CN114359275A - Hydraulic gear pump defect detection method and system based on artificial intelligence - Google Patents
Hydraulic gear pump defect detection method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN114359275A CN114359275A CN202210254838.8A CN202210254838A CN114359275A CN 114359275 A CN114359275 A CN 114359275A CN 202210254838 A CN202210254838 A CN 202210254838A CN 114359275 A CN114359275 A CN 114359275A
- Authority
- CN
- China
- Prior art keywords
- target area
- difference
- acquiring
- sub
- index
- 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
- 230000007547 defect Effects 0.000 title claims abstract description 60
- 238000001514 detection method Methods 0.000 title claims abstract description 38
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 16
- 238000000034 method Methods 0.000 claims abstract description 25
- 230000002159 abnormal effect Effects 0.000 claims abstract description 20
- 239000011159 matrix material Substances 0.000 claims description 13
- 230000005856 abnormality Effects 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 239000003086 colorant Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P80/00—Climate change mitigation technologies for sector-wide applications
- Y02P80/10—Efficient use of energy, e.g. using compressed air or pressurized fluid as energy carrier
Landscapes
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of hydraulic system part defect detection, in particular to a hydraulic gear pump defect detection method and system based on artificial intelligence. The method comprises the following steps: acquiring a tooth surface target area of a gear pump image; acquiring the complexity of a target area; acquiring the entropy of a target area under each color channel, acquiring a spatial distribution vector according to the entropy of the target area, and acquiring an abnormal index according to the difference between the spatial distribution vector and the spatial distribution vector of a normal tooth surface; acquiring the difference between any two pixel points in the target area, and acquiring the difference index of the target area according to the difference between the difference and the difference of the normal tooth surface; and acquiring a defect confidence coefficient according to the complexity, the abnormal index and the difference index, wherein when the defect confidence coefficient is greater than a preset threshold value, the gear pump has defects. The target area is comprehensively considered from multiple aspects, different characteristic parameters are obtained, and the accuracy of detecting the tooth surface defects is improved.
Description
Technical Field
The invention relates to the technical field of hydraulic system part defect detection, in particular to a hydraulic gear pump defect detection method and system based on artificial intelligence.
Background
With the automation development of modern industrial enterprises, hydraulic equipment in each industrial enterprise is rising, and in order to ensure normal and efficient operation of the hydraulic equipment, detection, maintenance and maintenance of components, parts and the like of the hydraulic equipment are generally required in the actual application process. The gear pump is a power element of a hydraulic system, the gear is an important transmission part commonly used in hydraulic equipment, and the quality and the precision of the tooth surface of the gear pump have important influence on the comprehensive performance and the service life of the whole hydraulic equipment.
Most of the existing fault detection on the gear pump of the hydraulic equipment is carried out by workers with abundant experience, but the detection on the gear pump by the workers is usually carried out according to a complicated detection flow, so that the workload is large and the detection efficiency is not high; and the manual detection capability is limited, the comprehensiveness of the gear detection cannot be ensured, and the false detection rate is high.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a hydraulic gear pump defect detection method and system based on artificial intelligence, and the adopted technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a hydraulic gear pump defect detection method based on artificial intelligence, including the following steps:
acquiring a target area of a gear pump image of a hydraulic system, wherein the target area comprises a tooth surface area of the gear pump; acquiring the complexity of the target area;
dividing the target area into a plurality of sub-blocks parallel to the gear pump tooth crest, wherein the sub-block close to the gear pump tooth crest is a first sub-block; acquiring the entropies of all the sub-blocks under each color channel, and performing weighted summation on the entropies of all the sub-blocks by taking the distance between each sub-block and the first sub-block as a weight to obtain the entropy of the target area; acquiring a spatial distribution vector according to the entropy of the target area under each color channel, and acquiring an abnormal index according to the difference between the spatial distribution vector and the spatial distribution vector of the normal tooth surface;
acquiring the gradient size and direction of each pixel point in the target area, acquiring the difference between any two pixel points according to the gradient size and direction, and acquiring the difference index of the target area according to the difference between the difference and the difference of a normal tooth surface;
and acquiring a defect confidence coefficient according to the complexity, the abnormal index and the difference index, wherein when the defect confidence coefficient is greater than a preset threshold value, the gear pump has defects.
Preferably, the step of obtaining the complexity of the target region includes:
dividing the target area into a plurality of windows with the same size, and filtering each window to obtain a characteristic value of each window;
and acquiring the ratio of the characteristic value of each window in the target area, and acquiring the complexity of the target area according to the ratio.
Preferably, the step of performing weighted summation on the entropies of all the sub-blocks by taking the distance between each sub-block and the first sub-block as a weight to obtain the entropy of the target region further includes:
equally dividing the value range of each color channel into a plurality of grades; obtaining the number of the levels in each sub-block, and optimizing the entropy of each sub-block according to the number of the levels in each sub-block to obtain the optimized entropy of each sub-block;
and taking the distance between each sub-block and the first sub-block as a weight to perform weighted summation on the optimized entropy to obtain the entropy of the target area.
Preferably, the step of obtaining a spatial distribution vector according to the entropy of the target region in each color channel includes:
acquiring the mean value of the entropies of the target area corresponding to all levels in each color channel, and acquiring the variance of the target area in each color channel according to the mean value;
and forming a spatial distribution vector by using the mean and variance of the target area under all color channels.
Preferably, the step of obtaining the difference between any two pixel points according to the gradient magnitude and the gradient direction includes:
forming a binary group by the gradient size and the gradient direction of each pixel point;
taking any pixel point in the target area as a central pixel point, wherein pixel points in the neighborhood of the central pixel point are to-be-processed pixel points; obtaining the similarity between the binary groups corresponding to any two to-be-processed pixel points; the difference degree and the similarity degree are in a negative correlation relationship.
Preferably, the step of obtaining the difference index of the target area according to the difference between the difference degree and the difference degree in the normal tooth surface includes:
forming a difference vector by using the difference degree between the pixels to be processed in each neighborhood, and acquiring an autocorrelation matrix of the difference vector; obtaining a gradient index according to the mean value of all elements in the autocorrelation matrix;
and acquiring gradient indexes corresponding to all pixel points in the target region as a gradient index sequence, and calculating the difference between the gradient index sequence and the normal region gradient index sequence as the difference index.
Preferably, the step of obtaining a defect confidence according to the complexity, the anomaly indicator and the difference indicator includes:
constructing a defect confidence coefficient model according to the complexity, the abnormal index and the difference index as follows:
wherein,
representing the defect confidence;
representing the complexity of the target region;
an abnormality index representing the target region;
a difference indicator representing the target area;
a parameter representing the abnormality index;
a parameter representing the indicator of difference.
In a second aspect, another embodiment of the present invention provides an artificial intelligence based hydraulic gear pump defect detection system, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.
The invention has the following beneficial effects: by acquiring a target area image of the gear pump, analyzing the complexity of the target area; in order to avoid the influence of color on tooth surface detection, the entropy of the target area is further calculated under different color channels, a space index is obtained according to the entropy of the target area, and the difference between the space index under the different color channels and the space index of the normal target area is combined to be used as an abnormal index of the target area. And then obtaining the gradient characteristic of each pixel point in the target area, and obtaining the difference index of the current detection target area according to the difference between the gradient characteristic of the current detection target area and the gradient characteristic of the normal target area. And obtaining the defect confidence of the target area according to the complexity, the abnormal index and the difference index of the target area. The characteristics of the target area are considered from multiple aspects, different characteristic parameters are obtained for analysis, the accuracy of tooth surface defect detection is improved, and error influence caused by single detection is avoided.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a method for detecting defects of a hydraulic gear pump based on artificial intelligence according to an embodiment of the present invention.
Detailed Description
In order to further explain the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description, the structure, the features and the effects of the method and the system for detecting the defect of the hydraulic gear pump based on artificial intelligence according to the present invention are provided with the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
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 embodiment of the invention is mainly applied to the detection of the tooth surface defects of the gear pump of the hydraulic system, and the complexity of a target area is analyzed by acquiring the target area image of the gear pump; further calculating the entropy of the target area under different color channels to obtain an abnormal index of the target area; then obtaining the gradient characteristics of each pixel point in the target area to obtain the difference index of the current detection target area; and obtaining the defect confidence of the target region by combining the complexity, the abnormal index and the difference index of the target region. The target area is analyzed from multiple aspects, different characteristic parameters are obtained, and the accuracy of detecting the tooth surface defects is improved.
The invention provides a hydraulic gear pump defect detection method and a hydraulic gear pump defect detection system based on artificial intelligence.
Referring to fig. 1, a flow chart of a method for detecting a defect of a hydraulic gear pump based on artificial intelligence according to an embodiment of the present invention is shown, where the method specifically includes the following steps:
s100, acquiring a target area of a gear pump image of a hydraulic system, wherein the target area comprises a tooth surface area of the gear pump; and acquiring the complexity of the target area.
A camera is arranged on one side, which is opposite to the tooth surface of the gear pump of the hydraulic system, so as to collect images of the tooth surface area, and the position of the camera is ensured to collect the whole area of one tooth surface of the gear pump, so that an initial image to be analyzed is obtained.
Because the initial image may include a plurality of tooth surfaces and non-tooth surface regions, in order to improve the detection accuracy and reduce the calculation amount, in the embodiment of the invention, the initial image is processed by using a semantic segmentation network, the obtained ROI of the tooth surface of the gear pump is a target region, the semantic segmentation network is structurally an encoder-decoder, and the specific training process is as follows:
(1) the input of the semantic segmentation network is an initial image of the gear pump;
(2) marking the initial image, marking the pixel point of the tooth surface area in the initial image as 1, and marking the pixel points of other areas as 0;
(3) performing feature extraction on the initial image by using an encoder, outputting a feature image, and performing up-sampling on the feature image by using a decoder;
(4) the loss function adopts a cross entropy loss function;
(5) the output of the semantic segmentation network is a tooth surface segmentation effect graph.
Furthermore, the tooth surface segmentation effect image output by the semantic segmentation network is used as a mask to be multiplied by the initial image to obtain a tooth surface ROI area to be analyzed, and the ROI area is used as a target area of subsequent defect analysis.
Extracting characteristic parameters of the obtained target area of the tooth surface so as to obtain the complexity of texture distribution of the tooth surface, dividing the target area into a plurality of windows with the same size, and filtering each window to obtain a characteristic value of each window; and acquiring the ratio of the characteristic value of each window in the target area, and acquiring the complexity of the target area according to the ratio.
The specific method for acquiring the complexity of the tooth surface target area comprises the following steps:
firstly, a target area is divided into a plurality of windows with the same size, and the target area of the tooth surface acquired by default in the embodiment of the invention is a rectangular area with basic rules. Preferably, the size of each window is set to
。
Secondly, a window filtering model is constructed for obtaining the characteristic value of each window center pixel point, and the specific calculation is as follows:
wherein,
is shown as
Characteristic value of the center point of each window;
representing coordinates of the center point of the window;
within a presentation window
Pixel values of the pixel points are processed;
within a presentation window
The weight corresponding to the pixel point is located;
indicating the size of the window, in an embodiment of the invention
。
Further, the weight corresponding to the pixel point in each window is calculated as:
wherein,
within a presentation window
The weight corresponding to the pixel point is located;
representing coordinates of the center point of the window;
and representing the coordinates of any pixel point in the window.
By analogy, obtaining the characteristic value of the central point of each window; obtaining the ratio of the window in the target area according to the characteristic value of the central point of each window as follows:
wherein,
is shown as
The proportion of each window in the target area;
is shown as
Characteristic value of the center point of each window;
representing the number of windows in the target area.
Further, analyzing the texture complexity of the target area specifically includes:
wherein,
representing the complexity of the target area;
is shown as
The proportion of each window in the target area;
representing the number of windows in the target area.
When in use
The larger the value of (a), the more complicated the texture feature of the target area is;
the smaller the value of (a), the simpler the texture feature of the corresponding target region.
Step S200, dividing a target area into a plurality of sub-blocks parallel to the gear pump tooth crest, wherein the sub-block close to the gear pump tooth crest is a first sub-block; acquiring the entropies of all sub-blocks under each color channel, and performing weighted summation on the entropies of all sub-blocks by taking the distance between each sub-block and the first sub-block as a weight to obtain the entropy of a target area; and acquiring a spatial distribution vector according to the entropy of the target area under each color channel, and acquiring an abnormal index according to the difference between the spatial distribution vector and the spatial distribution vector of the normal tooth surface.
Considering that the defect detection may be affected under different color channels, so that the result is inaccurate, the embodiment of the present invention analyzes the target region under different color channels.
Specifically, a target area of the gear pump is divided into a plurality of sub-blocks parallel to the tooth tops of the gear pump in an equal proportion, the number of the sub-blocks is set to be 5 in the embodiment of the invention, namely, the target area is divided into five rectangular areas in an equal proportion to obtain five sub-blocks; due to the fact that in actual work, tooth crest areas are common areas where defects exist due to meshing between gears, the subblock close to the tooth crest of the gear pump is used as the first subblock, the distance between each subblock and the first subblock is obtained, and the distance is used as the attention degree for analyzing the defects possibly occurring on each subblock.
Further, the value range of each color channel is equally divided into a plurality of grades; obtaining the number of levels in each subblock, and optimizing the entropy of each subblock according to the number of levels in each subblock to obtain the optimized entropy of each subblock; and taking the distance between each sub-block and the first sub-block as a weight to perform weighted summation on the optimized entropy to obtain the entropy of the target region.
Specifically, in the embodiment of the present invention, the three RGB color channels of the target region are analyzed, in order to reduce the subsequent calculation amount, the value range of each color channel is equally divided into 10 levels, and each sub-block is analyzed in each color channel.
As an example, taking a red channel of a target region as an example, marking a level corresponding to a pixel point in each sub-block, counting the number of different levels of the sub-block appearing in the red channel, and designing a factor model according to the number of levels in each sub-block as follows:
wherein,
is shown as
A factor of the individual block;
is shown as
The number of different levels present in the individual blocks;
indicating the number of levels divided under the red channel,
。
further, the probability of each level occurring in each sub-block is calculated as:
wherein,
is shown as
Is ranked at the first
Probability of occurrence in an individual block;
is shown as
Is ranked at the first
The number of occurrences in an individual block;
and the number of all pixel points in the sub-block is represented.
And aiming at the difference of each sub-block in spatial distribution, taking the distance between each sub-block and the first sub-block as the weight corresponding to the sub-block, wherein the closer the distance between a certain sub-block and the first sub-block is, the larger the weight corresponding to the sub-block is. Taking the weight value corresponding to each sub-block as a weight, and performing weighted summation on the entropy of each grade in each sub-block by combining the factor corresponding to each sub-block to obtain the entropy of each grade in the target area, wherein the entropy of each grade in the target area is specifically:
wherein,
is shown as
Entropy of the individual levels in the target area;
is shown as
A factor of the individual block;
is shown as
Distance between the sub-block and the first sub-blockSeparating;
is shown as
Is ranked at the first
Probability of occurrence in individual sub-blocks.
Further, obtaining the mean value of the entropies of the target areas corresponding to all levels in each color channel, and obtaining the variance of the target areas in each color channel according to the mean value; the spatial distribution vector is constructed with the mean and variance of the target region under all color channels.
Specifically, the mean value of the entropy values of all levels in the target region in the red channel is obtained as follows:
wherein,
is shown as
Entropy of the individual levels in the target area;
representing the mean of the entropy values of all levels in the target region.
Further obtaining the corresponding variance according to the mean value of the entropy values of all levels in the red channel in the target area as follows:
wherein,
is shown as
Entropy of the individual levels in the target area;
means representing the mean of the entropy values of all levels in the target region;
representing the variance of the entropy values of all levels in the target region.
In the embodiment of the invention, the obtained mean value and variance are used as space indexes in a red channel; and by analogy, the spatial index of the target area in the green channel and the spatial index of the target area in the blue channel are obtained. In order to embody the brightness distribution information of the target area, the HSV color space conversion is carried out on the target area image, and the space index of the target area in the brightness channel is calculated.
The space distribution vector formed by the space indexes in each channel corresponding to the target area is as follows:
wherein,
represents the mean of all levels in the target region under the red channel;
representing the variance of all levels in the target region under the red channel;
represents the mean of all levels in the target area under the green channel;
representing the variance of all levels in the target area under the green channel;
represents the mean of all levels in the target region under the blue channel;
represents the variance of all levels in the target region under the blue channel;
representing the mean value of all levels in the target area under the brightness channel;
representing the variance of all levels in the target region under the luminance channel.
Comparing the obtained space distribution vector with the space distribution vector of the target area corresponding to the normal tooth surface, and obtaining the abnormal index of the target area of the current detection tooth surface as follows:
wherein,
an abnormality index indicating a target region;
representing a spatially distributed vector
An element;
representing normal spatial distribution vector
And (4) each element.
Therefore, the currently detected target area is analyzed under different color channels, the spatial distribution condition of the target area can be visually detected, and the error influence caused by the quantization of colors and brightness values is avoided.
Step S300, obtaining the gradient size and direction of each pixel point in the target area, obtaining the difference degree between any two pixel points according to the gradient size and direction, and obtaining the difference index of the target area according to the difference between the difference degree and the difference degree in the normal tooth surface.
