CN117952482B - Product quality accident grading method and system based on convolutional neural network - Google Patents
Product quality accident grading method and system based on convolutional neural network Download PDFInfo
- Publication number
- CN117952482B CN117952482B CN202410349918.0A CN202410349918A CN117952482B CN 117952482 B CN117952482 B CN 117952482B CN 202410349918 A CN202410349918 A CN 202410349918A CN 117952482 B CN117952482 B CN 117952482B
- Authority
- CN
- China
- Prior art keywords
- accident
- quality
- product
- quality accident
- image
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000012216 screening Methods 0.000 claims abstract description 13
- 238000003908 quality control method Methods 0.000 claims abstract description 10
- 238000011156 evaluation Methods 0.000 claims description 65
- 238000013139 quantization Methods 0.000 claims description 42
- 230000008439 repair process Effects 0.000 claims description 36
- 238000004458 analytical method Methods 0.000 claims description 22
- 238000004364 calculation method Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 17
- 238000011002 quantification Methods 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 6
- 238000011158 quantitative evaluation Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 229930014626 natural product Natural products 0.000 claims 1
- 238000013145 classification model Methods 0.000 abstract description 6
- 230000006870 function Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 238000010206 sensitivity analysis Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/771—Feature selection, e.g. selecting representative features from a multi-dimensional feature space
-
- 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
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Educational Administration (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of quality control, and particularly discloses a product quality accident grading method and system based on a convolutional neural network, wherein the method comprises the following steps: collecting product quality accident sample data, and grading the accident severity of each quality accident product; extracting each image characteristic element of each quality accident product, and screening effective characteristic elements; and according to each effective characteristic element, training to obtain a product quality accident grading model, and grading the product quality accidents. According to the invention, the quality accident picture of the quality accident product is compared with the initial picture, the economic loss caused by the quality accident of the product is analyzed, the accident is scientifically and reasonably classified, meanwhile, the quality accident severity and the image characteristic elements are utilized for model training, the quality accident of the product is classified through the quality accident classification model, the subjective error caused by manual judgment is reduced, and the objectivity and consistency of the quality accident assessment result are improved.
Description
Technical Field
The invention relates to the technical field of quality control, in particular to a product quality accident grading method and system based on a convolutional neural network.
Background
The product quality accident classification refers to a method for classifying quality problems or accidents of products in the production and use processes, and in actual production, the quality accidents of the products can cause different degrees of influence, such as production stagnation, economic loss, safety accidents and the like. The traditional product quality accident grading method is more dependent on manual experience, has stronger subjectivity and lacks objectivity and accuracy, and cannot identify and evaluate the accident severity timely and accurately, so that the quality management decision of enterprises is affected.
For example, the invention patent with publication number CN112241832B discloses a product quality grading evaluation standard design method and system, comprising: acquiring a data set consisting of the whole process parameters and corresponding mechanical performance indexes; screening the technological parameters based on mutual information of the technological parameters and the mechanical performance indexes, and constructing a feature subset and a mechanical performance prediction model; performing multi-output sensitivity analysis on the mechanical performance prediction model to obtain estimation of sensitivity indexes of all process parameters in the characteristic subset, and obtaining a vector omega formed by the sensitivity indexes of all the process parameters; updating a column vector X formed by the feature subsets to omega X; and carrying out spectral clustering on the sample set corresponding to omega X to divide sample categories, and establishing corresponding relations between the sample categories and a final quality index narrow window according to different sample categories to form corresponding quality grading evaluation standards. The application simulates the quality index range corresponding to various technological parameters in advance based on actual production history big data, thereby facilitating the quantitative management and control of the quality performance of the product for steel enterprises.
Based on the above scheme, there are some disadvantages in quality control at present, which are specifically embodied in the following aspects: (1) The current quality accident classification has no unified standard, the analysis of the quality accidents of the products is not comprehensive enough, for example, the analysis from the aspects of the images of the quality accidents of the products and the economic losses is lacking, and the accurate classification of the quality accidents of the products is not possible.
(2) The current quality accident grading mainly depends on human judgment, subjective errors are often introduced, and the non-uniform result of the quality accident grading of the product is easily caused.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a product quality accident grading method and system based on a convolutional neural network, which can effectively solve the problems related to the background art.
In order to achieve the above purpose, the invention is realized by the following technical scheme: the first aspect of the invention provides a product quality accident grading method based on a convolutional neural network, which comprises the following steps: and collecting sample data of the quality accidents of the products, marking the products with the quality accidents as the quality accident products, and analyzing to obtain the accident severity assessment index of the quality accident products.
Extracting each image characteristic element of each quality accident product, analyzing to obtain an accident correlation degree evaluation value of each image characteristic element according to the accident severity evaluation index of each quality accident product, and screening the effective characteristic elements according to the accident correlation degree evaluation value of each image characteristic element to obtain each effective characteristic element.
According to each effective characteristic element, training to obtain a product quality accident grading model, obtaining a target product, and grading the product quality accidents through the product quality accident grading model.
