CN112288000A - Intelligent ferrographic image identification method based on support vector machine - Google Patents

Intelligent ferrographic image identification method based on support vector machine Download PDF

Info

Publication number
CN112288000A
CN112288000A CN202011169534.9A CN202011169534A CN112288000A CN 112288000 A CN112288000 A CN 112288000A CN 202011169534 A CN202011169534 A CN 202011169534A CN 112288000 A CN112288000 A CN 112288000A
Authority
CN
China
Prior art keywords
ferrographic
image
support vector
vector machine
intelligent
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.)
Pending
Application number
CN202011169534.9A
Other languages
Chinese (zh)
Inventor
樊红卫
田鑫杭
高烁琪
马宁阁
刘琦
黄利平
曹现刚
张旭辉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian University of Science and Technology
Original Assignee
Xian University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xian University of Science and Technology filed Critical Xian University of Science and Technology
Priority to CN202011169534.9A priority Critical patent/CN112288000A/en
Publication of CN112288000A publication Critical patent/CN112288000A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/56—Extraction of image or video features relating to colour

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Genetics & Genomics (AREA)
  • Physiology (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a ferrographic image intelligent identification method based on a support vector machine, which combines the support vector machine and ferrographic analysis technology and comprises the following processes: the abrasive grain ferrographic image prepared by the ferrographic instrument is preprocessed, characteristic parameters of the abrasive grain ferrographic image are extracted and used as input of a support vector machine, parameters of the support vector machine are optimized by utilizing a genetic algorithm, an intelligent identification model based on the support vector machine is established, intelligent identification of the ferrographic image is achieved, and diagnosis efficiency and intelligent degree are improved.

Description

Intelligent ferrographic image identification method based on support vector machine
Technical Field
The invention relates to the field of mechanical fault diagnosis, in particular to a mechanical wear diagnosis direction, and specifically relates to a ferrographic image intelligent identification method based on a support vector machine.
Background
With the rapid development of science and technology and industry, mechanical equipment is more and more complex and develops towards automation and intellectualization. In the working process of mechanical equipment, abrasion is a ubiquitous phenomenon, and mechanical parts can affect the performance and the service life of the equipment after being abraded for a long time, so that equipment faults are caused. Once the equipment is out of order, various losses can be caused, for example, maintenance cost required by mechanical equipment, indirect loss caused by shutdown due to equipment failure, and serious and even casualties can be caused. Therefore, in order to ensure safe and effective operation of the core equipment, the mechanical equipment should be diagnosed and predicted to prevent serious faults.
When mechanical equipment runs, the main functions of lubricating oil are lubrication, rust prevention and auxiliary cooling, and the lubricating oil is an indispensable part. During the operation of mechanical equipment, the mechanical parts rub against each other, which can generate a large amount of abrasive particles and enter the lubrication system of the equipment. Because the abrasive grains generated by different friction states of the mechanical parts have different characteristics such as shapes, colors, textures and the like, the abrasive grains contain a large amount of information which can reflect the wear condition of the mechanical parts. Therefore, the method for diagnosing and predicting the faults of the mechanical equipment is to extract the lubricating oil of the mechanical equipment by utilizing a ferrographic analysis technology, analyze abrasive particles in the lubricating oil and study the relation between the lubricating oil and the faults of the mechanical equipment. Although ferrographic analysis is effective in fault diagnosis, it is not easy to extract information on abrasive grains related to equipment, and the traditional method needs personnel with relevant working experience in the field to observe and judge, and judge the type of the abrasive grains through characteristics such as shape, color, texture and the like so as to analyze the operation state of the machine. This method is dependent on the experience level of the staff, is of limited efficiency and is subject to personal factors. At present, how to process ferrographic images is more important, and a subsequent intelligent diagnosis method is lacked.
In recent years, machine learning methods have been widely used in the field of fault diagnosis of mechanical equipment. The support vector machine provides a feasible scheme for intelligent fault diagnosis. Compared with a deep learning method, the support vector machine is more suitable for intelligent identification of small samples, and because the time for manufacturing qualified ferrographic image samples by using a ferrographic analysis technology is long, a large number of effective samples are often difficult to obtain in a short time, the support vector machine is an ideal solution for intelligent identification of ferrographic images.
Disclosure of Invention
The invention aims to provide a ferrographic image intelligent identification method based on a support vector machine, which aims to solve the problems in the prior art, realizes an intelligent diagnosis process of automatically extracting and identifying ferrographic image characteristics, has high identification efficiency and high diagnosis precision, is suitable for online monitoring and early warning of gear faults of mechanical equipment, and can effectively reduce the fault occurrence rate of the mechanical equipment so as to improve the utilization rate of the gear faults of the mechanical equipment.
In order to achieve the purpose, the invention adopts the following technical scheme:
an intelligent ferrographic image identification method based on a support vector machine comprises the following steps;
1) constructing a ferrographic image data set: collecting an oil sample of mechanical equipment, collecting a wear particle ferrographic image by utilizing a ferrographic technology, preprocessing the wear particle ferrographic image, expanding the ferrographic image, establishing a ferrographic image data set, and dividing the ferrographic image data set into a training set and a testing set;
2) and (3) ferrographic image data set normalization: extracting image features in the ferrographic image data set in the step 1), and normalizing each image of the ferrographic image data set;
3) constructing an intelligent recognition model: constructing an intelligent identification model for the ferrographic image data set normalized in the step 2) by using a support vector machine;
4) parameter optimization of the intelligent recognition model: optimizing the parameters of the support vector machine in the intelligent recognition model in the step 3), determining the parameters of the support vector machine, and obtaining the optimized intelligent recognition model;
5) intelligent ferrographic image identification based on a support vector machine: collecting lubricating oil of mechanical equipment in an unknown state, preparing a ferrographic image by using a ferrographic technology, inputting the ferrographic image into the intelligent identification model in the step 4) to classify faults of the ferrographic image, and completing intelligent identification of the ferrographic image based on a support vector machine.
Further, the preprocessing of the ferrographic image in the step 1) includes grayscale image segmentation, binarization and morphological open-close operation, firstly, in order to distinguish the abrasive grain image from the background, the originally bright color background is grayed by a weighting method to obtain a grayscale image, and the formula of the weighting method is as follows:
Gray＝(R*0.3+G*0.59+B*0.11)/3
wherein: gray is the pixel value of the processed image; r, G, B is the RGB color space value of a pixel;
the method comprises the steps of realizing image segmentation by using a three-section threshold segmentation method on the basis of a gray map, extracting a foreground part in the gray map, endowing the extracted pixel points and the rest pixel points with different pixel values, realizing binarization, eliminating fine particles of the binary map by using morphological open operation after the ferrographic image is binarized, filling the hole of the image by using closed operation, and finishing image preprocessing.
Further, in the step 1), the ferrographic image is subjected to four operations of translation, rotation, contrast enhancement and turning in a data enhancement mode, so that the number of the ferrographic images is increased, and the expansion is completed.
Further, the support vector machine model in step 2) takes the shape parameters and texture parameters of the ferrographic image as the extracted image features, and realizes feature extraction for the ferrographic image, and the specific steps are as follows:
2.1) extracting three shape parameters of a ferrographic image: the ratio of perimeter, area, major and minor axes; drawing an area-perimeter point diagram generated by combining two shape parameters, and separating symbols of different types of abrasive particles by adopting a straight line to realize linear feature extraction;
2.2) extracting four texture parameters of the ferrographic image by utilizing a gray level co-occurrence matrix: energy, entropy, moment of inertia, local stability;
the formula for calculating the gray value of the pixel point by the gray level co-occurrence matrix is as follows:
P(i，j)＝#{[(x1，y1)，(x2，y2)]∈M*N|f(x1，y1)＝i，f(x2，y2)＝j)
the energy is calculated by the formula: e (d, theta) ═ Σi,j{i,j|d,θ}2
The formula for calculating the entropy is: h (d, theta) ═ Σi,j{P(i,j|d,θ)}×log10{P(i,j|d,θ)}
The calculation formula of the moment of inertia is as follows: i (d, theta) ═ Σi,j(i,j)2P(i,j|d,θ)
The calculation formula of the local stability is as follows:
Figure BDA0002745316950000031
wherein (x)1,y1),(x2,y2) For a pair of pixel points in a certain direction, i, j represents the gray value of the pair of pixel points, M, N represents the width and height of the ferrographic image, P (i, j) represents the probability that the gray values of the pair of pixel points are i and j respectively, d is the distance between two pixel points, and theta is the included angle between the two pixel points and the abscissa axis.
Further, the normalization processing in the step 2) is realized through a polynomial kernel function, the support vector machine selects the polynomial kernel function to map the original space to a high-dimensional space, and an optimal separation hyperplane is constructed in the high-dimensional space, so that the nonlinear data in the original space is segmented, and the normalization is realized.
Further, the intelligent identification model in the step 3) is specifically constructed as follows:
3.1) dividing a training set and a test set of a ferrographic image data set into equal L groups;
3.2) parameters are given, a training set is used for constructing the model, and a testing set is used for checking the accuracy of the model, wherein the parameters comprise a penalty factor c and a kernel function self-carried parameter g.
Further, the parameter optimization in the step 4) realizes the parameter optimization of the support vector machine through a genetic algorithm, wherein the parameters include a penalty factor c and a parameter g carried by a kernel function, the penalty factor c is a regularization parameter and limits the importance of each point, and g is used for controlling the width of a gaussian kernel, and the parameter optimization process based on the genetic algorithm specifically comprises the following steps:
4.1) setting a support vector machine parameter grid: setting the same L group values as the training set and the test set for c and g respectively;
4.2) parameter coding: transforming the parameters c and g into binary bit strings in a binary coding mode;
4.3) setting the initial population: setting the initial population size to be Q;
4.4) calculating a fitness function: converting the binary bit string in the step 4.2) into a real number code, and calculating a fitness function;
4.5) genetic manipulation:
4.5.1) selection operation: selecting parents for participating in crossing and mutation operations through a roulette strategy, and calculating the size of a relative adaptive value according to the adaptive value of each individual;
4.5.2) interleaving: randomly selecting one bit in the binary bit string in the 4.2) by using a single-point hybridization mode, and exchanging the binary bit string behind the bit in the father body according to the cross probability to generate a new individual;
4.5.3) mutation operation: bit-wise negating the binary bit string of the selected individual according to the variation probability to prevent the algorithm from falling into local optimum;
4.6) combination verification: and (4) sequentially using the combination of the two parameters g and c to construct different support vector machine models, calculating the accuracy of the models, selecting a group with the highest model accuracy as the optimal parameters of the support vector machine, and completing parameter optimization.
Compared with the prior art, the invention has the following beneficial technical effects:
1) aiming at the problem of training ferrographic image samples by a support vector machine model, the invention adopts a data enhancement method to efficiently obtain a large amount of qualified ferrographic image data. And graying the obtained ferrographic color sample by utilizing a gray processing technology to obtain a processed gray image sample. The gray level threshold segmentation method is utilized to convert the gray level image into the binary image to obtain the gray level binary image, the pixel points of the image are divided into a plurality of regions, and the regions have the same attribute, so that the calculation is simpler, and the calculation speed is improved.
2) The support vector machine method adopted by the invention has the advantages of high identification efficiency and high diagnosis precision, is suitable for monitoring and early warning of the abrasion fault of the mechanical equipment, and can effectively reduce the fault occurrence rate of the mechanical equipment so as to improve the utilization rate of the mechanical equipment.
Drawings
FIG. 1 is a flow chart of the intelligent ferrographic image identification method based on a support vector machine in the invention;
FIG. 2 is a flow chart of ferrographic image preparation;
FIG. 3 is an image pre-processing process;
FIG. 4 is a diagram of image preprocessing effect;
FIG. 5 is a development program software interface;
fig. 6 is a program classification result interface.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
referring to fig. 1, the intelligent iron spectrum image identification method based on the support vector machine mainly comprises the following steps:
1) the method comprises the steps of collecting oil after the gear transmission system operates for a period of time by using a gear transmission system built in a laboratory, and preparing a ferrograph image by using a ferrograph technology.
The gear transmission system is driven by a variable frequency motor, a magnetic powder brake simulates a load, and comprises a secondary planetary gear train and a secondary straight gear reducer, the secondary planetary gear train and the secondary straight gear reducer are lubricated by using 600XP150 gear oil, after collection of lubricating oil is completed, a YTF-8 analytical ferrography platform is used for analyzing the collected oil, and a process for specifically preparing a ferrography image is shown in an attached figure 2. The first step is as follows: and extracting oil liquid. The step is an important link, the representative oil liquid is ensured, the part of the oil liquid extracted, the sampling time interval, the sampling normalization and the like are required to be noticed during sampling, and the extracted oil liquid is required to be recorded in detail. The second step is that: and (6) oil sample treatment. For the extracted oil, due to the action of gravity, abrasive particles in the oil can naturally settle, so that in order to prepare a proper ferrograph sheet, the extracted oil needs to be heated and oscillated before the ferrograph sheet is prepared, and meanwhile, an oil sample needs to be diluted. The third step: and (5) preparing the iron sheet. The manufacture of the iron sheet is completed on an iron spectrometer, firstly, a proper glass substrate is selected, the substrate is placed, an oil sample is conveyed, the substrate is cleaned after the oil sample is conveyed, and then the iron sheet is taken out and dried. The fourth step: and (4) observing through a microscope. And observing the settled abrasive particles on the substrate by using a microscope, and adjusting the visual angle of the microscope to enable the abrasive particles to be displayed in the center of the visual field as much as possible. The fifth step: and collecting an image. And (4) observing the abrasive particles in the ferrographic sheet by using a microscope, collecting ferrographic images by using a computer, and finishing recording and storing.
2) And manually classifying the obtained ferrographic images according to categories, numbering the ferrographic images, and taking 123 training samples and 45 test samples according to category proportions.
3) And preprocessing the acquired ferrographic image, including graying, binarization and morphological opening and closing operation.
The ferrographic image is intelligently identified, relevant features are required to be extracted to serve as identification standards, the ferrographic image is required to be preprocessed for facilitating subsequent feature extraction of the ferrographic image, the process is as shown in the attached figure 3, the ferrographic image conversion method comprises the step of converting a ferrographic image into a gray-scale image by a weighting method, and the weighting method is used for calculating a formula:
a weighting method: gray ═ 0.3+ G × (0.59 + B × (0.11)/3
Wherein: gray is the pixel value of the processed image; r, G, B is the RGB color space value of a pixel.
Converting the gray-scale image into a binary image, and performing morphological opening and closing operation on the binary image, wherein the opening operation is to corrode and expand firstly, and the effect is to smooth the image contour, eliminate small abrasive particles or noise points and eliminate fine burrs; the closed operation is to expand first and then corrode, and has the effect of making up the gap between two adjacent targets and filling up small cavities.
And finally, dividing the pixel points into a plurality of classes according to the gray level by utilizing a gray binary image so as to divide a pixel set, wherein the purpose is to conveniently extract texture features subsequently, and FIG. 4 is an effect image after pretreatment.
4) And extracting the features in the preprocessed ferrographic image.
Because the ferrographic image needs to be intelligently identified and relevant features need to be extracted as identification standards, the method extracts 7 feature parameters including area, perimeter, major and minor axes, energy, entropy, moment of inertia and correlation. The feature matrix extracted at last is:
(xi,yi)(i＝1,2,...,n)
in the formula: x is the number ofi-a feature vector; y isi-a two-class label.
Wherein xiIs (x)1i,x2i,...,xmi)TVector of (2), label y of two classesiIs of the value [ -1,1 [)]I.e. each sample contains m (here 7) features and a label yi. The two values of n are respectively the training sample 123 and the test sample 45.
5) And learning the characteristics of the extracted training samples by using a support vector machine.
The matrix of the extracted features is normalized and then used as the input of a support vector machine, a polynomial kernel function with the best effect is selected after a kernel function of the support vector machine is tested, the support vector machine has two parameters which need to be adjusted, namely a punishment parameter c and a kernel function parameter g, the initial population size Q is set to be 20, the cross probability is 40%, the variation probability is 85%, the two parameters of the support vector machine are optimized by using a genetic algorithm, and an intelligent identification model is obtained after learning. Because the support vector machine solves the problem of two-classification, and the fault diagnosis is the problem of multi-classification, a one-to-one method is adopted to train a plurality of support intelligent identifications (6 models in the case), so that the support vector machine is realized to solve the problem of multi-classification.
6) After an intelligent recognition model based on a support vector machine is obtained, 45 ferrographic image test samples are taken, image preprocessing, feature extraction and normalization are also carried out, and features of the ferrographic image test samples are classified by the support vector machine, so that intelligent recognition of the ferrographic image is realized.
7) With the help of designed intelligent recognition software, see fig. 5 and fig. 6, the trained model performance is detected by utilizing a ferrographic sample set, and the result is shown in the following table 1.
TABLE 1 model Final Properties
Figure BDA0002745316950000081
While the invention has been described in connection with specific embodiments thereof, it will be understood that these should not be construed as limiting the scope of the invention, which is defined in the following claims, as any modification which may vary from the scope of the invention.

Claims (7)

1. An intelligent ferrographic image identification method based on a support vector machine is characterized by comprising the following steps;
1) constructing a ferrographic image data set: collecting an oil sample of mechanical equipment, collecting a wear particle ferrographic image by utilizing a ferrographic technology, preprocessing the wear particle ferrographic image, expanding the ferrographic image, establishing a ferrographic image data set, and dividing the ferrographic image data set into a training set and a testing set;
2) and (3) ferrographic image data set normalization: extracting image features in the ferrographic image data set in the step 1), and normalizing each image of the ferrographic image data set;
3) constructing an intelligent recognition model: constructing an intelligent identification model for the ferrographic image data set normalized in the step 2) by using a support vector machine;
4) parameter optimization of the intelligent recognition model: optimizing the parameters of the support vector machine in the intelligent recognition model in the step 3), determining the parameters of the support vector machine, and obtaining the optimized intelligent recognition model;
5) intelligent ferrographic image identification based on a support vector machine: collecting lubricating oil of mechanical equipment in an unknown state, preparing a ferrographic image by using a ferrographic technology, inputting the ferrographic image into the intelligent identification model in the step 4) to classify faults of the ferrographic image, and completing intelligent identification of the ferrographic image based on a support vector machine.
2. The intelligent ferrographic image identification method based on the support vector machine as claimed in claim 1, wherein the preprocessing of the ferrographic image in step 1) includes grayscale image segmentation, binarization and morphological open-close operation, and in order to distinguish the abrasive grain image from the background, the originally bright color background is grayed by a weighting method to obtain a grayscale image, and the formula of the weighting method is as follows:
Gray＝(R*0.3+G*0.59+B*0.11)/3
wherein: gray is the pixel value of the processed image; r, G, B is the RGB color space value of a pixel;
the method comprises the steps of realizing image segmentation by using a three-section threshold segmentation method on the basis of a gray map, extracting a foreground part in the gray map, endowing the extracted pixel points and the rest pixel points with different pixel values, realizing binarization, eliminating fine particles of the binary map by using morphological open operation after the ferrographic image is binarized, filling the hole of the image by using closed operation, and finishing image preprocessing.
3. The intelligent ferrographic image identification method based on the support vector machine as claimed in claim 1, wherein the ferrographic image in step 1) is subjected to four operations of translation, rotation, contrast enhancement and inversion by using a data enhancement mode, the number of ferrographic images is increased, and the expansion is completed.
4. The intelligent ferrographic image identification method based on the support vector machine according to claim 1, wherein the support vector machine model in step 2) takes shape parameters and texture parameters of the ferrographic image as the extracted image features, and implements feature extraction for the ferrographic image, specifically comprising the steps of:
2.1) extracting three shape parameters of a ferrographic image: the ratio of perimeter, area, major and minor axes; drawing an area-perimeter point diagram generated by combining two shape parameters, and separating symbols of different types of abrasive particles by adopting a straight line to realize linear feature extraction;
2.2) extracting four texture parameters of the ferrographic image by utilizing a gray level co-occurrence matrix: energy, entropy, moment of inertia, local stability;
the formula for calculating the gray value of the pixel point by the gray level co-occurrence matrix is as follows:
P(i，j)＝#{[(x1，y1)，(x2，y2)]∈M*N|f(x1，y1)＝i，f(x2，y2)＝j)
the energy is calculated by the formula: e (d, theta) ═ Σi,j{i,j|d,θ}2
The formula for calculating the entropy is: h (d, theta) ═ Σi,j{P(i,j|d,θ)}×log10{P(i,j|d,θ)}
The calculation formula of the moment of inertia is as follows: i (d, theta) ═ Σi,j(i,j)2P(i,j|d,θ)
The calculation formula of the local stability is as follows:
Figure FDA0002745316940000021
wherein (x)1,y1),(x2,y2) For a pair of pixel points in a certain direction, i, j represents the gray value of the pair of pixel points, M, N represents the width and height of the ferrographic image, P (i, j) represents the probability that the gray values of the pair of pixel points are i and j respectively, d is the distance between two pixel points, and theta is the included angle between the two pixel points and the abscissa axis.
5. The intelligent ferrographic image identification method based on the support vector machine as claimed in claim 1, wherein the normalization process in step 2) is implemented by a polynomial kernel function, the support vector machine selects the polynomial kernel function to map the original space to a high-dimensional space, and an optimal separation hyperplane is constructed in the high-dimensional space, so that the nonlinear data in the original space is segmented to realize normalization.
6. The intelligent ferrographic image recognition method based on the support vector machine according to claim 1, wherein the intelligent recognition model in the step 3) is specifically constructed as follows:
3.1) dividing a training set and a test set of a ferrographic image data set into equal L groups;
3.2) parameters are given, a training set is used for constructing the model, and a testing set is used for checking the accuracy of the model, wherein the parameters comprise a penalty factor c and a kernel function self-carried parameter g.
7. The intelligent ferrographic image identification method based on the support vector machine according to claim 6, wherein the parameter optimization in step 4) is implemented by a genetic algorithm to optimize parameters of the support vector machine, wherein the parameters include a penalty factor c and a kernel function self-carried parameter g, the penalty factor c is a regularization parameter and limits the importance of each point, and g is used to control the width of a gaussian kernel, and the genetic algorithm-based parameter optimization process specifically includes:
4.1) setting a support vector machine parameter grid: setting the same L group values as the training set and the test set for c and g respectively;
4.2) parameter coding: transforming the parameters c and g into binary bit strings in a binary coding mode;
4.3) setting the initial population: setting the initial population size to be Q;
4.4) calculating a fitness function: converting the binary bit string in the step 4.2) into a real number code, and calculating a fitness function;
4.5) genetic manipulation:
4.5.1) selection operation: selecting parents for participating in crossing and mutation operations through a roulette strategy, and calculating the size of a relative adaptive value according to the adaptive value of each individual;
4.5.2) interleaving: randomly selecting one bit in the binary bit string in the 4.2) by using a single-point hybridization mode, and exchanging the binary bit string behind the bit in the father body according to the cross probability to generate a new individual;
4.5.3) mutation operation: bit-wise negating the binary bit string of the selected individual according to the variation probability to prevent the algorithm from falling into local optimum;
4.6) combination verification: and (4) sequentially using the combination of the two parameters g and c to construct different support vector machine models, calculating the accuracy of the models, selecting a group with the highest model accuracy as the optimal parameters of the support vector machine, and completing parameter optimization.
CN202011169534.9A 2020-10-27 2020-10-27 Intelligent ferrographic image identification method based on support vector machine Pending CN112288000A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011169534.9A CN112288000A (en) 2020-10-27 2020-10-27 Intelligent ferrographic image identification method based on support vector machine