In order to further improve the accuracy of defect detection and analysis of the target region, the gradient amplitude and the direction of the surface pixel points are randomly distributed when the tooth surface has a defect region. Therefore, in the embodiment of the invention, the gradient index of each pixel point of the image data of the tooth surface target area is extracted, and the gradient size and the gradient direction of each pixel point form a binary group; taking any pixel point in the target area as a central pixel point, and taking a pixel point in the neighborhood of the central pixel point as a pixel point to be processed; obtaining the similarity between the binary groups corresponding to any two pixel points to be processed; the difference and the similarity are in a negative correlation relationship.
Specifically, the gradient size and the gradient direction of all pixel points in the target area are obtained, and the gradient size and the gradient direction corresponding to each pixel point form a binary group, that is, each pixel point corresponds to one binary group. And taking any pixel point of the target area as a central pixel point, and taking a neighborhood pixel point of the central pixel point as a pixel point to be processed. Preferably, in the embodiment of the present invention, eight neighborhoods of the central pixel point are selected for analysis, that is, each central pixel point corresponds to eight to-be-processed pixel points.
Calculating the similarity between every two adjacent to-be-processed pixel points in the neighborhood, wherein the similarity between every two adjacent to-be-processed pixel points is calculated by using a correlation coefficient method in the embodiment of the invention, and the method specifically comprises the following steps:
wherein,
is shown as
A pixel point to be processed and the second
Similarity among the pixel points to be processed;
is shown as
The second group corresponding to the pixel point to be processed
An element;
is shown as
The second group corresponding to the pixel point to be processed
And (4) each element.
The difference between every two adjacent to-be-processed pixel points in the neighborhood is in a negative correlation with the similarity, and then the difference is:
wherein,
is shown as
A pixel point to be processed and the second
The difference degree between the pixel points to be processed;
is shown as
A pixel point to be processed and the second
Similarity between the pixels to be processed.
Based on the same principle, obtaining the difference between all adjacent two pixels to be processed in each neighborhood, forming a difference vector by the difference between the pixels to be processed in each neighborhood, and obtaining an autocorrelation matrix of the difference vector; obtaining a gradient index according to the mean value of all elements in the autocorrelation matrix; and acquiring gradient indexes corresponding to all pixel points in the target region as a gradient index sequence, and calculating the difference between the gradient index sequence and the normal region gradient index sequence as a difference index.
Specifically, in the embodiment of the present invention, the pixel point to be processed in each neighborhood is respectively marked as 1 in the clockwise direction
8, the corresponding difference vector of each neighborhood center pixel point is
Wherein
expressing the difference degree between the 1 st pixel point to be processed and the 2 nd pixel point to be processed;
and expressing the difference degree between the 1 st pixel point to be processed and the 4 th pixel point to be processed.
Further, for each central pixel pointAnd analyzing the corresponding difference vector, wherein the gradient distribution condition around the central pixel point is analyzed by acquiring the autocorrelation matrix of each difference vector. The autocorrelation matrix is of a size of
Each element value in the matrix represents the similarity between every two elements in the difference vector, and specifically, the matrix is as follows:
wherein,
represents the second in the disparity vector
An element and
the similarity between the elements, i.e. the first in the autocorrelation matrix
Line of
The element values of the columns;
represents the second in the disparity vector
An element;
represents the second in the disparity vector
An element;
represents a constant with the value range:
。
it should be noted that, when the similarity between each two elements in the disparity vector is larger, the gradient change between the two elements is more similar, and the corresponding element value in the autocorrelation matrix is closer to 1.
Further, a mean value of all the element values in the autocorrelation matrix is calculated, and the mean value is used as a gradient index corresponding to the central pixel point.
Based on the method for obtaining the same gradient index, the gradient indexes corresponding to all the pixel points in the target area of the gear pump are obtained to form a gradient index sequence.
Comparing the obtained gradient index sequence with a normal gradient index sequence corresponding to a target area under a normal tooth surface; in the embodiment of the invention, the gradient index sequence and the normal gradient index sequence are analyzed by adopting a maximum weighted mean difference method, and in other embodiments, a dynamic time warping or Pearson correlation algorithm can be adopted.
And finally obtaining a difference index between the gradient index sequence of the current target area and the normal gradient index sequence of the normal tooth surface target area, wherein when the value of the difference index is larger, the difference of the gradient index of the current target area and the normal target area is larger, and the tooth surface defect degree of the current detected target area is more serious.
And S400, acquiring defect confidence coefficient according to the complexity, the abnormal index and the difference index, wherein when the defect confidence coefficient is greater than a preset threshold value, the gear pump has defects.
