CN118392732A - Method and system for detecting quality of high-frequency low-loss ceramic powder - Google Patents
Method and system for detecting quality of high-frequency low-loss ceramic powder Download PDFInfo
- Publication number
- CN118392732A CN118392732A CN202410824929.XA CN202410824929A CN118392732A CN 118392732 A CN118392732 A CN 118392732A CN 202410824929 A CN202410824929 A CN 202410824929A CN 118392732 A CN118392732 A CN 118392732A
- Authority
- CN
- China
- Prior art keywords
- ceramic powder
- particle
- image
- powder
- pixel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000000843 powder Substances 0.000 title claims abstract description 312
- 239000000919 ceramic Substances 0.000 title claims abstract description 193
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000001514 detection method Methods 0.000 claims abstract description 33
- 239000002245 particle Substances 0.000 claims description 180
- 230000003068 static effect Effects 0.000 claims description 71
- 238000005070 sampling Methods 0.000 claims description 40
- 238000009826 distribution Methods 0.000 claims description 39
- 238000010586 diagram Methods 0.000 claims description 27
- 238000013441 quality evaluation Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 25
- 238000012549 training Methods 0.000 claims description 21
- 238000007781 pre-processing Methods 0.000 claims description 12
- 230000009466 transformation Effects 0.000 claims description 12
- 238000004458 analytical method Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 9
- 230000000877 morphologic effect Effects 0.000 claims description 9
- 239000013598 vector Substances 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000001303 quality assessment method Methods 0.000 claims description 6
- 238000003707 image sharpening Methods 0.000 claims description 4
- 230000000903 blocking effect Effects 0.000 claims description 3
- 239000006185 dispersion Substances 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000002360 preparation method Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012512 characterization method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000001035 drying Methods 0.000 description 2
- 238000002156 mixing Methods 0.000 description 2
- 238000005245 sintering Methods 0.000 description 2
- 238000005054 agglomeration Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000000746 purification Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000000790 scattering method Methods 0.000 description 1
- 238000003756 stirring Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
- G01N15/02—Investigating particle size or size distribution
- G01N15/0205—Investigating particle size or size distribution by optical means
- G01N15/0227—Investigating particle size or size distribution by optical means using imaging; using holography
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N11/00—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/85—Investigating moving fluids or granular solids
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- 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
- G06V10/54—Extraction of image or video features relating to texture
-
- 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
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- 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
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N11/00—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties
- G01N2011/006—Determining flow properties indirectly by measuring other parameters of the system
- G01N2011/008—Determining flow properties indirectly by measuring other parameters of the system optical properties
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/85—Investigating moving fluids or granular solids
- G01N2021/8592—Grain or other flowing solid samples
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Software Systems (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Dispersion Chemistry (AREA)
- Quality & Reliability (AREA)
- Geometry (AREA)
- Probability & Statistics with Applications (AREA)
- Investigating Or Analyzing Materials By The Use Of Electric Means (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of ceramic powder detection, in particular to a method and a system for detecting the quality of high-frequency low-loss ceramic powder.
Description
Technical Field
The invention relates to the technical field of ceramic powder detection, in particular to a method and a system for detecting the quality of high-frequency low-loss ceramic powder.
Background
The fineness and particle distribution of the ceramic raw material and the glaze material influence the technological properties and physicochemical properties of the product, such as plasticity, drying shrinkage, green body drying strength, sintering shrinkage, sintering property and the like. At present, the quality detection of ceramic powder is mainly performed by a laser scattering method, the measurement mode of equipment is complex, the real-time requirement after industrialization is difficult to meet, and the loss is high.
For example, chinese patent application publication No. CN117191839a discloses a fine powder SEM particle sample preparation apparatus and a particle size distribution detection method. The sample preparation equipment comprises a base, a sample stage arranged on the base and a high-speed airflow disperser arranged above the sample stage. The powder sample is scattered by the sample preparation equipment and falls into the sample stage, so that the problem that the existing sample is directly placed on the sample stage to be detected and agglomeration exists, so that automatic detection cannot be performed, and only manual detection can be relied on is effectively solved.
The problems proposed in the background art exist in the above patents: at present, complicated equipment and technology are required for detecting the quality of ceramic powder, the economic cost and the loss rate are high, the high requirements on pretreatment and preparation of samples are possibly only applicable to ceramic powder of a specific type or a specific form, and the method and the system for detecting the quality of the high-frequency low-loss ceramic powder are not applicable to other types.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a method and a system for detecting the quality of high-frequency low-loss ceramic powder, wherein a sampling device is used for collecting a ceramic powder sample, an image sensor is used for collecting a ceramic powder static image and a ceramic powder dynamic image, the ceramic powder static image and the ceramic powder dynamic image are processed, the particle distribution characteristics and the particle uniformity characteristics are extracted, and finally an input parameter is trained through a ceramic powder quality evaluation network, so that the quality detection result of the ceramic powder is output.
In order to achieve the above purpose, the present invention provides the following technical solutions:
the method for detecting the quality of the high-frequency low-loss ceramic powder comprises the following steps of:
s1: collecting a ceramic powder sample through a sampling device, and collecting a ceramic powder static image and a ceramic powder dynamic image through an image sensor;
s2: processing the ceramic powder static image and extracting particle distribution characteristics;
s3: processing the ceramic powder dynamic image, and extracting uniform particle characteristics;
s4: and constructing a ceramic powder quality evaluation network, taking the particle distribution characteristics and the particle uniformity characteristics as input parameters of the ceramic powder quality evaluation network, training the input parameters through the ceramic powder quality evaluation network, and outputting a quality detection result of the ceramic powder.
The sampling device comprises a powder mixer, a powder sampling port, a powder sampling disc and a conveying device, and the image sensor comprises a digital microscope and a high-speed camera.
The specific steps for processing the ceramic powder static image are as follows:
s2.1: preprocessing a ceramic powder static image, wherein the preprocessing comprises graying, image sharpening and image denoising;
S2.2: according to the preprocessed ceramic powder static image, calculating the energy gradient of the ceramic powder static image, determining the block number of the ceramic powder static image according to the energy gradient, blocking the ceramic powder static image, obtaining a histogram of each sub-block of the ceramic powder static image, generating an equalization histogram of each sub-block through histogram equalization, and obtaining the ceramic powder static equalization image, wherein the block number of the ceramic powder static image has the following calculation formula:
，
Wherein Q represents the number of blocks of the ceramic powder still image, m { } represents a downward rounding function, a represents a single pixel of the ceramic powder still image, I represents the total number of pixels of the ceramic powder still image, Represents the energy gradient of the a-th pixel point, F represents the total energy gradient, L represents the length of the ceramic powder static image, W represents the width of the ceramic powder static image,Indicating the degree of dispersion of the a-th pixel,Represents the gray-scale distribution kurtosis value of the a-th pixel point,A gray level distribution uniformity value representing the a-th pixel point;
S2.3: performing discrete wavelet transformation on the ceramic powder static equilibrium image to obtain a low-frequency component and a high-frequency component of the ceramic powder static equilibrium image, and enhancing the low-frequency component and the high-frequency component through the segmented normalized detail coefficient to obtain a powder particle enhanced image;
S2.4: calculating the density index of each pixel point, clustering the pixel points with similar density indexes through a clustering algorithm, and acquiring a particle area diagram according to a clustering result;
S2.5: the method comprises the steps of performing weighted sampling on a particle area diagram by an up-sampling function, reducing the number of 4C channels of the particle area diagram to 2C according to regularization and dimension transformation, measuring the particle area diagram, and extracting particle characteristics of the measured particle area diagram through residual connection, wherein the particle characteristics comprise texture characteristics and morphological characteristics;
S2.6: the texture features and the morphological features are fused by convolution kernels of 1×1 size, and the particle distribution features are calculated.
The extraction steps of the texture features are as follows:
s2.5.1: extracting key points of a particle area diagram, taking the key points as the center, placing a first virtual camera according to an included angle of 30 degrees between the camera and the center plane of the model, taking a Z axis as a rotating shaft, placing a next virtual camera every 30 degrees, and obtaining twelve views of a three-dimensional model object, wherein the virtual cameras meet the condition that the shooting direction and the center of the model are on the same straight line;
s2.5.2: constructing a view feature classification network, taking the twelve views as input parameters of the view feature classification network, training the view feature classification network, and outputting classification probability of each view Where P represents the view feature classification probability, Z represents the total number of view feature classes,Representing the classification probability of the view corresponding to the b-th view feature class;
s2.5.3: voting the view feature categories according to the view feature classification probability by each view, and taking the view feature category with the largest statistics vote number as the texture feature of the particle area diagram;
The morphological features include particle area, particle perimeter, particle roundness, particle convex hull area, particle aspect ratio, particle convex perimeter, particle solidity, and particle convexity;
The calculation formula of the particle distribution characteristics is as follows:
，
Wherein, Representing grain distribution characteristics, f representing convolution operations of the convolution kernel, R representing texture characteristics,The characteristic of the morphology is represented by,Representing the largest pixel value in the particle area map,Representing the smallest pixel value in the particle area map,The vector point-of-time is represented,Low frequency feature vectors representing the particle region map,The high-frequency gradient vector representing the particle area map, i represents the particle area map pixel abscissa, j represents the particle area map pixel ordinate, N represents the particle area map horizontal total pixel number, M represents the particle area map vertical total pixel number, and D (i, j) represents the image edge value of the area image at the (i, j) position.
The specific steps for processing the ceramic powder dynamic image are as follows:
s3.1: preprocessing the ceramic powder dynamic image, wherein the preprocessing comprises linear enhancement and denoising;
S3.2: traversing each pixel point in the pretreated ceramic powder dynamic image, calculating the sum of gray weighted differences between the neighborhood of the pixel point and the pixel point, comparing the sum of the gray weighted differences with a binarization threshold value, and setting the pixel point to be zero if the sum of the gray weighted differences is smaller than the binarization threshold value to obtain a powder binarization image;
S3.3: carrying out peak statistics on the powder binary image through Hough transformation, determining the horizontal coordinate of a parallel line according to the peak value, acquiring a powder flow area according to the upper and lower boundary coordinates of the powder binary image and the horizontal coordinate of the parallel line, projecting all white pixels in the powder flow area, and extracting a flowing powder image;
S3.4: marking pixel edge point coordinates of each flowing powder image, extracting area characteristics of a particle area, and calculating particle flow characteristics according to variances of the area characteristics of the particle area;
S3.5: dividing subareas according to the horizontal axis direction and the vertical axis direction of each flowing powder image in an equidistant mode, extracting area features of the subareas, and calculating particle stability features according to standard deviations of the area features of the subareas;
S3.6: the particle flow characteristics and the particle stability characteristics are fused through convolution kernels with the size of 1 multiplied by 1, the particle uniformity characteristics are calculated, and the calculation formula of the particle uniformity characteristics is as follows:
，
Wherein, Indicating the uniformity of the particles and,Indicating the flow characteristics of the particles, H indicating the stability characteristics of the particles,An average projection matrix of white pixels of a powder flow region is represented, T represents a matrix transpose, Q represents an abscissa of a flowing powder image pixel, K represents an ordinate of a flowing powder image pixel, Q represents a total number of horizontal pixels of the flowing powder image, K represents a total number of vertical pixels of the flowing powder image, I (Q, K) represents a gradation value of the flowing powder image at a (Q, K) position,The average gray value of the flowing powder image is represented.
The construction of the ceramic powder quality assessment network comprises network parameter pre-training, the network parameter pre-training optimizes network parameters according to a powder quality disclosure data set, the network parameters comprise a learning rate, a node threshold, an initial weight and a cross entropy, and the ceramic powder quality assessment network comprises:
the input layer establishes a characteristic sequence according to input parameters and calculates a fitting value of the characteristic sequence;
An implicit layer, converting the fitting value of the characteristic sequence of the input layer into a high-dimensional space, and calculating the output value of the implicit layer in the high-dimensional space;
And the output layer is applied to outputting the quality detection result of the ceramic powder.
Taking the particle distribution characteristics and the particle uniformity characteristics as input parameters of the ceramic powder quality evaluation network, training the input parameters through the ceramic powder quality evaluation network, and outputting a quality detection result of the ceramic powder, wherein the method comprises the following steps:
initializing the ceramic powder quality evaluation network according to the network parameters, and determining the connection weight and the node threshold of an input layer and an hidden layer;
Biasing input parameters through the connection weights of the input layer and the hidden layer to obtain output of the hidden layer, and updating the output of the hidden layer through the node threshold to obtain input of the output layer;
Calculating an error input by an output layer according to an error function of the output layer, judging whether the error input by the output layer meets a convergence requirement of the output layer, and updating the connection weight and the node threshold value if the error input by the output layer does not meet the convergence requirement;
And outputting a quality detection result of the ceramic powder if the convergence requirement is met, wherein the quality detection result comprises a high-grade, a medium-grade and a low-grade.
The system comprises a powder sample acquisition module, a powder sample analysis module and a powder sample evaluation module;
the powder sample acquisition module is used for acquiring a ceramic powder sample through the sampling device, and acquiring a ceramic powder static image and a ceramic powder dynamic image of the ceramic powder sample through the image sensor;
The powder sample analysis module is used for processing the ceramic powder static image and the ceramic powder dynamic image to obtain particle distribution characteristics and particle uniformity characteristics;
the powder sample evaluation module is used for training the particle distribution characteristics and the particle uniformity characteristics through a ceramic powder quality evaluation network and outputting a quality detection result of the ceramic powder;
the powder sample collection module includes:
the powder sample sampling unit is used for collecting ceramic powder samples through the sampling device;
the powder static image acquisition unit is used for acquiring a ceramic powder static image according to the digital microscope;
the powder dynamic image acquisition unit is used for acquiring ceramic powder dynamic images according to the high-speed camera;
The powder sample analysis module includes:
the static image processing unit is used for processing the ceramic powder static image and extracting particle distribution characteristics;
the dynamic image processing unit is used for processing the ceramic powder dynamic image and extracting the uniform particle characteristics;
the powder sample evaluation module includes:
The network pre-training unit is used for optimizing network parameters according to the powder quality disclosure data set;
The powder quality evaluation unit is used for training the input parameters through the ceramic powder quality evaluation network and outputting the quality detection result of the ceramic powder.
Compared with the prior art, the invention has the beneficial effects that:
1. The invention combines the image processing technology to detect the quality of the ceramic powder, avoids the requirements of professional equipment in other detection works, reduces the loss of the powder through nondestructive detection, and improves the detection accuracy;
2. According to the invention, the powder image is divided into a static image and a dynamic image, the particle size and the distribution degree of the powder are comprehensively evaluated from the aspects of fluidity and uniformity, and the comprehensiveness of powder detection is improved;
3. According to the invention, the particle image is enhanced by an image enhancement technology in the powder image processing process, so that the problem of particle information loss caused by insufficient resolution of a camera image is avoided, and in addition, similar pixel points with similar density indexes are aggregated by a clustering algorithm, so that the accuracy of particle extraction is improved.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, made with reference to the following drawings in which:
FIG. 1 is a flow chart of the method for detecting the quality of high-frequency low-loss ceramic powder according to the embodiment 1 of the invention;
FIG. 2 is a schematic diagram of a sampling flow of a powder sampling apparatus according to embodiment 1 of the present invention;
FIG. 3 is an exploded view of the wavelet transform according to embodiment 1 of the present invention;
FIG. 4 is a diagram of a view feature classification network according to embodiment 1 of the present invention;
FIG. 5 is a block diagram of a system for detecting the quality of a high frequency low loss ceramic powder according to embodiment 2 of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments.
Example 1
Referring to fig. 1, an embodiment of the present invention is provided: the method for detecting the quality of the high-frequency low-loss ceramic powder comprises the following steps of:
s1: collecting a ceramic powder sample through a sampling device, and collecting a ceramic powder static image and a ceramic powder dynamic image through an image sensor;
s2: processing the ceramic powder static image and extracting particle distribution characteristics;
s3: processing the ceramic powder dynamic image, and extracting uniform particle characteristics;
s4: and constructing a ceramic powder quality evaluation network, taking the particle distribution characteristics and the particle uniformity characteristics as input parameters of the ceramic powder quality evaluation network, training the input parameters through the ceramic powder quality evaluation network, and outputting a quality detection result of the ceramic powder.
The sampling device comprises a powder mixer, a powder sampling port, a powder sampling disc and a conveying device, wherein the image sensor comprises a digital microscope and a high-speed camera, the image sensor is used for collecting ceramic powder static images and the ceramic powder dynamic images comprise a digital microscope for collecting ceramic powder static images and a high-speed camera for collecting ceramic powder dynamic images, the digital microscope is used for collecting ceramic powder static images, the powder sampling disc is arranged on the conveying device, the powder sampling disc is transmitted to the lower part of the digital microscope after the sampling is finished, the rotary disc is used for shooting ceramic powder static images corresponding to different sampling points on the disc ring by the digital microscope, the high-speed camera is used for collecting ceramic powder dynamic images, the ceramic powder dynamic images are poured into a powder hopper, and the high-speed camera is used for collecting ceramic powder dynamic images in the falling process of the ceramic powder hopper;
Referring to fig. 2, a schematic diagram of a sampling flow of a powder sampling device according to an embodiment of the present invention is shown, a powder sample is placed in a powder mixer, the powder mixer performs powder mixing, the powder presents self-dropping along with movement of a stirring paddle, a disc is placed at a discharge port at the lower part of the mixer, along with continuous self-dropping movement of powder in the mixer during mixing, powder samples at different sampling point positions in the mixer are collected, when the powder on the disc exceeds the height of the disc, sampling is stopped, excessive powder is scraped by a doctor blade, the sampling disc is placed on a conveying device, after sampling is finished, the sampling disc is conveyed to the lower part of a digital microscope by the conveying device, an image is collected by the digital microscope, and after the static image is collected, the sampling disc is conveyed to a powder hopper by the conveying device, the powder hopper is placed in the powder hopper by adopting a hyperbolic hopper, and the powder dynamic image is photographed by a high-speed camera;
The specific steps for processing the ceramic powder static image are as follows:
S2.1: preprocessing a ceramic powder static image, wherein the preprocessing comprises graying, image sharpening and image denoising, the graying is used for carrying out image graying on the ceramic powder static image by a weighted average method, the image sharpening comprises histogram equalization and linear gray scale transformation, and the image denoising comprises sensor noise denoising and field light noise denoising;
S2.2: according to the preprocessed ceramic powder static image, calculating the energy gradient of the ceramic powder static image, determining the block number of the ceramic powder static image according to the energy gradient, blocking the ceramic powder static image, obtaining a histogram of each sub-block of the ceramic powder static image, generating an equalization histogram of each sub-block through histogram equalization, and obtaining the ceramic powder static equalization image, wherein the block number of the ceramic powder static image has the following calculation formula:
，
Wherein Q represents the number of blocks of the ceramic powder still image, m { } represents a downward rounding function, a represents a single pixel of the ceramic powder still image, I represents the total number of pixels of the ceramic powder still image, Represents the energy gradient of the a-th pixel point, F represents the total energy gradient, L represents the length of the ceramic powder static image, W represents the width of the ceramic powder static image,Indicating the degree of dispersion of the a-th pixel,Represents the gray-scale distribution kurtosis value of the a-th pixel point,A gray level distribution uniformity value representing the a-th pixel point;
S2.3: performing discrete wavelet transformation on the ceramic powder static equilibrium image to obtain a low-frequency component and a high-frequency component of the ceramic powder static equilibrium image, and enhancing the low-frequency component and the high-frequency component through the segmented normalized detail coefficient to obtain a powder particle enhanced image;
referring to fig. 3, a schematic diagram of decomposition of wavelet transformation according to an embodiment of the present invention is shown, where a low-pass filter and a high-pass filter process a static equilibrium image of ceramic powder in two directions of rows and columns, respectively, and decompose the static equilibrium image into a high-frequency part and a low-frequency part, where the low-frequency part is Low frequency information, high frequency part isHigh frequency information in the horizontal direction,Vertical high frequency informationThe high frequency information of the diagonal line is provided,Is used for reflecting the overall view of the static equilibrium image of the ceramic powder,、AndThe method is used for reflecting the detail part of the static equilibrium image of the ceramic powder, carrying out wavelet transformation on the low-frequency information again, and obtaining the low-frequency information of the next layerHigh frequency part information、And；
S2.4: calculating the density index of each pixel point, clustering the pixel points with similar density indexes through a clustering algorithm, and acquiring a particle area diagram according to a clustering result;
Specifically, through combining two clustering algorithms of DBSCAN and K-means, clustering analysis is carried out on the pixel points, and a more accurate clustering result is obtained through iterative optimization of a central cluster, so that the corresponding relation between the pixel points and particles is better known, and the purification efficiency of powder particles is improved. DBSCAN is a dense clustering algorithm that identifies closely connected data points in space and divides them into clusters, and can be used for preliminary clustering to obtain core points, which refers to a core region containing at least a specified number of data points within a given radius, and since each core point may represent a cluster, the number of core points can be used to preliminarily determine the number of clusters in the center of the cluster.
Taking a nuclear point obtained by performing primary clustering on DBSCAN as an initial center cluster of a K-means clustering algorithm, and providing an initial cluster center position for the K-means algorithm; the Euclidean distance is used for cluster division, and each pixel point is distributed to the nearest central cluster by calculating the Euclidean distance from the density index of each pixel point to each initial central cluster. The cluster average error is an average value of Euclidean distances from all data points to the center, the cluster average error can be used for updating an initial center cluster to improve the clustering effect, the position of the center cluster can be adjusted according to the calculated cluster average error, so that the center cluster is updated, a convergence condition is defined, whether the updated center cluster meets the condition is judged to determine whether iteration is ended, and if the convergence condition meets the condition, a clustering result is output;
S2.5: the method comprises the steps of performing weighted sampling on a particle area diagram by an up-sampling function, reducing the number of 4C channels of the particle area diagram to 2C according to regularization and dimension transformation, measuring the particle area diagram, and extracting particle characteristics of the measured particle area diagram through residual connection, wherein the particle characteristics comprise texture characteristics and morphological characteristics;
S2.6: the texture features and the morphological features are fused by convolution kernels of 1×1 size, and the particle distribution features are calculated.
The extraction steps of the texture features are as follows:
s2.5.1: extracting key points of a particle area diagram, taking the key points as the center, placing a first virtual camera according to an included angle of 30 degrees between the camera and the center plane of the model, taking a Z axis as a rotating shaft, placing a next virtual camera every 30 degrees, and obtaining twelve views of a three-dimensional model object, wherein the virtual cameras meet the condition that the shooting direction and the center of the model are on the same straight line;
s2.5.2: constructing a view feature classification network, taking the twelve views as input parameters of the view feature classification network, training the view feature classification network, and outputting classification probability of each view Where P represents the view feature classification probability, Z represents the total number of view feature classes,Representing the classification probability of the view corresponding to the b-th view feature class;
s2.5.3: voting the view feature categories according to the view feature classification probability by each view, and taking the view feature category with the largest statistics vote number as the texture feature of the particle area diagram;
Referring to fig. 4, a view feature classification network structure diagram according to an embodiment of the present invention includes a view characterization layer, a view classification layer, and a view feature voting layer;
the view characterization layer is used for extracting view shallow layer characteristics of the input parameters through the VGG-11 network according to the input parameters of the view characteristic classification network;
The view classification layer takes view shallow features as input and outputs classification probability of each view shallow feature through an N-gram feature learning unit;
And the view feature voting layer is used for voting the view feature category according to the view feature classification probability, the view feature category with the largest statistical vote number is used as the surface texture feature of the three-dimensional model object, if the condition that the vote number is the same occurs, the voting probability value is increased to be used as the additional weight to recalculate the vote number, and the concrete voting method is as follows:
The method comprises the steps of finding out the maximum classification probability in a view, voting a category to which the maximum classification probability belongs, re-ordering the rest classification probabilities of the view, taking the maximum classification probability in the rest classification probabilities, comparing the maximum classification probability with the maximum classification probability, and voting a category corresponding to the maximum classification probability in the rest classification probabilities if the difference value of the maximum classification probability and the maximum classification probability is smaller than a comparison threshold, wherein the comparison threshold is determined by a person skilled in the art according to a large number of experiments;
The morphological characteristics comprise particle area, particle perimeter, particle roundness, particle convex hull area, particle length-width ratio, particle convex perimeter, particle firmness and particle convexity, wherein the particle area refers to the number of pixels occupied by particles, the particle perimeter refers to the number of pixels at the edge of particles, the particle roundness refers to the ratio of the particle area to the minimum circumcircle area of particles, the particle convex hull area refers to the number of pixels occupied by the inside of the particle convex hull, the particle length-width ratio refers to the ratio of the minimum circumcircle rectangular length to the width, the particle convex perimeter refers to the number of pixels at the edge of the particle convex hull, the particle firmness refers to the ratio of the particle convex hull area to the particle area, and the particle convexity refers to the ratio of the particle convex hull perimeter to the particle perimeter.
The calculation formula of the particle distribution characteristics is as follows:
，
Wherein, Representing grain distribution characteristics, f representing convolution operations of the convolution kernel, R representing texture characteristics,The characteristic of the morphology is represented by,Representing the largest pixel value in the particle area map,Representing the smallest pixel value in the particle area map,The vector point-of-time is represented,Low frequency feature vectors representing the particle region map,The high-frequency gradient vector representing the particle area map, i represents the particle area map pixel abscissa, j represents the particle area map pixel ordinate, N represents the particle area map horizontal total pixel number, M represents the particle area map vertical total pixel number, and D (i, j) represents the image edge value of the area image at the (i, j) position.
The specific steps for processing the ceramic powder dynamic image are as follows:
s3.1: preprocessing the ceramic powder dynamic image, wherein the preprocessing comprises linear enhancement and denoising;
S3.2: traversing each pixel point in the pretreated ceramic powder dynamic image, calculating the sum of gray weighted differences between the neighborhood of the pixel point and the pixel point, comparing the sum of the gray weighted differences with a binarization threshold value, and setting the pixel point to be zero if the sum of the gray weighted differences is smaller than the binarization threshold value to obtain a powder binarization image;
S3.3: carrying out peak statistics on the powder binary image through Hough transformation, determining the horizontal coordinate of a parallel line according to the peak value, acquiring a powder flow area according to the upper and lower boundary coordinates of the powder binary image and the horizontal coordinate of the parallel line, projecting all white pixels in the powder flow area, and extracting a flowing powder image;
S3.4: marking pixel edge point coordinates of each flowing powder image, extracting area characteristics of a particle area, and calculating particle flow characteristics according to variances of the area characteristics of the particle area;
S3.5: dividing subareas according to the horizontal axis direction and the vertical axis direction of each flowing powder image in an equidistant mode, extracting area features of the subareas, and calculating particle stability features according to standard deviations of the area features of the subareas;
S3.6: the particle flow characteristics and the particle stability characteristics are fused through convolution kernels with the size of 1 multiplied by 1, the particle uniformity characteristics are calculated, and the calculation formula of the particle uniformity characteristics is as follows:
，
Wherein, Indicating the uniformity of the particles and,Indicating the flow characteristics of the particles, H indicating the stability characteristics of the particles,An average projection matrix of white pixels of a powder flow region is represented, T represents a matrix transpose, Q represents an abscissa of a flowing powder image pixel, K represents an ordinate of a flowing powder image pixel, Q represents a total number of horizontal pixels of the flowing powder image, K represents a total number of vertical pixels of the flowing powder image, I (Q, K) represents a gradation value of the flowing powder image at a (Q, K) position,The average gray value of the flowing powder image is represented.
The construction of the ceramic powder quality assessment network comprises network parameter pre-training, the network parameter pre-training optimizes network parameters according to a powder quality disclosure data set, the network parameters comprise a learning rate, a node threshold, an initial weight and a cross entropy, and the ceramic powder quality assessment network comprises:
the input layer establishes a characteristic sequence according to input parameters and calculates a fitting value of the characteristic sequence;
An implicit layer, converting the fitting value of the characteristic sequence of the input layer into a high-dimensional space, and calculating the output value of the implicit layer in the high-dimensional space;
And the output layer is applied to outputting the quality detection result of the ceramic powder.
Taking the particle distribution characteristics and the particle uniformity characteristics as input parameters of the ceramic powder quality evaluation network, training the input parameters through the ceramic powder quality evaluation network, and outputting a quality detection result of the ceramic powder, wherein the method comprises the following steps:
initializing the ceramic powder quality evaluation network according to the network parameters, and determining the connection weight and the node threshold of an input layer and an hidden layer;
Biasing input parameters through the connection weights of the input layer and the hidden layer to obtain output of the hidden layer, and updating the output of the hidden layer through the node threshold to obtain input of the output layer;
Calculating an error input by an output layer according to an error function of the output layer, judging whether the error input by the output layer meets a convergence requirement of the output layer, and updating the connection weight and the node threshold value if the error input by the output layer does not meet the convergence requirement;
outputting a quality detection result of the ceramic powder if the convergence requirement is met, wherein the quality detection result comprises a high-grade, a medium-grade and a low-grade;
the calculation formula of the error input by the output layer is as follows:
Wherein, Representing the error of the input of the output layer,Represents the learning rate, U represents the single node of the hidden layer, U represents the total number of nodes of the hidden layer, V represents the single node of the output layer, V represents the total number of nodes of the output layer,Representing the output of the hidden layer (u) th node,Representing connection weights for hidden layer and output layer nodes,A node threshold representing an output layer, E representing cross entropy;
Specifically, the low-voltage power distribution network loss analysis network is provided with an input layer, an hidden layer and an output layer, wherein the input layer is used for establishing a characteristic sequence according to input parameters of the low-voltage power distribution network loss analysis network, accumulating the characteristic sequence, establishing a gray differential equation, calculating least square parameters of the gray differential equation, and calculating fitting values of the characteristic sequence by discretizing least square parameters. The hidden layer is used for converting the characteristic sequence fitting value of the input layer into a high-dimensional space through mapping according to the connection weight between the input layer and the hidden layer, summing the characteristic sequence fitting value in the high-dimensional space after nonlinear weighting, and calculating the output value of the hidden layer through offset. The output layer is used for obtaining input of the output layer, calculating average error of input values of the output layer, judging whether the average error meets convergence conditions, updating the connection weight and the node threshold value if the average error does not meet the convergence requirements, and outputting the loss type and the loss rate of the low-voltage distribution network if the average error meets the convergence requirements.
Example 2
Referring to fig. 5, the present invention provides an embodiment: the system comprises a powder sample acquisition module, a powder sample analysis module and a powder sample evaluation module;
the powder sample acquisition module is used for acquiring a ceramic powder sample through the sampling device, and acquiring a ceramic powder static image and a ceramic powder dynamic image of the ceramic powder sample through the image sensor;
The powder sample analysis module is used for processing the ceramic powder static image and the ceramic powder dynamic image to obtain particle distribution characteristics and particle uniformity characteristics;
the powder sample evaluation module is used for training the particle distribution characteristics and the particle uniformity characteristics through a ceramic powder quality evaluation network and outputting a quality detection result of the ceramic powder;
the powder sample collection module includes:
the powder sample sampling unit is used for collecting ceramic powder samples through the sampling device;
the powder static image acquisition unit is used for acquiring a ceramic powder static image according to the digital microscope;
the powder dynamic image acquisition unit is used for acquiring ceramic powder dynamic images according to the high-speed camera;
The powder sample analysis module includes:
the static image processing unit is used for processing the ceramic powder static image and extracting particle distribution characteristics;
the dynamic image processing unit is used for processing the ceramic powder dynamic image and extracting the uniform particle characteristics;
the powder sample evaluation module includes:
The network pre-training unit is used for optimizing network parameters according to the powder quality disclosure data set;
The powder quality evaluation unit is used for training the input parameters through the ceramic powder quality evaluation network and outputting the quality detection result of the ceramic powder.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.
Claims (8)
1. The method for detecting the quality of the high-frequency low-loss ceramic powder is characterized by comprising the following steps of:
s1: collecting a ceramic powder sample through a sampling device, and collecting a ceramic powder static image and a ceramic powder dynamic image through an image sensor;
s2: processing the ceramic powder static image and extracting particle distribution characteristics;
s3: processing the ceramic powder dynamic image, and extracting uniform particle characteristics;
s4: and constructing a ceramic powder quality evaluation network, taking the particle distribution characteristics and the particle uniformity characteristics as input parameters of the ceramic powder quality evaluation network, training the input parameters through the ceramic powder quality evaluation network, and outputting a quality detection result of the ceramic powder.
2. The method for detecting the quality of high-frequency low-loss ceramic powder according to claim 1, wherein the sampling device comprises a powder mixer, a powder sampling port, a powder sampling disc and a conveyor, and the image sensor comprises a digital microscope and a high-speed camera.
3. The method for detecting the quality of high-frequency low-loss ceramic powder according to claim 2, wherein the specific steps of processing the ceramic powder still image are as follows:
s2.1: preprocessing a ceramic powder static image, wherein the preprocessing comprises graying, image sharpening and image denoising;
S2.2: according to the preprocessed ceramic powder static image, calculating the energy gradient of the ceramic powder static image, determining the block number of the ceramic powder static image according to the energy gradient, blocking the ceramic powder static image, obtaining a histogram of each sub-block of the ceramic powder static image, generating an equalization histogram of each sub-block through histogram equalization, and obtaining the ceramic powder static equalization image, wherein the block number of the ceramic powder static image has the following calculation formula:
，
Wherein Q represents the number of blocks of the ceramic powder still image, m { } represents a downward rounding function, a represents a single pixel of the ceramic powder still image, I represents the total number of pixels of the ceramic powder still image, Represents the energy gradient of the a-th pixel point, F represents the total energy gradient, L represents the length of the ceramic powder static image, W represents the width of the ceramic powder static image,Indicating the degree of dispersion of the a-th pixel,Represents the gray-scale distribution kurtosis value of the a-th pixel point,A gray level distribution uniformity value representing the a-th pixel point;
S2.3: performing discrete wavelet transformation on the ceramic powder static equilibrium image to obtain a low-frequency component and a high-frequency component of the ceramic powder static equilibrium image, and enhancing the low-frequency component and the high-frequency component through the segmented normalized detail coefficient to obtain a powder particle enhanced image;
S2.4: calculating the density index of each pixel point, clustering the pixel points with similar density indexes through a clustering algorithm, and acquiring a particle area diagram according to a clustering result;
S2.5: the method comprises the steps of performing weighted sampling on a particle area diagram by an up-sampling function, reducing the number of 4C channels of the particle area diagram to 2C according to regularization and dimension transformation, measuring the particle area diagram, and extracting particle characteristics of the measured particle area diagram through residual connection, wherein the particle characteristics comprise texture characteristics and morphological characteristics;
S2.6: the texture features and the morphological features are fused by convolution kernels of 1×1 size, and the particle distribution features are calculated.
4. The method for detecting the quality of high-frequency low-loss ceramic powder according to claim 3, wherein the step of extracting the texture features is as follows:
s2.5.1: extracting key points of a particle area diagram, taking the key points as the center, placing a first virtual camera according to an included angle of 30 degrees between the camera and the center plane of the model, taking a Z axis as a rotating shaft, placing a next virtual camera every 30 degrees, and obtaining twelve views of a three-dimensional model object, wherein the virtual cameras meet the condition that the shooting direction and the center of the model are on the same straight line;
s2.5.2: constructing a view feature classification network, taking the twelve views as input parameters of the view feature classification network, training the view feature classification network, and outputting classification probability of each view Where P represents the view feature classification probability, Z represents the total number of view feature classes,Representing the classification probability of the view corresponding to the b-th view feature class;
s2.5.3: voting the view feature categories according to the view feature classification probability by each view, and taking the view feature category with the largest statistics vote number as the texture feature of the particle area diagram;
The morphological features include particle area, particle perimeter, particle roundness, particle convex hull area, particle aspect ratio, particle convex perimeter, particle solidity, and particle convexity;
The calculation formula of the particle distribution characteristics is as follows:
，
Wherein, Representing grain distribution characteristics, f representing convolution operations of the convolution kernel, R representing texture characteristics,The characteristic of the morphology is represented by,Representing the largest pixel value in the particle area map,Representing the smallest pixel value in the particle area map,The vector point-of-time is represented,Low frequency feature vectors representing the particle region map,The high-frequency gradient vector representing the particle area map, i represents the particle area map pixel abscissa, j represents the particle area map pixel ordinate, N represents the particle area map horizontal total pixel number, M represents the particle area map vertical total pixel number, and D (i, j) represents the image edge value of the area image at the (i, j) position.
5. The method for detecting the quality of high-frequency low-loss ceramic powder according to claim 2, wherein the specific steps of processing the ceramic powder dynamic image are as follows:
s3.1: preprocessing the ceramic powder dynamic image, wherein the preprocessing comprises linear enhancement and denoising;
S3.2: traversing each pixel point in the pretreated ceramic powder dynamic image, calculating the sum of gray weighted differences between the neighborhood of the pixel point and the pixel point, comparing the sum of the gray weighted differences with a binarization threshold value, and setting the pixel point to be zero if the sum of the gray weighted differences is smaller than the binarization threshold value to obtain a powder binarization image;
S3.3: carrying out peak statistics on the powder binary image through Hough transformation, determining the horizontal coordinate of a parallel line according to the peak value, acquiring a powder flow area according to the upper and lower boundary coordinates of the powder binary image and the horizontal coordinate of the parallel line, projecting all white pixels in the powder flow area, and extracting a flowing powder image;
S3.4: marking pixel edge point coordinates of each flowing powder image, extracting area characteristics of a particle area, and calculating particle flow characteristics according to variances of the area characteristics of the particle area;
S3.5: dividing subareas according to the horizontal axis direction and the vertical axis direction of each flowing powder image in an equidistant mode, extracting area features of the subareas, and calculating particle stability features according to standard deviations of the area features of the subareas;
S3.6: the particle flow characteristics and the particle stability characteristics are fused through convolution kernels with the size of 1 multiplied by 1, the particle uniformity characteristics are calculated, and the calculation formula of the particle uniformity characteristics is as follows:
，
Wherein, Indicating the uniformity of the particles and,Indicating the flow characteristics of the particles, H indicating the stability characteristics of the particles,An average projection matrix of white pixels of a powder flow region is represented, T represents a matrix transpose, Q represents an abscissa of a flowing powder image pixel, K represents an ordinate of a flowing powder image pixel, Q represents a total number of horizontal pixels of the flowing powder image, K represents a total number of vertical pixels of the flowing powder image, I (Q, K) represents a gradation value of the flowing powder image at a (Q, K) position,The average gray value of the flowing powder image is represented.
6. The method for high frequency low loss ceramic powder quality detection of claim 1, wherein the constructing a ceramic powder quality assessment network comprises a network parameter pre-training that optimizes network parameters according to a powder quality disclosure dataset, the network parameters comprising a learning rate, a node threshold, an initial weight, and a cross entropy, the ceramic powder quality assessment network comprising:
the input layer establishes a characteristic sequence according to input parameters and calculates a fitting value of the characteristic sequence;
An implicit layer, converting the fitting value of the characteristic sequence of the input layer into a high-dimensional space, and calculating the output value of the implicit layer in the high-dimensional space;
And the output layer is applied to outputting the quality detection result of the ceramic powder.
7. The method for high-frequency low-loss ceramic powder quality detection according to claim 6, wherein the step of taking the particle distribution characteristics and the particle uniformity characteristics as input parameters of the ceramic powder quality evaluation network, training the input parameters through the ceramic powder quality evaluation network, and outputting a quality detection result of ceramic powder comprises:
initializing the ceramic powder quality evaluation network according to the network parameters, and determining the connection weight and the node threshold of an input layer and an hidden layer;
Biasing input parameters through the connection weights of the input layer and the hidden layer to obtain output of the hidden layer, and updating the output of the hidden layer through the node threshold to obtain input of the output layer;
Calculating an error input by an output layer according to an error function of the output layer, judging whether the error input by the output layer meets a convergence requirement of the output layer, and updating the connection weight and the node threshold value if the error input by the output layer does not meet the convergence requirement;
And outputting a quality detection result of the ceramic powder if the convergence requirement is met, wherein the quality detection result comprises a high-grade, a medium-grade and a low-grade.
8. The system for detecting the quality of the high-frequency low-loss ceramic powder is realized based on the method for detecting the quality of the high-frequency low-loss ceramic powder according to any one of claims 1 to 7, and is characterized by comprising a powder sample acquisition module, a powder sample analysis module and a powder sample evaluation module;
the powder sample acquisition module is used for acquiring a ceramic powder sample through the sampling device, and acquiring a ceramic powder static image and a ceramic powder dynamic image of the ceramic powder sample through the image sensor;
The powder sample analysis module is used for processing the ceramic powder static image and the ceramic powder dynamic image to obtain particle distribution characteristics and particle uniformity characteristics;
the powder sample evaluation module is used for training the particle distribution characteristics and the particle uniformity characteristics through a ceramic powder quality evaluation network and outputting a quality detection result of ceramic powder.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410824929.XA CN118392732B (en) | 2024-06-25 | 2024-06-25 | Method and system for detecting quality of high-frequency low-loss ceramic powder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410824929.XA CN118392732B (en) | 2024-06-25 | 2024-06-25 | Method and system for detecting quality of high-frequency low-loss ceramic powder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118392732A true CN118392732A (en) | 2024-07-26 |
| CN118392732B CN118392732B (en) | 2024-09-27 |
Family
ID=92007968
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410824929.XA Active CN118392732B (en) | 2024-06-25 | 2024-06-25 | Method and system for detecting quality of high-frequency low-loss ceramic powder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118392732B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150077282A1 (en) * | 2013-09-17 | 2015-03-19 | Farrokh Mohamadi | Real-time, two dimensional (2-d) tracking of first responders with identification inside premises |
| CN106018159A (en) * | 2016-05-26 | 2016-10-12 | 廊坊市高瓷新材料科技有限公司 | Detection method for content of ceramic powder of organic ceramic mixture |
| CN109876569A (en) * | 2019-04-19 | 2019-06-14 | 济南大学 | A kind of bag filter purification gas index on-line measuring device, system and method |
| CN113636846A (en) * | 2021-10-14 | 2021-11-12 | 季华实验室 | Ceramic powder modifying method |
| CN116768637A (en) * | 2023-06-20 | 2023-09-19 | 河北鼎瓷电子科技有限公司 | Preparation method of high-toughness high-strength high-heat-conductivity silicon nitride ceramic substrate |
| CN117292076A (en) * | 2023-10-09 | 2023-12-26 | 厦门大学 | Dynamic three-dimensional reconstruction method and system for local operation scene of engineering machinery |
-
2024
- 2024-06-25 CN CN202410824929.XA patent/CN118392732B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150077282A1 (en) * | 2013-09-17 | 2015-03-19 | Farrokh Mohamadi | Real-time, two dimensional (2-d) tracking of first responders with identification inside premises |
| CN106018159A (en) * | 2016-05-26 | 2016-10-12 | 廊坊市高瓷新材料科技有限公司 | Detection method for content of ceramic powder of organic ceramic mixture |
| CN109876569A (en) * | 2019-04-19 | 2019-06-14 | 济南大学 | A kind of bag filter purification gas index on-line measuring device, system and method |
| CN113636846A (en) * | 2021-10-14 | 2021-11-12 | 季华实验室 | Ceramic powder modifying method |
| CN116768637A (en) * | 2023-06-20 | 2023-09-19 | 河北鼎瓷电子科技有限公司 | Preparation method of high-toughness high-strength high-heat-conductivity silicon nitride ceramic substrate |
| CN117292076A (en) * | 2023-10-09 | 2023-12-26 | 厦门大学 | Dynamic three-dimensional reconstruction method and system for local operation scene of engineering machinery |
Non-Patent Citations (4)
| Title |
|---|
| 周传强，等: "铁矿粉制粒特性模糊聚类分析", 烧结球团, vol. 34, no. 3, 30 June 2009 (2009-06-30), pages 16 - 20 * |
| 张光磊，等: "电子封装陶瓷的研究进展", 桃子学报, vol. 42, no. 5, 31 October 2021 (2021-10-31), pages 732 - 740 * |
| 李伟博，等: "掺铝氧化锌纳米粉体的制备及性能研究", 湖北大学学报（自然科学版）, vol. 33, no. 1, 31 March 2011 (2011-03-31), pages 72 - 75 * |
| 郝玉红， 吴立敏: "显微图像法测量颗粒粒径的不确定度评定", 中国粉体技术, vol. 21, no. 3, 30 June 2015 (2015-06-30), pages 73 - 75 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118392732B (en) | 2024-09-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115082467B (en) | Building material welding surface defect detection method based on computer vision | |
| CN113450307A (en) | Product edge defect detection method | |
| CN116664559B (en) | Machine vision-based memory bank damage rapid detection method | |
| CN107622277B (en) | Bayesian classifier-based complex curved surface defect classification method | |
| CN101059425A (en) | Method and device for identifying different variety green tea based on multiple spectrum image texture analysis | |
| CN114549446B (en) | Cylinder liner defect mark detection method based on deep learning | |
| CN114324078B (en) | Particle size identification method, device, equipment and medium | |
| CN116246174B (en) | Sweet potato variety identification method based on image processing | |
| CN114049618B (en) | Graph-point-graph transformation-based pavement three-dimensional disease PCI calculation method | |
| CN112150430A (en) | Numerical analysis method utilizing rock microscopic structure digital image | |
| CN114596551A (en) | Vehicle-mounted forward-looking image crack detection method | |
| CN116309577B (en) | Intelligent detection method and system for high-strength conveyor belt materials | |
| CN116255912A (en) | Method and system for measuring static volume of package | |
| CN118392732B (en) | Method and system for detecting quality of high-frequency low-loss ceramic powder | |
| CN117635615B (en) | Defect detection method and system for realizing punching die based on deep learning | |
| CN112750119B (en) | Detection and measurement method for weak defects on surface of white glass cover plate | |
| CN114627081A (en) | Wheat imperfect grain identification method based on Mask R-CNN | |
| CN116823827B (en) | Ore crushing effect evaluation method based on image processing | |
| CN113052234A (en) | Jade classification method based on image features and deep learning technology | |
| CN116934678A (en) | Method for detecting pit defects of aircraft skin under different scales based on point cloud data | |
| CN115471494A (en) | Wo citrus quality inspection method, device, equipment and storage medium based on image processing | |
| CN114972166B (en) | Aggregate grading data statistics method based on visual detection image | |
| CN115308247B (en) | Method for detecting deslagging quality of aluminum oxide powder | |
| CN117036635B (en) | Shale multi-scale digital rock core construction method based on image texture classification | |
| CN116563286B (en) | Mobile hard disk box production quality rapid detection system |
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 |