As a further method, the analysis is performed to obtain an accident severity assessment index of each quality accident product, and the specific analysis process is as follows: according to the product quality accident sample data, extracting an initial image and a quality accident image of each quality accident product, comparing the initial image and the quality accident image of each quality accident product, and analyzing to obtain a damage degree quantification index of each quality accident product.
And obtaining the repair cost of each quality accident product, counting the repair time cost of each quality accident product, and processing to obtain the influence degree quantification index of each quality accident product.
And comprehensively analyzing to obtain accident severity assessment indexes of the accident products of all qualities.
As a further method, the quantitative indexes of the damage degree of each quality accident product are specifically quantitative evaluation data obtained by comparing and analyzing the initial image and the quality accident image of each quality accident product, and are used for quantitatively evaluating the damage degree of each quality accident product and providing data basis for quality accident classification.
As a further method, the analysis obtains an accident-related degree evaluation value of each image characteristic element, and the specific analysis process is as follows: according to the quality accident images of the quality accident products, the image entropy values and the image contrast indexes of the quality accident products are obtained through processing, and the characteristic quantized values of the image characteristic elements of the quality accident products are obtained through comprehensive analysis.
And analyzing and obtaining the accident correlation degree evaluation value of each image characteristic element according to the accident severity evaluation index of each quality accident product and the characteristic quantification value of each image characteristic element of each quality accident product.
As a further method, the effective feature element screening is performed according to the accident correlation degree evaluation value of each image feature element, so as to obtain each effective feature element, and the specific analysis process is as follows: and comparing the accident correlation degree evaluation value of each image characteristic element with a correlation degree evaluation threshold value stored in a quality control database, and marking the image characteristic element as an effective characteristic element if the accident correlation degree evaluation value of a certain image characteristic element is larger than or equal to the correlation degree evaluation threshold value.
As a further method, the accident severity assessment index of each quality accident product is specifically a quantitative assessment value obtained by analyzing the damage degree quantitative index and the influence degree quantitative index of each quality accident product, and is used for quantitatively assessing the accident severity of each quality accident product and providing a data basis for quality accident classification.
As a further method, the feature quantization value of each image feature element of each quality accident product is specifically quantization data of each image feature element obtained by performing quantization analysis on each image feature element from two dimensions of an image entropy value and an image contrast index, and the quantization data are used for quantitatively evaluating the severity of each image feature element and providing a data basis for screening effective feature elements.
As a further method, the accident severity assessment index of each quality accident product is specifically calculated as: in the above, the ratio of/> Represents the/>Accident severity assessment index for individual quality accident products,/>Representing natural constant,/>Represents the/>Quantitative index of damage degree of individual quality accident products,/>Represents the/>Quantitative index of influence degree of individual quality accident products,/>Accident severity influencing factors corresponding to the set damage degree quantification indexes,/>, are representedAnd the accident severity influence factor corresponding to the set influence degree quantification index is represented.
As a further method, the characteristic quantization value of each image characteristic element of each quality accident product is specifically calculated as: in the above, the ratio of/> Represents the/>Product of personal quality accident/>Feature quantization value of each image feature element,/>Representing natural constant,/>Represents the/>Image entropy value of individual quality accident products,/>Represents the/>Image contrast index of individual quality accident products,/>Indicate the set/>Characteristic quantization influence factors of image entropy values corresponding to image characteristic elements,/>Indicate the set firstThe individual image feature elements correspond to feature quantization influencing factors of the image contrast index.
The second aspect of the invention provides a product quality accident grading system based on a convolutional neural network, which comprises: the quality accident grading module is used for collecting product quality accident sample data, marking products with quality accidents as quality accident products, and analyzing to obtain accident severity assessment indexes of the quality accident products.
The image feature element extraction module is used for extracting each image feature element of each quality accident product, analyzing and obtaining an accident correlation degree evaluation value of each image feature element according to the accident severity evaluation index of each quality accident product, and screening the effective feature elements according to the accident correlation degree evaluation value of each image feature element to obtain each effective feature element.
And the grading model training module is used for training to obtain a product quality accident grading model according to each effective characteristic element, obtaining a target product and grading the product quality accident through the product quality accident grading model.
The quality control database is used for storing product quality accident sample data, accident grades corresponding to the accident severity evaluation index intervals and related degree evaluation thresholds.
Compared with the prior art, the embodiment of the invention has at least the following advantages or beneficial effects:
(1) The invention provides a product quality accident grading method and system based on a convolutional neural network, which are used for comparing quality accident pictures of quality accident products with initial pictures, analyzing economic losses caused by the quality accidents of the products, scientifically and reasonably grading the accidents, carrying out model training by utilizing the severity degree of the quality accidents and image characteristic elements, grading the quality accidents of the products through a quality accident grading model, reducing subjective errors caused by manual judgment, and improving objectivity and consistency of quality accident assessment results.