The complexity, the abnormal index and the difference index of the target area are respectively obtained in the steps S100, S200 and S300, and the defect confidence level is in positive correlation with the complexity, the abnormal index and the difference index, so that the defect confidence level model is obtained as follows:
wherein,
representing a defect confidence level;
representing the complexity of the target area;
an abnormality index indicating a target region;
a difference index representing a target area;
a parameter indicating an abnormality index;
a parameter representing a discrepancy indicator.
Preferably, the device is arranged in the embodiment of the invention
。
Furthermore, in order to facilitate the working personnel to know the tooth surface defect condition of the gear pump of the hydraulic system in real time and avoid the problems of abnormal gear operation, reduced working efficiency and the like caused by the overlarge tooth surface defect degree of the gear pump, the defect confidence coefficient preset threshold is set as
That is, when the obtained defect confidence of the target area is greater than the preset threshold, it indicates that the tooth surface defect degree of the gear pump is too high, which will affect the operation process of the gear, resulting in the reduction of the working efficiency of the gear pump, and should prompt the worker to timely overhaul and maintain the tooth surface of the gear pumpAnd the generation of a great safety problem is prevented.
In summary, in the embodiment of the present invention, the target area image of the gear pump is obtained, and then the complexity of the target area is analyzed; in order to avoid the influence of different colors, the entropy of the target area is further calculated under different color channels, a spatial index is obtained according to the entropy of the target area, a spatial distribution vector is obtained by combining the spatial indexes under the different color channels, and the difference between the currently detected spatial distribution vector of the target area and the spatial distribution vector of the normal target area is used as an abnormal index of the target area. And then obtaining the gradient characteristic of each pixel point in the target area, and obtaining the difference index of the current detection target area according to the difference between the gradient characteristic of the current detection target area and the gradient characteristic of the normal target area. And obtaining the defect confidence of the target region by combining the complexity, the abnormal index and the difference index of the target region. The target area is analyzed from multiple aspects, different characteristic parameters are obtained, the accuracy of tooth surface defect detection is improved, and error influence caused by single detection is avoided.
Based on the same inventive concept as the method embodiment, the embodiment of the invention also provides a hydraulic gear pump defect detection system based on artificial intelligence, which comprises: a processor, a memory, and a computer program stored in the memory and executable on the processor. The processor, when executing the computer program, implements the steps of one embodiment of the artificial intelligence based hydraulic gear pump defect detection method described above, such as the steps shown in fig. 1. The method for detecting the defect of the hydraulic gear pump based on artificial intelligence is described in detail in the above embodiments, and is not repeated.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (8)
1. A hydraulic gear pump defect detection method based on artificial intelligence is characterized by comprising the following steps:
acquiring a target area of a gear pump image of a hydraulic system, wherein the target area comprises a tooth surface area of the gear pump; acquiring the complexity of the target area;
dividing the target area into a plurality of sub-blocks of a rectangular area parallel to the gear pump tooth crest in an equal proportion, wherein the sub-block close to the gear pump tooth crest is a first sub-block; acquiring the entropies of all the sub-blocks under each color channel, and performing weighted summation on the entropies of all the sub-blocks by taking the distance between each sub-block and the first sub-block as a weight to obtain the entropy of the target area; acquiring a spatial distribution vector according to the entropy of the target area under each color channel, and acquiring an abnormal index according to the difference between the spatial distribution vector and the spatial distribution vector of the normal tooth surface;
acquiring the gradient size and direction of each pixel point in the target area, acquiring the difference between any two pixel points according to the gradient size and direction, and acquiring the difference index of the target area according to the difference between the difference and the difference of a normal tooth surface;
and acquiring a defect confidence coefficient according to the complexity, the abnormal index and the difference index, wherein when the defect confidence coefficient is greater than a preset threshold value, the gear pump has defects.
2. The method of claim 1, wherein the step of obtaining the complexity of the target region comprises:
dividing the target area into a plurality of windows with the same size, and filtering each window to obtain a characteristic value of each window;
and acquiring the ratio of the characteristic value of each window in the target area, and acquiring the complexity of the target area according to the ratio.
3. The method according to claim 1, wherein the step of performing weighted summation on the entropies of all the sub-blocks by using the distance between each sub-block and the first sub-block as a weight to obtain the entropy of the target region further comprises:
equally dividing the value range of each color channel into a plurality of grades; obtaining the number of the levels in each sub-block, and optimizing the entropy of each sub-block according to the number of the levels in each sub-block to obtain the optimized entropy of each sub-block;
and taking the distance between each sub-block and the first sub-block as a weight to perform weighted summation on the optimized entropy to obtain the entropy of the target area.