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011169534.9A CN112288000A (en) 2020-10-27 2020-10-27 Intelligent ferrographic image identification method based on support vector machine

Publications (1)

Publication Number Publication Date
CN112288000A true CN112288000A (en) 2021-01-29

Family

ID=74373460

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011169534.9A Pending CN112288000A (en) 2020-10-27 2020-10-27 Intelligent ferrographic image identification method based on support vector machine

Country Status (1)

Country Link
CN (1) CN112288000A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115115340A (en) * 2022-06-30 2022-09-27 广州高新工程顾问有限公司 Intelligent digital supervision multi-target cooperative control method for engineering project

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108389216A (en) * 2018-02-06 2018-08-10 西安交通大学 Local auto-adaptive threshold segmentation method towards on-line ferrograph image wear Particles Recognition
US20200271720A1 (en) * 2020-05-09 2020-08-27 Hefei University Of Technology Method for diagnosing analog circuit fault based on vector-valued regularized kernel function approximation

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108389216A (en) * 2018-02-06 2018-08-10 西安交通大学 Local auto-adaptive threshold segmentation method towards on-line ferrograph image wear Particles Recognition
US20200271720A1 (en) * 2020-05-09 2020-08-27 Hefei University Of Technology Method for diagnosing analog circuit fault based on vector-valued regularized kernel function approximation

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
邱丽娟: "基于支持向量机的铁谱磨粒图像识别技术", 中国优秀硕士论文电子期刊网, pages 3 - 5 *
颜欢;韩超;张贤明;: "铁谱技术智能化应用研究进展", 重庆工商大学学报(自然科学版), no. 05 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115115340A (en) * 2022-06-30 2022-09-27 广州高新工程顾问有限公司 Intelligent digital supervision multi-target cooperative control method for engineering project
CN115115340B (en) * 2022-06-30 2023-11-24 广州高新工程顾问有限公司 Intelligent digital supervision multi-target cooperative control method for engineering project

Similar Documents

Publication Publication Date Title
Wang et al. Integrated model of BP neural network and CNN algorithm for automatic wear debris classification
CN112254964A (en) Rolling bearing fault diagnosis method based on rapid multi-scale convolution neural network
CN105678332A (en) Converter steel-making endpoint determination method and system based on flame image CNN recognizing and modeling process
CN112734734A (en) Railway tunnel crack detection method based on improved residual error network
CN112257530A (en) Rolling bearing fault diagnosis method based on blind signal separation and support vector machine
CN114358124B (en) New fault diagnosis method for rotary machinery based on deep countermeasure convolutional neural network
CN110208124A (en) The development approach of mechanical wear system based on Abrasive Wear Mechanism
CN111753891A (en) Rolling bearing fault diagnosis method based on unsupervised feature learning
CN108509701A (en) A kind of direct intelligent diagnosing method of rotating machinery fault based on vibration signal
CN116883393B (en) Metal surface defect detection method based on anchor frame-free target detection algorithm
CN112907561A (en) Notebook appearance flaw detection method based on deep learning
CN111402236A (en) Hot-rolled strip steel surface defect grading method based on image gray value
CN112231971A (en) Blast furnace fault diagnosis method based on relative integral trend diffusion fault sample generation
Wang et al. Automated 3D ferrograph image analysis for similar particle identification with the knowledge-embedded double-CNN model
CN112288000A (en) Intelligent ferrographic image identification method based on support vector machine
CN115587290A (en) Aero-engine fault diagnosis method based on variational self-coding generation countermeasure network
CN114997211A (en) Cross-working-condition fault diagnosis method based on improved countermeasure network and attention mechanism
CN114445397A (en) Strip steel defect detection method based on shallow neural network
CN113469252A (en) Extra-high voltage converter valve operation state evaluation method considering unbalanced samples
CN113159225A (en) Multivariable industrial process fault classification method
Liu et al. Unsupervised segmentation of wear particle’s image using local texture feature
He et al. CBAM-YOLOv5: a promising network model for wear particle recognition
Gupta et al. Online Gear Wear Particle Detection and Categorization Using a Convolutional Neural Network Algorithm Integrated with Cascade Classifier
Wang et al. Wear debris classification and quantity and size calculation using convolutional neural network
CN112149804B (en) Novel convolutional neural network model and application

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