(2) The invention helps to grade the quality accidents of the products by comparing the difference between the quality accident pictures and the initial pictures, and can rapidly evaluate the influence degree of the quality accidents by visual picture comparison, thereby providing basis for quality accident grading.
(3) According to the invention, the damage caused by the product quality accident is analyzed, including the product repair cost and the repair time cost, and the influence range and the severity of the accident can be evaluated by analyzing the damage caused by the product quality accident, so that the accident is scientifically and reasonably classified, and the accuracy of product accident classification is improved.
(4) According to the invention, the quality accident classification model is trained to classify the quality accident of the product, and the quality accident classification model is used to classify the quality accident of the product, so that subjective errors caused by manual judgment are reduced, and objectivity and consistency of the quality accident assessment result are improved.
Drawings
The invention will be further described with reference to the accompanying drawings, in which embodiments do not constitute any limitation of the invention, and other drawings can be obtained by one of ordinary skill in the art without inventive effort from the following drawings.
FIG. 1 is a schematic flow chart of the method of the present invention.
Fig. 2 is a schematic diagram of system module connection according to the present invention.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments, and all other embodiments obtained by those skilled in the art without making creative efforts based on the embodiments of the present invention are included in the protection scope of the present invention.
Referring to fig. 1, a first aspect of the present invention provides a product quality accident classification method based on a convolutional neural network, including: and collecting sample data of the quality accidents of the products, marking the products with the quality accidents as the quality accident products, and analyzing to obtain the accident severity assessment index of the quality accident products.
It should be understood that the products referred to in the specification refer to products having entities such as electronic products, appliances, living goods, and the like.
In this embodiment, the product quality accident sample data is accident sample data obtained by collecting products with quality accidents, including an initial image, a quality accident image, repair cost and repair time cost of each quality accident product. The product quality accident sample data is historical data of quality accident grading in the past and is used for training a quality accident grading model.
Specifically, the accident severity assessment index of each quality accident product is obtained through analysis, and the specific analysis process is as follows: according to the product quality accident sample data, extracting an initial image and a quality accident image of each quality accident product, comparing the initial image and the quality accident image of each quality accident product, and analyzing to obtain a damage degree quantification index of each quality accident product.
It should be understood that the initial image of each quality accident product is a planar image acquired before each quality accident product is put into use, and the quality accident image of each quality accident product is a planar image acquired after each quality accident product has a quality accident. By comparing the image differences of the products after the quality accident, the damage degree of the quality accident products can be estimated.
In a specific embodiment, the damage degree quantization index of each quality accident product can be obtained by not only deeply analyzing each performance parameter of the product through a data analysis tool and technology, so as to obtain the damage degree quantization index, but also collecting feedback of a user to the accident product, quantizing the damage degree of the product according to the severity degree of the feedback of the user, and further obtaining the damage degree quantization index through the following calculation method, disposing a plurality of pixel points, extracting the pixel value of each pixel point of an initial image of each quality accident product, extracting the pixel value of each pixel point corresponding to the quality accident image of each quality accident product, and comprehensively calculating the damage degree quantization index of each quality accident product, wherein the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Quantitative index of damage degree of individual quality accident products,/>Represents the/>Quality incident image of individual quality incident product/>The pixel values of the individual pixel points,Represents the/>Initial image of individual quality accident product/>Pixel value of each pixel point,/>Representing the set allowable deviation pixel value,/>Indicating the set product damage degree quantized correction factor,/>The number of each quality accident product is indicated,，/>Representing the total number of quality accident products,/>The number of each pixel point is indicated,，/>Representing the total number of pixels.
In a specific embodiment, the quality accident classification of the product is assisted by comparing the difference between the quality accident picture and the initial picture, and the influence degree of the quality accident can be rapidly estimated by visual picture comparison, so that a basis is provided for quality accident classification.
And obtaining the repair cost of each quality accident product, counting the repair time cost of each quality accident product, and processing to obtain the influence degree quantification index of each quality accident product.
It should be understood that the repair costs of each quality accident product refer to the total cost of repairing each quality accident product to normal use, and the repair time costs of each quality accident product refer to the total time taken for each quality accident product to normal use. The influence degree caused by the quality accidents of the products of each quality accident can be evaluated by analyzing the repair cost and the repair time cost of the product quality accidents.
It should be understood that the quantitative evaluation index of the influence degree of each quality accident product is specifically obtained by analyzing the repair cost and repair time cost of each quality accident product, and is used for evaluating the influence degree of the quality accident of each quality accident product and providing a data basis for quality accident classification.
In a specific embodiment, the quantitative index of the influence degree of each quality accident product can be obtained by collecting and analyzing the quality accident data generated in the past, the indexes of the frequency, the influence range, the influence degree and the like of the occurrence of the quality accident can be obtained, the detailed analysis can be performed on the influence of the quality accident, including the influence of the accident on the product performance, the service life and the safety, the influence on the user satisfaction and the like, the influence degree of the quality accident on the product can be obtained through analysis, the quantitative index of the influence degree of the quality accident on the product can be obtained through the following calculation method, the critical repair cost and the critical repair time cost of the predefined quality accident product can be respectively obtained, the quantitative index of the influence degree of each quality accident product can be comprehensively calculated, and the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Quantitative index of influence degree of individual quality accident products,/>Representing natural constant,/>Represents the/>Repair cost of individual quality Accident products,/>Represents the/>Repair time cost of individual quality Accident products,/>Representing a predefined critical repair cost,/>Representing a predefined critical repair time cost,/>Quantized weighting factor indicating influence level corresponding to set repair cost,/>A quantitative weighting factor indicating the degree of influence corresponding to the set repair time cost,Quantized weighting factor indicating influence degree corresponding to set repair cost unit value,/>And the influence degree quantization weight factor corresponding to the set repair time cost unit value is represented.
In a specific embodiment, the damage caused by the product quality accident is analyzed, including the product repair cost and repair time, and the influence range and severity of the accident can be evaluated by analyzing the damage caused by the product quality accident, so that the accident is scientifically and reasonably classified, and the accuracy of product accident classification is improved.
And comprehensively analyzing to obtain accident severity assessment indexes of the accident products of all qualities.
The quantitative evaluation value is obtained by analyzing the damage degree quantitative index and the influence degree quantitative index of each quality accident product, is used for quantitatively evaluating the accident severity of each quality accident product and provides a data basis for quality accident grading.
Specifically, the accident severity assessment index of each quality accident product can not only judge the severity of the current accident by comparing the severity of similar quality accidents in history, by reference and referencing, but also indirectly assess the severity of the accident by collecting and analyzing data related to the accident, such as the product percent of pass, customer satisfaction rate, return rate and the like, and can also be obtained by the following calculation modes, wherein the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Accident severity assessment index for individual quality accident products,/>Representing natural constant,/>Represents the/>Quantitative index of damage degree of individual quality accident products,/>Represents the/>Quantitative index of influence degree of individual quality accident products,/>Accident severity influencing factors corresponding to the set damage degree quantification indexes,/>, are representedAnd the accident severity influence factor corresponding to the set influence degree quantification index is represented.
The quantitative evaluation data is used for quantitatively evaluating the damage degree of each quality accident product and providing a data basis for quality accident grading.
It should be understood that, in this embodiment, the accident severity assessment index interval and the accident level are in a one-to-one correspondence, and the accident level of the quality accident product can be obtained by analyzing the accident severity assessment index interval to which the accident severity assessment index of the quality accident product belongs.
It should be understood that the accident level in this embodiment refers to the accident level of product quality, and is generally classified according to factors such as the impact range of the accident, economic loss, casualties, social influence, etc., including a slight accident level, a general accident level, a significant accident level, and a particularly significant accident level.
Extracting each image characteristic element of each quality accident product, analyzing to obtain an accident correlation degree evaluation value of each image characteristic element according to the accident severity evaluation index of each quality accident product, and screening the effective characteristic elements according to the accident correlation degree evaluation value of each image characteristic element to obtain each effective characteristic element.
It should be appreciated that each image feature element of each quality incident product, including damage area, damage depth, damage morphology, etc., may be quantified by two dimensions, an image entropy value and an image contrast index.
Specifically, accident correlation degree evaluation values of the image characteristic elements are obtained through analysis, and the specific analysis process is as follows: according to the quality accident images of the quality accident products, the image entropy values and the image contrast indexes of the quality accident products are obtained through processing, and the characteristic quantized values of the image characteristic elements of the quality accident products are obtained through comprehensive analysis.
It should be understood that the image entropy and image contrast index in this specification are only processed for quality incident images of respective quality incident products.
It should be understood that the image entropy value in this embodiment is a measure of the amount of information in an image, reflecting the randomness and uncertainty of the information in the image, and is typically used to describe the texture complexity or distribution of the information in image processing. The image contrast index is data obtained by quantifying the image contrast, is a measure for describing the degree of difference of light and dark areas in the image, and high contrast means that the image has obvious light and dark changes, while low contrast means that the image is uniform and the light and dark changes are not obvious. And quantifying the image characteristic elements through two dimensions of the image entropy value and the image contrast index to help analyze the correlation of each image characteristic element and the accident level.
In a specific embodiment, the entropy value of the image of each quality accident product can be obtained by converting the image from a space domain to a frequency domain, for example, using a Fast Fourier Transform (FFT), calculating the entropy value of each frequency component in the frequency domain, combining the entropy values to obtain the entropy value of the whole image, classifying the pixels or regions of the image by using a machine learning algorithm, calculating the entropy value according to the classification result, reflecting the classification uncertainty, converting the quality accident image of each quality accident product into a gray image, extracting the pixel value of each pixel point of the gray image, and performing gray level division (for example, 256 gray levels) on each pixel point according to the pixel value, simultaneously obtaining the total number of the pixel points of the quality accident image of each quality accident product and the number of the pixel points of each gray level, and comprehensively calculating the image entropy value of each quality accident product, wherein the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Image entropy value of individual quality accident products,/>Representing the set image entropy correction factor,/>A number representing each gray level is provided,，/>Representing the total number of gray levels,/>Represents the/>Product of personal quality accident/>Probability of individual gray levels. The probability of the gray level can be further obtained by analyzing the number of pixel points, and the specific calculation formula is as follows: in the above, the ratio of/> Represents the/>The total number of quality accident image pixels for each quality accident product,Represents the/>Product of personal quality accident/>Number of pixels for each gray level.
In a specific embodiment, the image contrast index of each quality accident product can be obtained by not only decomposing the image into components with different scales and directions through wavelet transformation, obtaining the contrast quantization index of the image through analyzing the contrast of the components, but also quantifying the image contrast through analyzing the edge strength of the image and measuring the edge sharpness of the image, and can be obtained through the following calculation modes, extracting the pixel value, the maximum pixel value and the minimum pixel value of each pixel point of the quality accident image of each quality accident product, and comprehensively calculating the image contrast index of each quality accident product, wherein the specific calculation expression is as follows: In which, in the process, Represents the/>Image contrast index of individual quality accident products,/>Representing the set image contrast index correction factor,/>Represents the/>Quality incident image of individual quality incident product/>Pixel value of each pixel point,/>Represents the/>Maximum pixel value of quality accident image of individual quality accident products,/>Represents the/>Minimum pixel value of quality accident image of individual quality accident products,/>Image contrast index influence factor corresponding to set pixel value variance is expressed by/>And an image contrast index influence factor corresponding to the unit deviation pixel value between the set maximum pixel value and minimum pixel value is represented.
Specifically, the feature quantization value of each image feature element of each quality accident product is specifically obtained by performing quantization analysis on each image feature element from two dimensions of an image entropy value and an image contrast index, and the quantization data is used for quantitatively evaluating the severity of each image feature element and providing a data basis for screening effective feature elements.
Specifically, the feature quantization value of each image feature element of each quality accident product can be obtained not only by extracting the main change direction of image data through a principal component analysis method so as to quantize the main features of an image, but also by extracting complex image feature elements through a deep learning model such as a convolutional neural network and the like, obtaining quantized indexes through training, and further by obtaining the following calculation modes, wherein the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Product of personal quality accident/>Feature quantization value of each image feature element,/>Represents the/>Image entropy value of individual quality accident products,/>Represents the/>Image contrast index of individual quality accident products,/>Indicate the set/>Characteristic quantization influence factors of image entropy values corresponding to image characteristic elements,/>Indicate the set/>The individual image feature elements correspond to feature quantization influencing factors of the image contrast index.
And analyzing and obtaining the accident correlation degree evaluation value of each image characteristic element according to the accident severity evaluation index of each quality accident product and the characteristic quantification value of each image characteristic element of each quality accident product.
It should be understood that, in this embodiment, the evaluation value of the accident correlation degree of each image feature element is specifically a quantized evaluation value obtained by performing correlation analysis on the accident severity evaluation index of each quality accident product and the feature quantization value of each image feature element of each quality accident product, and is used for quantitatively evaluating the correlation degree of each image feature element and the accident severity, so as to provide a data basis for screening of effective feature elements.
In a specific embodiment, the accident correlation degree evaluation value of each image feature element can simulate the quality accident occurrence process by using a computer through a simulation method to generate a quality accident image, and compare and analyze the quality accident image with an actual quality accident image, observe the change of the quality accident image feature element through adjusting parameters in the simulation so as to deduce the relation between the feature and the severity, and can also be obtained through analyzing the existing product quality accident case, matching the quality accident image with the loss caused by the quality accident, establishing the mapping relation between the feature and the severity, and the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Accident correlation degree evaluation value of each image characteristic element,/>Represents the/>Accident severity assessment index for individual quality accident products,/>Represents the/>Product of personal quality accident/>Feature quantization value of each image feature element,/>An average accident severity assessment index representing quality accident products,/>，/>Represents the/>Average feature quantization value of individual image feature elements,/>，/>Indicating the set accident related degree correction factor.
Specifically, effective characteristic elements are screened according to the accident correlation degree evaluation values of the image characteristic elements, so that the effective characteristic elements are obtained, and the specific analysis process is as follows: and comparing the accident correlation degree evaluation value of each image characteristic element with a correlation degree evaluation threshold value stored in a quality control database, and marking the image characteristic element as an effective characteristic element if the accident correlation degree evaluation value of a certain image characteristic element is larger than or equal to the correlation degree evaluation threshold value.
It should be understood that, in this embodiment, the correlation degree evaluation threshold is a critical index for evaluating the correlation degree, if the accident correlation degree evaluation value of a certain image feature element is greater than or equal to the correlation degree evaluation threshold, it indicates that the image feature element has a higher correlation degree with the accident correlation degree, otherwise, if the accident correlation degree evaluation value of a certain image feature element is less than the correlation degree evaluation threshold, it indicates that the image feature element has a lower correlation degree with the accident correlation degree.
According to each effective characteristic element, training to obtain a product quality accident grading model, obtaining a target product, and grading the product quality accidents through the product quality accident grading model.
In a specific embodiment, the accident classification model of the quality accident product is obtained by training according to each effective characteristic element, and particularly, the accident classification function model is built according to the correlation degree of each effective characteristic element and the accident severity, and the accident classification index of the quality accident product is obtained by analyzing the accident classification function model.
It should be appreciated that the product quality incident image is analyzed using convolutional neural networks in this embodiment.
It should be understood that, in this embodiment, the accident classification index is specifically quantized data obtained by analyzing the effective feature elements in the product quality accident image, which is used for measuring the severity of the product quality accident reflected by the effective feature elements, so as to provide data support for product quality accident classification.
In a specific embodiment, product quality accidents are classified according to product quality accident image data, products needing quality accident classification are marked as target products, each effective characteristic element in quality accident images of the target products is extracted, each effective characteristic element is quantitatively processed according to the image entropy value and the image contrast index to obtain a characteristic quantization value of each effective characteristic element, accident classification indexes of the target products are comprehensively calculated, the accident classification indexes can be obtained by summarizing quality accidents occurring in history and analyzing reasons, influence ranges and treatment measures of the accidents, so that a quality accident classification system based on experience is formed, classification can be performed according to factors such as severity, influence ranges, economic loss and social influence of the accidents, and the like, the specific function equation is obtained by the following accident classification function model: in the above, the ratio of/> Accident grading index representing target product,/>Represent the firstFeature quantization value of each effective feature element,/>Indicate the set/>The grading index corresponding to each effective characteristic element affects the weight factor,/>Number representing each effective characteristic element,/>，/>Representing the total number of valid feature elements.
In this embodiment, the accident classification index of the target product is matched with the accident level corresponding to each set accident classification index interval, so as to obtain the accident level of the target product.
In a specific embodiment, the quality accident classification model is trained, the convolutional neural network is utilized to extract image characteristic elements, the quality accident of the product is classified, the quality accident classification model is utilized to classify the quality accident of the product, subjective errors caused by manual judgment are reduced, and objectivity and consistency of the quality accident assessment result are improved.
Referring to fig. 2, a second aspect of the present invention provides a product quality accident grading system based on a convolutional neural network, comprising: the system comprises a quality accident grading module, an image characteristic element extraction module, a grading model training module and a quality control database.
The quality accident grading module is used for collecting product quality accident sample data, marking products with quality accidents as quality accident products, and analyzing to obtain accident severity assessment indexes of the quality accident products.
The image feature element extraction module is used for extracting each image feature element of each quality accident product, analyzing and obtaining an accident correlation degree evaluation value of each image feature element according to the accident severity evaluation index of each quality accident product, and screening effective feature elements according to the accident correlation degree evaluation value of each image feature element to obtain each effective feature element.
And the grading model training module is used for training to obtain a product quality accident grading model according to each effective characteristic element, obtaining a target product and grading the product quality accident through the product quality accident grading model.
The quality control database is used for storing product quality accident sample data, accident grades corresponding to the accident severity evaluation index intervals and related degree evaluation thresholds.
In a specific embodiment, by providing a product quality accident grading method and system based on a convolutional neural network, quality accident pictures of quality accident products are compared with initial pictures, economic losses caused by the quality accidents of the products are analyzed, the accidents are scientifically and reasonably graded, meanwhile, the quality accidents are graded by utilizing the severity degree of the quality accidents and image characteristic elements, and the quality accidents are graded by a quality accident grading model, so that subjective errors caused by manual judgment are reduced, and objectivity and consistency of quality accident assessment results are improved.
The foregoing is merely illustrative of the structures of this invention and various modifications, additions and substitutions for those skilled in the art can be made to the described embodiments without departing from the scope of the invention or from the scope of the invention as defined in the accompanying claims.
Claims (6)
1. The product quality accident grading method based on the convolutional neural network is characterized by comprising the following steps of:
Collecting sample data of the quality accidents of the products, marking the products with the quality accidents as the quality accident products, and calculating to obtain the accident severity assessment index of the quality accident products;
Extracting each image characteristic element of each quality accident product, calculating to obtain an accident correlation degree evaluation value of each image characteristic element according to an accident severity evaluation index of each quality accident product, and screening effective characteristic elements according to the accident correlation degree evaluation value of each image characteristic element to obtain each effective characteristic element;
training according to each effective characteristic element to obtain a product quality accident grading model, obtaining a target product and grading the product quality accidents through the product quality accident grading model;
the accident severity evaluation index of each quality accident product is obtained through calculation, and the specific analysis process is as follows:
According to the product quality accident sample data, extracting an initial image and a quality accident image of each quality accident product, comparing the initial image and the quality accident image of each quality accident product, and calculating to obtain a damage degree quantification index of each quality accident product;
Obtaining the repair cost of each quality accident product, counting the repair time cost of each quality accident product, and obtaining the influence degree quantification index of each quality accident product through processing calculation;
comprehensively calculating to obtain accident severity assessment indexes of all quality accident products;
The accident correlation degree evaluation value of each image characteristic element is obtained through calculation, and the specific analysis process is as follows:
According to the quality accident images of the quality accident products, obtaining image entropy values and image contrast indexes of the quality accident products through processing, and comprehensively calculating to obtain characteristic quantized values of image characteristic elements of the quality accident products;
according to the accident severity evaluation index of each quality accident product and the characteristic quantization value of each image characteristic element of each quality accident product, calculating to obtain an accident correlation degree evaluation value of each image characteristic element;
The accident severity assessment index of each quality accident product comprises the following specific calculation expression:
，
In the method, in the process of the invention, Represents the/>Accident severity assessment index for individual quality accident products,/>Represents a natural constant of the natural product,Represents the/>Quantitative index of damage degree of individual quality accident products,/>Represents the/>Quantitative index of influence degree of individual quality accident products,/>Accident severity influencing factors corresponding to the set damage degree quantification indexes,/>, are representedThe accident severity influence factors corresponding to the set influence degree quantization indexes are represented;
The characteristic quantization value of each image characteristic element of each quality accident product is specifically calculated as follows:
，
In the method, in the process of the invention, Represents the/>Product of personal quality accident/>Feature quantization value of each image feature element,/>Representing natural constant,/>Represents the/>Image entropy value of individual quality accident products,/>Represents the/>Image contrast index of individual quality accident products,/>Indicate the set/>Characteristic quantization influence factors of image entropy values corresponding to image characteristic elements,/>Indicate the set/>Characteristic quantization influence factors of image contrast indexes corresponding to the individual image characteristic elements;
Calculating damage degree quantization indexes of each quality accident product, wherein the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Quantitative index of damage degree of individual quality accident products,/>Represents the/>Quality incident image of individual quality incident product/>The pixel values of the individual pixel points,Represents the/>Initial image of individual quality accident product/>Pixel value of each pixel point,/>Representing the set allowable deviation pixel value,/>Indicating the set product damage degree quantized correction factor,/>The number of each quality accident product is indicated,，/>Representing the total number of quality accident products,/>The number of each pixel point is indicated,，/>Representing the total number of pixel points;
calculating to obtain the influence degree quantization index of each quality accident product, wherein the specific calculation expression is as follows: in the above, the ratio of/> Represents the/>Quantitative index of influence degree of individual quality accident products,/>Representing natural constant,/>Represents the/>Repair cost of individual quality Accident products,/>Represents the/>Repair time cost of individual quality Accident products,/>Representing a predefined critical repair cost,/>Representing a predefined critical repair time cost,/>Quantized weighting factor indicating influence level corresponding to set repair cost,/>A quantitative weighting factor indicating the degree of influence corresponding to the set repair time cost,Quantized weighting factor indicating influence degree corresponding to set repair cost unit value,/>The influence degree quantization weight factor corresponding to the set repair time cost unit value is represented;
The accident correlation degree evaluation value of each image characteristic element is specifically calculated as follows: in the above, the ratio of/> Represents the/>Accident correlation degree evaluation value of each image characteristic element,/>Represents the/>Accident severity assessment index for individual quality accident products,/>Represents the/>Product of personal quality accident/>Feature quantization value of each image feature element,/>An average accident severity assessment index representing quality accident products,/>，/>Represents the/>Average feature quantization values of individual image feature elements,，/>Indicating the set accident related degree correction factor.
2. The convolutional neural network-based product quality incident classification method of claim 1, wherein: the quantitative evaluation data are obtained by comparing and analyzing the initial image and the quality accident image of each quality accident product, and are used for quantitatively evaluating the damage degree of each quality accident product and providing data basis for quality accident grading.
3. The convolutional neural network-based product quality incident classification method of claim 1, wherein: the effective characteristic elements are screened according to the accident correlation degree evaluation values of the image characteristic elements to obtain the effective characteristic elements, and the specific analysis process is as follows:
And comparing the accident correlation degree evaluation value of each image characteristic element with a correlation degree evaluation threshold value stored in a quality control database, and marking the image characteristic element as an effective characteristic element if the accident correlation degree evaluation value of a certain image characteristic element is larger than or equal to the correlation degree evaluation threshold value.
4. The convolutional neural network-based product quality incident classification method of claim 1, wherein: the damage degree quantization index and the influence degree quantization index of each quality accident product are analyzed to obtain a quantization evaluation value, and the quantization evaluation value is used for quantitatively evaluating the accident severity of each quality accident product and providing a data basis for quality accident grading.
5. The convolutional neural network-based product quality incident classification method of claim 1, wherein: the quantized data of each image characteristic element, which is obtained by performing quantization analysis on each image characteristic element from two dimensions of an image entropy value and an image contrast index, is used for quantitatively evaluating the severity of each image characteristic element, and provides a data basis for screening effective characteristic elements.
6. A system applying the convolutional neural network-based product quality incident classification method of any one of claims 1-5, characterized in that: comprising the following steps:
The quality accident grading module is used for collecting product quality accident sample data, marking products with quality accidents as quality accident products, and analyzing to obtain accident severity assessment indexes of the quality accident products;
The image feature element extraction module is used for extracting each image feature element of each quality accident product, analyzing and obtaining an accident correlation degree evaluation value of each image feature element according to the accident severity evaluation index of each quality accident product, and screening effective feature elements according to the accident correlation degree evaluation value of each image feature element to obtain each effective feature element;
The grading model training module is used for training to obtain a product quality accident grading model according to each effective characteristic element, obtaining a target product and grading the product quality accident through the product quality accident grading model;
The quality control database is used for storing product quality accident sample data, accident grades corresponding to the accident severity evaluation index intervals and related degree evaluation thresholds.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410349918.0A CN117952482B (en) | 2024-03-26 | 2024-03-26 | Product quality accident grading method and system based on convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410349918.0A CN117952482B (en) | 2024-03-26 | 2024-03-26 | Product quality accident grading method and system based on convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117952482A CN117952482A (en) | 2024-04-30 |
| CN117952482B true CN117952482B (en) | 2024-06-14 |
Family
ID=90796539
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410349918.0A Active CN117952482B (en) | 2024-03-26 | 2024-03-26 | Product quality accident grading method and system based on convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117952482B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106327108A (en) * | 2016-10-11 | 2017-01-11 | 北京航空航天大学 | Product quality accident classification method based on failure mechanism and domain mapping |
| CN117726240A (en) * | 2024-02-18 | 2024-03-19 | 中国标准化研究院 | Quality evaluation classification method and system based on convolutional neural network |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6470303B2 (en) * | 1998-02-04 | 2002-10-22 | Injury Sciences Llc | System and method for acquiring and quantifying vehicular damage information |
| US8983168B2 (en) * | 2012-04-30 | 2015-03-17 | Ncr Corporation | System and method of categorising defects in a media item |
-
2024
- 2024-03-26 CN CN202410349918.0A patent/CN117952482B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106327108A (en) * | 2016-10-11 | 2017-01-11 | 北京航空航天大学 | Product quality accident classification method based on failure mechanism and domain mapping |
| CN117726240A (en) * | 2024-02-18 | 2024-03-19 | 中国标准化研究院 | Quality evaluation classification method and system based on convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117952482A (en) | 2024-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108549908B (en) | Chemical process fault detection method based on multi-sampling probability kernel principal component model | |
| CN111898839B (en) | Importance degree classification method and device for power users | |
| CN111415492A (en) | Slope landslide early warning method and system based on fuzzy comprehensive evaluation algorithm | |
| CN112132233A (en) | Criminal personnel dangerous behavior prediction method and system based on effective influence factors | |
| CN111402236A (en) | Hot-rolled strip steel surface defect grading method based on image gray value | |
| CN102254185B (en) | Background clutter quantizing method based on contrast ratio function | |
| CN111340098B (en) | STA-Net age prediction method based on shoe print image | |
| CN117193222A (en) | Intelligent quality control system based on industrial Internet of things and big data and control method thereof | |
| CN114679310A (en) | Network information security detection method | |
| CN115713403A (en) | Enterprise risk identification method, device and equipment based on self-coding neural network | |
| Sang et al. | Blind image quality assessment using a reciprocal singular value curve | |
| CN117172430B (en) | Deep learning-based water body environment assessment and prediction method and system | |
| CN110751170A (en) | Panel quality detection method, system, terminal device and computer readable medium | |
| CN117952482B (en) | Product quality accident grading method and system based on convolutional neural network | |
| Gaata et al. | No-reference quality metric for watermarked images based on combining of objective metrics using neural network | |
| CN117132383A (en) | Credit data processing method, device, equipment and readable storage medium | |
| CN111238927A (en) | Fatigue durability evaluation method and device, electronic equipment and computer readable medium | |
| Zhang et al. | Nature scene statistics approach based on ICA for no-reference image quality assessment | |
| CN111985826B (en) | Visual quality grading method and system for multi-index industrial products | |
| CN114625781A (en) | Commodity housing value-based batch evaluation method | |
| CN116012634A (en) | Strip steel shape classification method, device, computer medium and equipment | |
| CN112767365A (en) | Flaw detection method | |
| CN107704680B (en) | Method for calculating safety injury occurrence probability of consumer goods | |
| Saulo et al. | A bivariate fatigue-life regression model and its application to fracture of metallic tools | |
| Olsen et al. | Predicting dactyloscopic examiner fingerprint image quality assessments |
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 |