4. The method according to claim 3, wherein the step of obtaining the spatial distribution vector according to the entropy of the target region in each color channel comprises:
acquiring the mean value of the entropies of the target area corresponding to all levels in each color channel, and acquiring the variance of the target area in each color channel according to the mean value;
and forming a spatial distribution vector by using the mean and variance of the target area under all color channels.
5. The method according to claim 1, wherein the step of obtaining the difference between any two pixel points according to the gradient magnitude and direction comprises:
forming a binary group by the gradient size and the gradient direction of each pixel point;
taking any pixel point in the target area as a central pixel point, wherein pixel points in the neighborhood of the central pixel point are to-be-processed pixel points; obtaining the similarity between the binary groups corresponding to any two to-be-processed pixel points; the difference degree and the similarity degree are in a negative correlation relationship.
6. The method of claim 5, wherein the step of obtaining the measure of variance of the target area based on the difference between the measure of variance and a measure of variance in a normal tooth plane comprises:
forming a difference vector by using the difference degree between the pixels to be processed in each neighborhood, and acquiring an autocorrelation matrix of the difference vector; obtaining a gradient index according to the mean value of all elements in the autocorrelation matrix;
and acquiring gradient indexes corresponding to all pixel points in the target region as a gradient index sequence, and calculating the difference between the gradient index sequence and the normal region gradient index sequence as the difference index.
7. The method of claim 1, wherein the step of obtaining a defect confidence measure from the complexity measure, the anomaly measure, and the discrepancy measure comprises:
constructing a defect confidence coefficient model according to the complexity, the abnormal index and the difference index as follows:
wherein,
representing the defect confidence;
representing the complexity of the target region;
an abnormality index representing the target region;
a difference indicator representing the target area;
a parameter representing the abnormality index;
a parameter representing the indicator of difference.
8. An artificial intelligence based hydraulic gear pump defect detection system comprising a memory, a processor and a computer program stored in said memory and executable on said processor, wherein said processor when executing said computer program implements the method of claim 1
7 the steps of any one of the methods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210254838.8A CN114359275B (en) | 2022-03-16 | 2022-03-16 | Hydraulic gear pump defect detection method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210254838.8A CN114359275B (en) | 2022-03-16 | 2022-03-16 | Hydraulic gear pump defect detection method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114359275A true CN114359275A (en) | 2022-04-15 |
| CN114359275B CN114359275B (en) | 2022-06-07 |
Family
ID=81094961
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210254838.8A Expired - Fee Related CN114359275B (en) | 2022-03-16 | 2022-03-16 | Hydraulic gear pump defect detection method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114359275B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114943739A (en) * | 2022-07-26 | 2022-08-26 | 山东三微新材料有限公司 | Aluminum pipe quality detection method |
| CN114998341A (en) * | 2022-08-04 | 2022-09-02 | 启东万惠机械制造有限公司 | Gear defect detection method and system based on template matching |
| CN114937041B (en) * | 2022-07-25 | 2022-10-25 | 聊城市博源节能科技有限公司 | Method and system for detecting defects of copper bush of oil way of automobile engine |
| CN115294142A (en) * | 2022-10-10 | 2022-11-04 | 丰源纸业(南通)有限公司 | Quality detection method for chemical fiber paper tubes |
| CN116433665A (en) * | 2023-06-13 | 2023-07-14 | 中国民航大学 | Aircraft component defect online identification system based on visual detection |
| CN116630332A (en) * | 2023-07-26 | 2023-08-22 | 山东华航高分子材料有限公司 | PVC plastic pipe orifice defect detection method based on image processing |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080205747A1 (en) * | 2006-12-28 | 2008-08-28 | Toshimasa Kuchii | Defect detecting device, defect detecting method, image sensor device, image sensor module, defect detecting program, and computer-readable recording medium |
| CN109447946A (en) * | 2018-09-26 | 2019-03-08 | 中睿通信规划设计有限公司 | A kind of Overhead optical cable method for detecting abnormality |
| CN112258463A (en) * | 2020-10-13 | 2021-01-22 | 房桂丽 | Photovoltaic defect feature acquisition method and system based on camera parameter adjustment |
| CN113822915A (en) * | 2021-07-30 | 2021-12-21 | 济宁安泰矿山设备制造有限公司 | Image stereo matching method for intelligent pump cavity endoscope fault diagnosis |
-
2022
- 2022-03-16 CN CN202210254838.8A patent/CN114359275B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080205747A1 (en) * | 2006-12-28 | 2008-08-28 | Toshimasa Kuchii | Defect detecting device, defect detecting method, image sensor device, image sensor module, defect detecting program, and computer-readable recording medium |
| CN109447946A (en) * | 2018-09-26 | 2019-03-08 | 中睿通信规划设计有限公司 | A kind of Overhead optical cable method for detecting abnormality |
| CN112258463A (en) * | 2020-10-13 | 2021-01-22 | 房桂丽 | Photovoltaic defect feature acquisition method and system based on camera parameter adjustment |
| CN113822915A (en) * | 2021-07-30 | 2021-12-21 | 济宁安泰矿山设备制造有限公司 | Image stereo matching method for intelligent pump cavity endoscope fault diagnosis |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114937041B (en) * | 2022-07-25 | 2022-10-25 | 聊城市博源节能科技有限公司 | Method and system for detecting defects of copper bush of oil way of automobile engine |
| CN114943739A (en) * | 2022-07-26 | 2022-08-26 | 山东三微新材料有限公司 | Aluminum pipe quality detection method |
| CN114943739B (en) * | 2022-07-26 | 2022-10-21 | 山东三微新材料有限公司 | Aluminum pipe quality detection method |
| CN114998341A (en) * | 2022-08-04 | 2022-09-02 | 启东万惠机械制造有限公司 | Gear defect detection method and system based on template matching |
| CN114998341B (en) * | 2022-08-04 | 2024-01-05 | 启东万惠机械制造有限公司 | Gear defect detection method and system based on template matching |
| CN115294142A (en) * | 2022-10-10 | 2022-11-04 | 丰源纸业(南通)有限公司 | Quality detection method for chemical fiber paper tubes |
| CN116433665A (en) * | 2023-06-13 | 2023-07-14 | 中国民航大学 | Aircraft component defect online identification system based on visual detection |
| CN116433665B (en) * | 2023-06-13 | 2023-08-22 | 中国民航大学 | Aircraft component defect online identification system based on visual detection |
| CN116630332A (en) * | 2023-07-26 | 2023-08-22 | 山东华航高分子材料有限公司 | PVC plastic pipe orifice defect detection method based on image processing |
| CN116630332B (en) * | 2023-07-26 | 2023-09-26 | 山东华航高分子材料有限公司 | PVC plastic pipe orifice defect detection method based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114359275B (en) | 2022-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114359275B (en) | Hydraulic gear pump defect detection method and system based on artificial intelligence | |
| CN107172418B (en) | A kind of tone scale map image quality evaluating method based on exposure status analysis | |
| CN113808138B (en) | Artificial intelligence-based wire and cable surface defect detection method | |
| CN103455979B (en) | A kind of low illumination level video image enhancement | |
| CN111402247A (en) | Machine vision-based method for detecting defects of suspension clamp on power transmission line | |
| CN1845605A (en) | Image processing and recognizing method for real-time monitoring power high voltage circuit breaker switch state | |
| CN113205051A (en) | Oil storage tank extraction method based on high spatial resolution remote sensing image | |
| CN110942137A (en) | Power grid information operation and maintenance monitoring method based on deep learning | |
| CN112396635B (en) | Multi-target detection method based on multiple devices in complex environment | |
| CN107749058A (en) | A kind of machine vision detection method and system of boiler tubing surface defect | |
| CN112767385B (en) | No-reference image quality evaluation method based on significance strategy and feature fusion | |
| CN112241707A (en) | Wind-powered electricity generation field intelligence video identification device | |
| CN111833347A (en) | Transmission line damper defect detection method and related device | |
| CN116844114A (en) | Helmet detection method and device based on YOLOv7-WFD model | |
| CN116757972A (en) | Fabric defect detection method capable of resisting influence of shadow noise | |
| CN115082741A (en) | Waste textile classifying method based on image processing | |
| CN118038443A (en) | Track section state detection method and system | |
| CN117217531A (en) | Factory safe operation risk data processing method based on artificial intelligence | |
| JP6722651B2 (en) | How to choose a compression algorithm depending on the type of image | |
| CN105163076B (en) | The substation breakers state video intelligent parser mapped using transversal scanning | |
| CN114792392B (en) | Method for simulating lightning precursor development path based on Markov chain | |
| CN112766144B (en) | Island infrastructure inspection detail difference judging method, device, equipment and medium | |
| CN114841895A (en) | Image shadow removing method based on bidirectional mapping network | |
| CN110211096B (en) | Haze image detection method | |
| CN114418087A (en) | Model quantification method, device and equipment based on optimized kl divergence |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220607 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |