defect online detection method based on line scanning three-dimension |
| CN116758057A (en) * | 2023-08-10 | 2023-09-15 | 山东贺铭电气有限公司 | Communication equipment defect detection method based on artificial intelligence |
| CN117522873A (en) * | 2024-01-05 | 2024-02-06 | 惠汕绿创(江苏)科技有限公司 | Solar photovoltaic module production quality detection system |
Non-Patent Citations (2)
| Title |
|---|
| 杨世聪;张劲泉;姚国文;: "基于图像灰度分析的腐蚀钢绞线细观损伤行为", 固体力学学报, no. 03, 8 February 2018 (2018-02-08) * |
| 王卓;张云伟;喻勇;樊阳阳;: "主动热激励式红外热成像管道缺陷深度检测", 光学学报, no. 09, 27 April 2018 (2018-04-27) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117974666B (en) | 2024-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117173189B (en) | Visual inspection system for polishing effect of aluminum alloy surface | |
| CN114937055B (en) | Image self-adaptive segmentation method and system based on artificial intelligence | |
| CN115829883B (en) | Surface image denoising method for special-shaped metal structural member | |
| CN114359270B (en) | Computer vision-based automobile engine oil way copper sleeve defect detection method | |
| CN115601367A (en) | LED lamp wick defect detection method | |
| CN114219805B (en) | Intelligent detection method for glass defects | |
| CN116630813B (en) | Highway road surface construction quality intelligent detection system | |
| CN114581692B (en) | Vibration damper fault detection method and system based on intelligent pattern recognition | |
| CN112581434A (en) | Image identification method for product defect detection | |
| CN115953398B (en) | Defect identification method for strip steel surface | |
| CN115861320A (en) | Intelligent detection method for automobile part machining information | |
| CN116563288B (en) | Detection method for threaded hole of gear of automobile engine | |
| CN116091455A (en) | Steel mesh surface defect judging method based on machine vision | |
| CN116152242B (en) | Visual detection system of natural leather defect for basketball | |
| CN116402810B (en) | Image processing-based lubricating oil anti-abrasive particle quality detection method | |
| CN116883408B (en) | Integrating instrument shell defect detection method based on artificial intelligence | |
| CN116523923B (en) | Battery case defect identification method | |
| CN115311289A (en) | Method for detecting oil stain defects of plain-color cloth | |
| CN115018790A (en) | Workpiece surface defect detection method based on anomaly detection | |
| CN114581805A (en) | Coating roller surface defect detection method adopting 3D line laser profile technology | |
| CN115294097A (en) | Textile surface defect detection method based on machine vision | |
| CN115222735A (en) | Metal mold quality detection method based on pockmark defects | |
| CN117269168B (en) | New energy automobile precision part surface defect detection device and detection method | |
| CN117974666B (en) | Quality anomaly detection method for non-circular planetary gear | |
| CN110838101B (en) | Image segmentation quality evaluation method based on edge characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |