CN117849193A - Online crack damage monitoring method for neodymium iron boron sintering - Google Patents
Online crack damage monitoring method for neodymium iron boron sintering Download PDFInfo
- Publication number
- CN117849193A CN117849193A CN202410259815.5A CN202410259815A CN117849193A CN 117849193 A CN117849193 A CN 117849193A CN 202410259815 A CN202410259815 A CN 202410259815A CN 117849193 A CN117849193 A CN 117849193A
- Authority
- CN
- China
- Prior art keywords
- acoustic emission
- emission signal
- feature
- prototype
- training
- 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.)
- Withdrawn
Links
- 229910001172 neodymium magnet Inorganic materials 0.000 title claims abstract description 115
- QJVKUMXDEUEQLH-UHFFFAOYSA-N [B].[Fe].[Nd] Chemical compound [B].[Fe].[Nd] QJVKUMXDEUEQLH-UHFFFAOYSA-N 0.000 title claims abstract description 105
- 238000012544 monitoring process Methods 0.000 title claims abstract description 58
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000005245 sintering Methods 0.000 title claims abstract description 45
- 230000009466 transformation Effects 0.000 claims abstract description 75
- 238000012512 characterization method Methods 0.000 claims abstract description 62
- 238000000605 extraction Methods 0.000 claims abstract description 48
- 239000003623 enhancer Substances 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000005728 strengthening Methods 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 88
- 239000013598 vector Substances 0.000 claims description 76
- 238000013527 convolutional neural network Methods 0.000 claims description 27
- 238000010586 diagram Methods 0.000 claims description 14
- 238000009826 distribution Methods 0.000 claims description 12
- 238000003062 neural network model Methods 0.000 claims description 12
- 230000002787 reinforcement Effects 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 description 14
- 230000008569 process Effects 0.000 description 12
- 238000003860 storage Methods 0.000 description 11
- 238000001514 detection method Methods 0.000 description 8
- 239000011159 matrix material Substances 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000001902 propagating effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000002427 irreversible effect Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/04—Analysing solids
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/14—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object using acoustic emission techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/4454—Signal recognition, e.g. specific values or portions, signal events, signatures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/4481—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/023—Solids
- G01N2291/0234—Metals, e.g. steel
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/028—Material parameters
- G01N2291/0289—Internal structure, e.g. defects, grain size, texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Acoustics & Sound (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
Abstract
Discloses an online crack damage monitoring method for neodymium iron boron sintering. Firstly, carrying out gram angle and field transformation on an acoustic emission signal from a detected neodymium iron boron sintered body to obtain gram angle and field images, then carrying out feature extraction on the gram angle and the field images through an acoustic emission signal feature extractor to obtain an acoustic emission signal transformation domain feature map, carrying out feature map representation strengthening on the acoustic emission signal transformation domain feature map through a feature map enhancer based on a heavy parameterization layer to obtain a strengthened acoustic emission signal transformation domain feature map, then processing the strengthened acoustic emission signal transformation domain feature map through a prototype class feature extraction network to obtain acoustic emission signal prototype characterization features, and finally determining whether crack damage exists in the detected neodymium iron boron sintered body based on the acoustic emission signal prototype characterization features. Thus, whether the neodymium iron boron sintered body has crack damage can be detected in real time and nondestructively.
Description
Technical Field
The application relates to the field of crack damage monitoring, and more particularly relates to an online crack damage monitoring method for neodymium iron boron sintering.
Background
The neodymium iron boron sintered body is a high-performance permanent magnet material and is widely applied to the fields of motors, generators, sensors and the like. However, neodymium iron boron sintered bodies are prone to crack damage during preparation and use, affecting their performance and life. Therefore, it is very important to monitor crack damage of the neodymium iron boron sintered body on line.
However, the conventional crack damage monitoring method is usually performed manually, and has problems of subjectivity and operability, and a professional technician is required for manual monitoring, and for a large-scale produced neodymium iron boron sintered body, the manual monitoring efficiency is low and the cost is high. In addition, conventional monitoring methods typically require removal or downtime of the NdFeB sintered body for inspection, which can lead to downtime in the production process and interruption of the production line. Meanwhile, failure to monitor crack damage in real time may delay the timely discovery and treatment of problems.
Thus, an optimized online crack damage monitoring scheme for neodymium iron boron sintering is desired.
Disclosure of Invention
In view of this, the application provides an online monitoring method for crack damage of neodymium iron boron sintering, which can analyze acoustic emission signals of neodymium iron boron sintering body by introducing signal processing and analysis algorithms based on artificial intelligence and deep learning into data analysis software after collecting the acoustic emission signals, so as to judge whether crack damage exists in the neodymium iron boron sintering body.
According to an aspect of the present application, there is provided an online monitoring method for crack damage of neodymium iron boron sintering, including:
acquiring an acoustic emission signal from the detected neodymium iron boron sintered body;
performing a gram angle and field transformation on the acoustic emission signal to obtain a gram angle and field image;
carrying out feature extraction on the gram angles and the field images through an acoustic emission signal feature extractor based on a deep neural network model so as to obtain an acoustic emission signal transformation domain feature map;
performing feature map representation enhancement on the acoustic emission signal transformation domain feature map by using a feature map enhancer based on a re-parameterized layer to obtain an enhanced acoustic emission signal transformation domain feature map;
processing the enhanced acoustic emission signal transformation domain feature map by using a prototype class feature extraction network to obtain acoustic emission signal prototype characterization feature vectors serving as acoustic emission signal prototype characterization features; and
and determining whether the detected neodymium iron boron sintered body has crack damage or not based on the characterization characteristics of the acoustic emission signal prototype.
In the above online monitoring method for crack damage of neodymium iron boron sintering, the deep neural network model is a convolutional neural network model.
In the above method for online monitoring crack damage of neodymium iron boron sintering, performing feature map representation reinforcement on the acoustic emission signal transformation domain feature map by using a feature map enhancer based on a heavy parameterized layer to obtain a reinforced acoustic emission signal transformation domain feature map, including:
using the characteristic map enhancer based on the re-parameterization layer to carry out characteristic map representation enhancement on the acoustic emission signal transformation domain characteristic map according to the following enhancement formula so as to obtain the enhanced acoustic emission signal transformation domain characteristic map; wherein, the strengthening formula is:
wherein,transforming a global mean of a domain signature for the acoustic emission signal,for the variance of the acoustic emission signal transform domain feature map,is obtained by randomly sampling the Gaussian distribution of the acoustic emission signal transformation domain feature mapThe value of the one of the values,representing the multiplication by the position point,is the characteristic value of each position in the characteristic diagram of the transformation domain of the enhanced acoustic emission signal.
In the above method for online monitoring crack damage of neodymium iron boron sintering, the processing of the reinforced acoustic emission signal transformation domain feature map by using a prototype class feature extraction network to obtain an acoustic emission signal prototype characterization feature vector as an acoustic emission signal prototype characterization feature comprises the following steps:
Processing the enhanced acoustic emission signal transform domain feature map with the prototype formula using the prototype class feature extraction network to obtain the acoustic emission signal prototype characterization feature vector; wherein, the prototype formula is:
wherein,andvectorizing the characteristic matrixes along the channel dimension in the characteristic map of the transformation domain of the enhanced acoustic emission signals to obtain the enhanced acoustic emission signalsIn sequence of transform domain feature vectorsAnd (d)The enhanced acoustic emission signal transforms domain feature vectors,is a sequence of transform domain feature vectors of the enhanced acoustic emission signal,is a norm of the vector which is the one,the number of vectors in the sequence of transform domain feature vectors for the enhanced acoustic emission signal is-1,for the feature values of each position in the semantic fluctuation feature vector of the acoustic emission signal,is the length of the semantic fluctuation feature vector of the acoustic emission signal,is the characteristic vector of the acoustic emission signal prototype,is an exponential operation.
In the above method for online monitoring of crack damage of neodymium iron boron sintering, determining whether the detected neodymium iron boron sintered body has crack damage based on the acoustic emission signal prototype characterization feature comprises the following steps:
and the acoustic emission signal prototype characterization feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the detected neodymium-iron-boron sintered body has crack damage.
In the above online monitoring method for crack damage of neodymium iron boron sintering, the method further comprises the training steps of: the acoustic emission signal feature extractor is used for training the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier.
In the above online monitoring method for crack damage of neodymium iron boron sintering, the training step includes:
training data are acquired, wherein the training data comprise training acoustic emission signals from the detected neodymium iron boron sintered body and true values of whether the detected neodymium iron boron sintered body has crack damage or not;
carrying out gram angle and field transformation on the training acoustic emission signal to obtain training gram angle and field images;
performing feature extraction on the training gram angles and the field images through the acoustic emission signal feature extractor based on the convolutional neural network model to obtain a training acoustic emission signal transformation domain feature map;
performing feature map representation reinforcement on the training acoustic emission signal transformation domain feature map by using the feature map enhancer based on the re-parameterization layer to obtain a training reinforced acoustic emission signal transformation domain feature map;
Processing the training enhanced acoustic emission signal transformation domain feature map by using the prototype class feature extraction network to obtain a training acoustic emission signal prototype characterization feature vector;
passing the training acoustic emission signal prototype characterization feature vector through the classifier to obtain a classification loss function value; and
and training the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier by using the classification loss function value, wherein in each iteration of training, the training acoustic emission signal prototype characterization feature vector is optimized.
In the method, firstly, acoustic emission signals from a detected neodymium iron boron sintered body are subjected to gram angle and field transformation to obtain gram angles and field images, then, the gram angles and the field images are subjected to feature extraction through an acoustic emission signal feature extractor to obtain acoustic emission signal transformation domain feature images, then, the acoustic emission signal transformation domain feature images are subjected to feature image representation enhancement through a feature image enhancer based on a heavy parameterization layer to obtain enhanced acoustic emission signal transformation domain feature images, then, the enhanced acoustic emission signal transformation domain feature images are processed through a prototype category feature extraction network to obtain acoustic emission signal prototype characterization features, and finally, whether crack damage exists in the detected neodymium iron boron sintered body is determined based on the acoustic emission signal prototype characterization features. Thus, whether the neodymium iron boron sintered body has crack damage can be detected in real time and nondestructively.
Other features and aspects of the present application will become apparent from the following detailed description of the application with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments, features and aspects of the present application and together with the description, serve to explain the principles of the present application.
Fig. 1 shows a flow chart of a crack damage on-line monitoring method for neodymium iron boron sintering according to an embodiment of the application.
Fig. 2 shows a schematic architecture diagram of an online monitoring method for crack damage of neodymium iron boron sintering according to an embodiment of the application.
FIG. 3 shows a block diagram of an on-line crack damage monitoring system for NdFeB sintering according to an embodiment of the present application.
Fig. 4 shows an application scenario diagram of a crack damage online monitoring method for neodymium iron boron sintering according to an embodiment of the application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some, but not all embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present application without making any inventive effort, are also within the scope of the present application.
As used in this application and in the claims, the terms "a," "an," "the," and/or "the" are not specific to the singular, but may include the plural, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers indicate identical or functionally similar elements. Although various aspects of the embodiments are illustrated in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
In addition, numerous specific details are set forth in the following detailed description in order to provide a better understanding of the present application. It will be understood by those skilled in the art that the present application may be practiced without some of these specific details. In some instances, methods, means, elements, and circuits have not been described in detail as not to unnecessarily obscure the present application.
It should be appreciated that acoustic emission monitoring is a non-destructive inspection technique that detects acoustic emission signals generated by a material when subjected to a force and analyzes the signals to determine if the material is subject to crack damage. Therefore, in the technical scheme of the application, an online monitoring method for crack damage of neodymium iron boron sintering is provided, and the online monitoring for crack damage of neodymium iron boron sintering can be realized by utilizing an acoustic emission monitoring system. Specifically, the acoustic emission monitoring system mainly comprises an acoustic emission sensor, a signal collector and data analysis software. The acoustic emission sensor is arranged on the neodymium iron boron sintered body, and when the neodymium iron boron sintered body is stressed, acoustic emission signals are generated, and the signals are received by the sensor and transmitted to the signal collector. The signal collector amplifies and digitizes the signal and then transmits it to data analysis software. And the data analysis software analyzes the signals and judges whether the NdFeB sintered body has crack damage or not.
Correspondingly, in the online monitoring process of crack damage of neodymium iron boron sintering by utilizing the acoustic emission monitoring system, in order to ensure the accuracy of crack damage detection, the technical concept of the application is that after acoustic emission signals of neodymium iron boron sintering bodies are collected, signal processing and analysis algorithms based on artificial intelligence and deep learning are introduced into data analysis software to analyze the acoustic emission signals, so that whether the NdFeB sintering bodies have crack damage or not is judged. Therefore, whether the neodymium-iron-boron sintered body has crack damage can be detected in real time and nondestructively, so that early warning can be conveniently carried out in time to take corresponding maintenance measures, and the reliability and the service life of the neodymium-iron-boron sintered magnet are ensured.
Fig. 1 shows a flow chart of a crack damage on-line monitoring method for neodymium iron boron sintering according to an embodiment of the application. Fig. 2 shows a schematic architecture diagram of an online monitoring method for crack damage of neodymium iron boron sintering according to an embodiment of the application. As shown in fig. 1 and fig. 2, the online monitoring method for crack damage of neodymium iron boron sintering according to the embodiment of the application includes the following steps: s110, acquiring an acoustic emission signal from the detected neodymium iron boron sintered body; s120, carrying out the gram angle and field transformation on the acoustic emission signal to obtain gram angle and field images; s130, carrying out feature extraction on the Gellam angles and the field images through an acoustic emission signal feature extractor based on a deep neural network model so as to obtain an acoustic emission signal transformation domain feature map; s140, performing feature map representation enhancement on the acoustic emission signal transformation domain feature map by using a feature map enhancer based on a re-parameterization layer to obtain an enhanced acoustic emission signal transformation domain feature map; s150, processing the enhanced acoustic emission signal transformation domain feature map by using a prototype class feature extraction network to obtain acoustic emission signal prototype characterization feature vectors serving as acoustic emission signal prototype characterization features; and S160, determining whether the detected neodymium iron boron sintered body has crack damage or not based on the acoustic emission signal prototype characterization features.
It should be understood that in step S110, the acoustic emission signal generated by the neodymium iron boron sintered body is acquired by a sensor or other devices, and acoustic emission refers to an acoustic signal generated when a crack growth or deformation occurs in a material or a structure, and these acoustic emission signals may include crack damage information about the material. In step S120, the acoustic emission signal is subjected to a gram angle and field transformation to obtain corresponding gram angle and field images, which can provide information about the spectral features of the acoustic emission signal, facilitating subsequent feature extraction and analysis. In step S130, the deep neural network model is used to process the glahm angle and the field image, and the deep neural network model can learn and capture the features related to crack damage in the acoustic emission signal, so as to improve the accuracy of subsequent crack damage detection. In step S140, a feature map enhancer based on a re-parameterized layer is used to process the acoustic emission signal transform domain feature map so as to enhance the expression capability of the feature map, and the re-parameterized layer is a neural network layer, and the representation mode of the feature map can be changed by adjusting parameters, so that the degree of distinction and importance of the features are improved. In step S150, the enhanced acoustic emission signal transform domain feature map is processed using a prototype-class feature extraction network, which is a neural network model for learning and extracting feature vectors, that can transform complex acoustic emission signals into more differentiated feature representations. In step S160, crack damage is detected and judged by using the prototype characterization feature vector of the acoustic emission signal, and by comparing the known crack damage feature with the prototype characterization feature of the acoustic emission signal, it can be determined whether the detected neodymium-iron-boron sintered body has crack damage, and the result of this step can be used for predicting and evaluating the crack growth condition of the material, and corresponding repair and maintenance measures are adopted. The above is a detailed description of each step in the on-line monitoring method of the crack damage of the neodymium iron boron sintering, and the combination and the flow of the steps aim at accurately detecting and monitoring the crack damage of the neodymium iron boron sintering body through analysis and feature extraction of acoustic emission signals.
Specifically, in the technical scheme of the application, firstly, an acoustic emission signal from a detected neodymium iron boron sintered body is obtained. Next, the acoustic emission signal is subjected to a gram angle and field transformation to obtain a gram angle and field image. It should be appreciated that, since the Gram angle field (Gramian angular field, GAF) is based on Gram principles, it can migrate a time series under a classical cartesian coordinate system to a polar coordinate system for representation, that is, the Gram angle field can convert time series data into image data, can preserve information of signal integrity, and can well preserve the dependence and correlation of the acoustic emission signal on time, with timing characteristics similar to those of the original signal. In particular, the GAF can obtain a glamer angle sum field (Gramian angular sum field, GASF) and a glamer angle difference field (Gramian angular difference field, GADF) according to the difference of trigonometric functions used for encoding, and since the GADF is irreversible after conversion, in the technical scheme of the application, a GASF conversion mode capable of performing inverse conversion is selected to encode the acoustic emission signal. In particular, in one specific example of the present application, the encoding step of the acoustic emission signal to the GASF image is as follows: for a time series of C dimensions = { Q1, Q2, …, QC }, where each dimension contains n sampling points Qi = { Qi1, qi2, …, qi }, the data of each dimension is first normalized. Then, all values in the data are integrated into [ -1,1], and after integration, the normalized numerical value is replaced by the value of the trigonometric function value Cos, and the Cartesian coordinates are replaced by the polar coordinates, so that the absolute time relation of the sequence is reserved.
And then, carrying out feature extraction on the gram angles and the field images by using an acoustic emission signal feature extractor based on a convolutional neural network model, wherein the acoustic emission signal feature extractor has excellent performance in the aspect of implicit feature extraction of the images, so as to extract feature distribution information about acoustic emission signals in the gram angles and the field images, and further obtain an acoustic emission signal transformation domain feature map. It should be appreciated that by using the convolutional neural network model to perform feature analysis and capture of the transform domain of the acoustic emission signal, subtle changes and features in the acoustic emission signal can be captured, thereby facilitating more accurate detection of feature information in the acoustic emission signal that is related to crack damage.
Accordingly, in step S130, the deep neural network model is a convolutional neural network model, that is, the acoustic emission signal feature extractor based on the deep neural network model is an acoustic emission signal feature extractor based on the convolutional neural network model. It should be noted that the convolutional neural network (Convolutional Neural Network, abbreviated as CNN) is a deep learning model, and is particularly suitable for processing data with a grid structure. The core idea of the convolutional neural network is to construct a network structure through components such as a convolutional layer, a pooling layer, a full connection layer and the like. Among them, the convolution layer is an important component of CNN, which performs feature extraction on input data by sliding a convolution kernel (a small matrix of learnable parameters) over the input data. The convolution layer can effectively capture the local relation of the input data, and the characteristic of parameter sharing enables the model to have fewer parameter amounts, so that the risk of overfitting is reduced. Among the acoustic emission signal feature extractors, acoustic emission signal feature extractors based on convolutional neural network models use convolutional layers to process the glahm angle and field images from which features related to crack damage are extracted. The convolutional neural network can automatically extract key features in input data by learning convolutional kernel weights suitable for specific tasks, so that feature extraction and representation of acoustic emission signals are realized. In general, convolutional neural network models are deep learning models that process input data through convolutional layers to extract features related to tasks. Among the acoustic emission signal feature extractors, the acoustic emission signal feature extractor based on the convolutional neural network model is used for extracting features related to crack damage from the gram angle and the field image, and provides useful information for subsequent crack damage detection.
Further, in order to further improve the expression capability and distinguishing property of the acoustic emission signal transformation domain feature map about the acoustic emission signal feature, so as to better distinguish the situation that the neodymium-iron-boron sintered body has crack damage, in the technical scheme of the application, a feature map enhancer based on a heavy parameterization layer is further used for carrying out feature map expression enhancement on the acoustic emission signal transformation domain feature map so as to obtain an enhanced acoustic emission signal transformation domain feature map. By processing the feature map enhancer based on the re-parameterization layer, randomness can be introduced, and the original feature map is re-parameterized into richer feature representation, so that the expression capability of the acoustic emission signal transformation domain feature map is enhanced. In this process, the mean and variance of the acoustic emission signal transform domain feature map are extracted and used to generate a new feature map. The re-parameterized form can be regarded as a mode for data enhancement in a semantic feature space, which is helpful for improving the feature perception and recognition capability of the classifier on crack damages of different neodymium-iron-boron sintered bodies, so that the classifier is better suitable for detecting different crack damages of the neodymium-iron-boron sintered bodies, and the classification accuracy is improved.
Accordingly, in step S140, performing feature map representation enhancement on the acoustic emission signal transform domain feature map using a feature map enhancer based on a re-parameterized layer to obtain an enhanced acoustic emission signal transform domain feature map, including: using the characteristic map enhancer based on the re-parameterization layer to carry out characteristic map representation enhancement on the acoustic emission signal transformation domain characteristic map according to the following enhancement formula so as to obtain the enhanced acoustic emission signal transformation domain characteristic map; wherein, the strengthening formula is:
wherein,transforming a global mean of a domain signature for the acoustic emission signal,for the variance of the acoustic emission signal transform domain feature map,is obtained by randomly sampling the Gaussian distribution of the acoustic emission signal transformation domain feature mapThe value of the one of the values,representing the multiplication by the position point,is the characteristic value of each position in the characteristic diagram of the transformation domain of the enhanced acoustic emission signal.
It should be appreciated that, since the feature vectors contained in each feature matrix along the channel dimension in the transform domain feature map of the enhanced acoustic emission signal represent different feature information about the acoustic emission signal, some of these features have a greater contribution to the crack damage detection of the neodymium-iron-boron sintered body, and some have a lesser contribution, the less-contributing acoustic emission signal feature representation does not reflect the semantic intrinsic features of the acoustic emission signal that are associated with the crack damage detection of the neodymium-iron-boron sintered body. Therefore, in order to reduce the influence of the acoustic emission signal semantics deviating from the essential characteristics on the prototype calculation, in the technical scheme of the application, a prototype class feature extraction network is required to process the enhanced acoustic emission signal transformation domain feature map to obtain an acoustic emission signal prototype characterization feature vector. And processing the enhanced acoustic emission signal transformation domain feature map by using the prototype class feature extraction network, so that feature information related to crack damage in the acoustic emission signal can be extracted, and the features can be used for judging whether the neodymium-iron-boron sintered body has crack damage or not. That is, through the processing of the prototype class feature extraction network, important modes and structures in the acoustic emission signals can be captured, and the normal state and the crack damage state can be differentiated, so that whether the neodymium-iron-boron sintered body has crack damage can be judged more accurately.
Accordingly, in step S150, the enhanced acoustic emission signal transform domain feature map is processed using a prototype class feature extraction network to obtain acoustic emission signal prototype characterization feature vectors as acoustic emission signal prototype characterization features, including: processing the enhanced acoustic emission signal transform domain feature map with the prototype formula using the prototype class feature extraction network to obtain the acoustic emission signal prototype characterization feature vector; wherein, the prototype formula is:
wherein,andvectorizing each feature matrix along the channel dimension in the enhanced acoustic emission signal transform domain feature map to obtain the sequence of enhanced acoustic emission signal transform domain feature vectorsAnd (d)The enhanced acoustic emission signal transforms domain feature vectors,is a sequence of transform domain feature vectors of the enhanced acoustic emission signal,is a norm of the vector which is the one,the number of vectors in the sequence of transform domain feature vectors for the enhanced acoustic emission signal is-1,for the feature values of each position in the semantic fluctuation feature vector of the acoustic emission signal,is the length of the semantic fluctuation feature vector of the acoustic emission signal,is the acoustic emission signalThe prototype characterizes the feature vector(s), Is an exponential operation.
And then, the acoustic emission signal prototype characterization feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the detected NdFeB sintered body has crack damage or not. That is, the prototype characterization characteristic information of the acoustic emission signal is used for classifying treatment, so as to judge whether the NdFeB sintered body has crack damage. Therefore, whether the neodymium-iron-boron sintered body has crack damage can be detected in real time and nondestructively, so that early warning can be conveniently carried out in time to take corresponding maintenance measures, and the reliability and the service life of the neodymium-iron-boron sintered magnet are ensured.
Accordingly, in step S160, based on the acoustic emission signal prototype characterization feature, determining whether the detected neodymium-iron-boron sintered body has crack damage includes: and the acoustic emission signal prototype characterization feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the detected neodymium-iron-boron sintered body has crack damage. Specifically, the acoustic emission signal prototype characterization feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the detected neodymium-iron-boron sintered body has crack damage or not, and the method comprises the following steps: performing full-connection coding on the acoustic emission signal prototype characterization feature vector by using a full-connection layer of the classifier to obtain a coding classification feature vector; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
That is, in the technical solution of the present application, the label of the classifier includes that the detected neodymium-iron-boron sintered body has crack damage (first label) and that the detected neodymium-iron-boron sintered body does not have crack damage (second label), wherein the classifier determines, through a soft maximum function, to which classification label the acoustic emission signal prototype characterization feature vector belongs. It should be noted that the first tag p1 and the second tag p2 do not include the concept of artificial setting, and in fact, during the training process, the computer model does not have the concept of "whether the detected neodymium iron boron sintered body has crack damage", which is only two kinds of classification tags, and the probability that the output feature is under the two classification tags, that is, the sum of p1 and p2 is one. Therefore, the classification result of whether the detected neodymium iron boron sintered body has crack damage is actually converted into a classification probability distribution conforming to the natural rule through classifying the label, and the physical meaning of the natural probability distribution of the label is essentially used instead of the language text meaning of whether the detected neodymium iron boron sintered body has crack damage.
It should be appreciated that the role of the classifier is to learn the classification rules and classifier using a given class, known training data, and then classify (or predict) the unknown data. Logistic regression (logistics), SVM, etc. are commonly used to solve the classification problem, and for multi-classification problems (multi-class classification), logistic regression or SVM can be used as well, but multiple bi-classifications are required to compose multiple classifications, but this is error-prone and inefficient, and the commonly used multi-classification method is the Softmax classification function.
Further, in the technical scheme of the application, the online monitoring method for crack damage of neodymium iron boron sintering further comprises the training steps of: the acoustic emission signal feature extractor is used for training the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier.
It should be appreciated that the training step plays a critical role in the on-line monitoring of crack damage in neodymium iron boron sintering. Through the training step, the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier can be trained, so that the acoustic emission signal feature extractor, the prototype class feature extraction network and the classifier have the capability of extracting, enhancing and classifying acoustic emission signals. Specifically, the training step works as follows: 1. training an acoustic emission signal feature extractor: by marking and training a large number of known acoustic emission signals, the acoustic emission signal feature extractor can be made to learn a feature representation associated with crack damage. Thus, in the subsequent crack damage detection, the feature extractor can accurately extract the features from the acoustic emission signal, and powerful support is provided for judging the crack damage. 2. Training a feature map enhancer: the characteristic map enhancer processes the acoustic emission signal transformation domain characteristic map through the re-parameterization layer, and enhances the expression capability of the characteristic map. Through training, the feature map enhancer can learn an enhancement mode suitable for crack damage detection, so that the acoustic emission signal transformation domain feature map is more beneficial to crack damage judgment. 3. Training a prototype class feature extraction network: the prototype-class feature extraction network is used to learn and extract prototype-characterization feature vectors of acoustic emission signals. Through training, the prototype class feature extraction network can learn the feature representation related to crack damage, so that the prototype characterization feature vector of the acoustic emission signal is more differentiated, and the judgment and classification of the crack damage are facilitated. 4. Training a classifier: the classifier is used for determining crack damage according to the characteristics of the acoustic emission signals. Through training, the classifier can learn the association between the acoustic emission signal characteristics and the crack damage, so that the acoustic emission signal can be accurately classified into the situation with the crack damage or the situation without the crack damage. In summary, the training step trains the key model components by using known labeling data to provide them with the ability to extract, enhance and classify the features of the acoustic emission signal. Therefore, in actual on-line crack damage monitoring, the trained model components can accurately analyze the acoustic emission signals to judge whether the NdFeB sintered body has crack damage.
Wherein, in one example, the training step comprises: training data are acquired, wherein the training data comprise training acoustic emission signals from the detected neodymium iron boron sintered body and true values of whether the detected neodymium iron boron sintered body has crack damage or not; carrying out gram angle and field transformation on the training acoustic emission signal to obtain training gram angle and field images; performing feature extraction on the training gram angles and the field images through the acoustic emission signal feature extractor based on the convolutional neural network model to obtain a training acoustic emission signal transformation domain feature map; performing feature map representation reinforcement on the training acoustic emission signal transformation domain feature map by using the feature map enhancer based on the re-parameterization layer to obtain a training reinforced acoustic emission signal transformation domain feature map; processing the training enhanced acoustic emission signal transformation domain feature map by using the prototype class feature extraction network to obtain a training acoustic emission signal prototype characterization feature vector; passing the training acoustic emission signal prototype characterization feature vector through the classifier to obtain a classification loss function value; and training the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier by using the classification loss function value, wherein in each iteration of the training, the training acoustic emission signal prototype characterization feature vector is optimized.
In the technical scheme of the application, each feature matrix of the training enhancement acoustic emission signal transformation domain feature map expresses the enhanced image semantic features of the gram angle and the field image, so that the image semantic feature distribution of each feature matrix is inconsistent due to the channel distribution difference of the convolutional neural network model. In this way, when the prototype class feature extraction network is used for processing the image semantic feature expression of the training enhancement acoustic emission signal transformation domain feature map based on the feature matrix, the feature distribution information saliency of each feature matrix based on the image semantic feature distribution is also influenced by considering that each feature matrix is inconsistent with the image semantic feature distribution, so that the training acoustic emission signal prototype characterization feature vector is difficult to stably focus on the salient local distribution of the feature in the training process, and the expression effect of the training acoustic emission signal prototype characterization feature vector and the training speed of the model are influenced.
Based on the method, the applicant optimizes the training acoustic emission signal prototype characterization feature vector every time the training acoustic emission signal prototype characterization feature vector is subjected to iteration of classification regression through a classifier.
Accordingly, in one example, in each iteration of the training, the training acoustic emission signal prototype characterization feature vector is optimized with the following optimization formula to obtain an optimized training acoustic emission signal prototype characterization feature vector; wherein, the optimization formula is:
wherein,is the characteristic vector of the prototype characterization of the training acoustic emission signal,andthe training acoustic emission signal prototype characterizes the feature vectorSquare of 1-norm and 2-norm of (c),is the characteristic feature vector of the training acoustic emission signal prototypeAnd (2) length ofIs the weight of the parameter to be exceeded,representing the feature values of each position in the training acoustic emission signal prototype characterization feature vector,a logarithmic function with a base of 2 is shown,and representing the feature values of each position in the feature vector of the optimized training acoustic emission signal prototype characterization feature vector.
In particular, characterizing feature vectors based on the acoustic emission signal prototype by trainingGeometric registration of its high-dimensional feature manifold shape is performed with respect to the scale and structural parameters of the training acoustic emission signal prototype characterization feature vectorFeatures with rich feature semantic information in the feature set formed by the feature values of (a), namely distinguishable stable interest features which represent dissimilarity based on local context information when the classifier classifies, thereby realizing the feature vector characterization of the training acoustic emission signal prototype And the feature information significance is marked in the classification process, so that the training speed of the classifier is improved. Therefore, whether the neodymium iron boron sintered body has crack damage can be detected in real time and nondestructively based on the acoustic emission signal of the neodymium iron boron sintered body, so that early warning can be conveniently carried out in time to take corresponding maintenance measures, and the reliability and the service life of the neodymium iron boron sintered magnet are ensured.
In summary, according to the online monitoring method for crack damage of neodymium iron boron sintering provided by the embodiment of the application, whether crack damage exists in the neodymium iron boron sintered body can be detected in real time and in a nondestructive mode.
Fig. 3 shows a block diagram of an on-line crack damage monitoring system 100 for neodymium iron boron sintering according to an embodiment of the present application. As shown in fig. 3, the online crack damage monitoring system 100 for neodymium iron boron sintering according to an embodiment of the present application includes: a signal acquisition module 110, configured to acquire an acoustic emission signal from the detected neodymium iron boron sintered body; a gram angle and field transformation module 120 for performing gram angle and field transformation on the acoustic emission signal to obtain a gram angle and field image; the acoustic emission signal feature extraction module 130 is configured to perform feature extraction on the gram angle and the field image through an acoustic emission signal feature extractor based on a deep neural network model to obtain an acoustic emission signal transform domain feature map; a feature map enhancing module 140, configured to enhance the feature map of the acoustic emission signal transform domain feature map by using a feature map enhancer based on a heavily parameterized layer to obtain an enhanced acoustic emission signal transform domain feature map; a prototype class feature extraction module 150, configured to process the enhanced acoustic emission signal transform domain feature map using a prototype class feature extraction network to obtain an acoustic emission signal prototype characterization feature vector as an acoustic emission signal prototype characterization feature; and a crack damage analysis module 160, configured to determine whether the detected neodymium iron boron sintered body has crack damage based on the acoustic emission signal prototype characterization feature.
In one possible implementation, the deep neural network model is a convolutional neural network model.
Here, it will be appreciated by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described neodymium iron boron sintering crack damage on-line monitoring system 100 have been described in detail in the above description of the neodymium iron boron sintering crack damage on-line monitoring method with reference to fig. 1 to 2, and thus, repetitive descriptions thereof will be omitted.
As described above, the online monitoring system 100 for crack damage of neodymium iron boron sintering according to the embodiment of the present application may be implemented in various wireless terminals, such as a server with an online monitoring algorithm for crack damage of neodymium iron boron sintering. In one possible implementation, the neodymium iron boron sintered crack damage online monitoring system 100 according to embodiments of the present application may be integrated into a wireless terminal as a software module and/or hardware module. For example, the NdFeB sintered crack damage online monitoring system 100 may be a software module in the operating system of the wireless terminal, or may be an application developed for the wireless terminal; of course, the NdFeB sintered crack damage on-line monitoring system 100 can also be one of many hardware modules of the wireless terminal.
Alternatively, in another example, the neodymium iron boron sintered crack damage online monitoring system 100 and the wireless terminal may be separate devices, and the neodymium iron boron sintered crack damage online monitoring system 100 may be connected to the wireless terminal through a wired and/or wireless network and transmit interactive information in accordance with a agreed data format.
Fig. 4 shows an application scenario diagram of a crack damage online monitoring method for neodymium iron boron sintering according to an embodiment of the application. As shown in fig. 4, in this application scenario, firstly, an acoustic emission signal (for example, D illustrated in fig. 4) from a detected neodymium iron boron sintered body is acquired, and then, the acoustic emission signal is input into a server (for example, S illustrated in fig. 4) deployed with a neodymium iron boron sintered crack damage online monitoring algorithm, where the server can process the acoustic emission signal by using the neodymium iron boron sintered crack damage online monitoring algorithm to obtain a classification result for indicating whether the detected neodymium iron boron sintered body has crack damage.
In an exemplary embodiment, a non-transitory computer readable storage medium is also provided, such as a memory including computer program instructions executable by a processing component of an apparatus to perform the above-described method.
The present application may be a system, method, and/or computer program product. The computer program product may include a computer readable storage medium having computer readable program instructions embodied thereon for causing a processor to implement aspects of the present application.
The computer readable storage medium may be a tangible device that can hold and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: portable computer disks, hard disks, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static Random Access Memory (SRAM), portable compact disk read-only memory (CD-ROM), digital Versatile Disks (DVD), memory sticks, floppy disks, mechanical coding devices, punch cards or in-groove structures such as punch cards or grooves having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media, as used herein, are not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., optical pulses through fiber optic cables), or electrical signals transmitted through wires.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The embodiments of the present application have been described above, the foregoing description is exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
Claims (7)
1. The online crack damage monitoring method for neodymium iron boron sintering is characterized by comprising the following steps of:
acquiring an acoustic emission signal from the detected neodymium iron boron sintered body;
performing a gram angle and field transformation on the acoustic emission signal to obtain a gram angle and field image;
carrying out feature extraction on the gram angles and the field images through an acoustic emission signal feature extractor based on a deep neural network model so as to obtain an acoustic emission signal transformation domain feature map;
performing feature map representation enhancement on the acoustic emission signal transformation domain feature map by using a feature map enhancer based on a re-parameterized layer to obtain an enhanced acoustic emission signal transformation domain feature map;
processing the enhanced acoustic emission signal transformation domain feature map by using a prototype class feature extraction network to obtain acoustic emission signal prototype characterization feature vectors serving as acoustic emission signal prototype characterization features; and determining whether the detected NdFeB sintered body has crack damage or not based on the acoustic emission signal prototype characterization features.
2. The online monitoring method of crack damage of neodymium iron boron sintering according to claim 1, wherein the deep neural network model is a convolutional neural network model.
3. The method of claim 2, wherein the enhancing the acoustic emission signal transform domain signature using a signature enhancer based on a heavily parameterized layer to obtain an enhanced acoustic emission signal transform domain signature comprises:
using the characteristic map enhancer based on the re-parameterization layer to carry out characteristic map representation enhancement on the acoustic emission signal transformation domain characteristic map according to the following enhancement formula so as to obtain the enhanced acoustic emission signal transformation domain characteristic map; wherein, the strengthening formula is:
wherein (1)>Transforming a global mean value of a domain feature map for said acoustic emission signal,/for>For the acoustic emission signal the variance of the transform domain feature map,/->Is obtained by randomly sampling the Gaussian distribution of the acoustic emission signal transformation domain feature map>Personal value (s)/(s)>Representing multiplication by location +.>Is the characteristic value of each position in the characteristic diagram of the transformation domain of the enhanced acoustic emission signal.
4. A method of on-line crack damage monitoring for neodymium-iron-boron sintering according to claim 3, wherein processing the enhanced acoustic emission signal transform domain feature map using a prototype class feature extraction network to obtain acoustic emission signal prototype characterization feature vectors as acoustic emission signal prototype characterization features comprises:
Processing the enhanced acoustic emission signal transform domain feature map with the prototype formula using the prototype class feature extraction network to obtain the acoustic emission signal prototype characterization feature vector; wherein, the prototype formula is:
wherein,/>and->The first part of the sequence of the characteristic vectors of the transformation domain of the enhanced acoustic emission signal, which is obtained by vectorizing and arranging all characteristic matrixes along the channel dimension in the characteristic diagram of the transformation domain of the enhanced acoustic emission signal>And->Transformation domain feature vectors of enhanced acoustic emission signals +.>Is a sequence of transformation domain feature vectors of the enhanced acoustic emission signal,>is a norm of the vector, +.>-1, < > -number of vectors in the sequence of transform domain feature vectors for said enhanced acoustic emission signal>For the characteristic value of each position in the semantic fluctuation characteristic vector of the acoustic emission signal,/for the acoustic emission signal>Is the length of the semantic fluctuation feature vector of the acoustic emission signal, < >>Is the characterization feature vector of the acoustic emission signal prototype, < >>Is an exponential operation.
5. The on-line monitoring method of crack damage for neodymium iron boron sintering according to claim 4, wherein determining whether the detected neodymium iron boron sintered body has crack damage based on the acoustic emission signal prototype characterization feature comprises:
And the acoustic emission signal prototype characterization feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the detected neodymium-iron-boron sintered body has crack damage.
6. The on-line monitoring method of crack damage in neodymium iron boron sintering according to claim 5, further comprising the training step of: the acoustic emission signal feature extractor is used for training the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier.
7. The online monitoring method of crack damage in neodymium iron boron sintering according to claim 6, wherein the training step comprises:
training data are acquired, wherein the training data comprise training acoustic emission signals from the detected neodymium iron boron sintered body and true values of whether the detected neodymium iron boron sintered body has crack damage or not;
carrying out gram angle and field transformation on the training acoustic emission signal to obtain training gram angle and field images;
performing feature extraction on the training gram angles and the field images through the acoustic emission signal feature extractor based on the convolutional neural network model to obtain a training acoustic emission signal transformation domain feature map;
Performing feature map representation reinforcement on the training acoustic emission signal transformation domain feature map by using the feature map enhancer based on the re-parameterization layer to obtain a training reinforced acoustic emission signal transformation domain feature map;
processing the training enhanced acoustic emission signal transformation domain feature map by using the prototype class feature extraction network to obtain a training acoustic emission signal prototype characterization feature vector;
passing the training acoustic emission signal prototype characterization feature vector through the classifier to obtain a classification loss function value; and training the acoustic emission signal feature extractor based on the convolutional neural network model, the feature map enhancer based on the re-parameterization layer, the prototype class feature extraction network and the classifier by using the classification loss function value, wherein in each iteration of the training, the training acoustic emission signal prototype characterization feature vector is optimized.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410259815.5A CN117849193A (en) | 2024-03-07 | 2024-03-07 | Online crack damage monitoring method for neodymium iron boron sintering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410259815.5A CN117849193A (en) | 2024-03-07 | 2024-03-07 | Online crack damage monitoring method for neodymium iron boron sintering |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117849193A true CN117849193A (en) | 2024-04-09 |
Family
ID=90543773
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410259815.5A Withdrawn CN117849193A (en) | 2024-03-07 | 2024-03-07 | Online crack damage monitoring method for neodymium iron boron sintering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117849193A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118035798A (en) * | 2024-04-11 | 2024-05-14 | 克拉玛依市城投油砂矿勘探有限责任公司 | Intelligent monitoring system and method for oil sand production |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113313059A (en) * | 2021-06-16 | 2021-08-27 | 燕山大学 | One-dimensional spectrum classification method and system |
| CN114487129A (en) * | 2022-01-24 | 2022-05-13 | 中北大学 | Damage identification method for flexible materials based on acoustic emission technology |
| CN115187832A (en) * | 2022-06-24 | 2022-10-14 | 同济大学 | Energy system fault diagnosis method based on deep learning and gram angular field image |
| CN115456012A (en) * | 2022-08-24 | 2022-12-09 | 华能新能源股份有限公司 | Wind power plant fan major component state monitoring system and method |
| CN115791969A (en) * | 2022-12-08 | 2023-03-14 | 深圳量云能源网络科技有限公司 | Jacket underwater crack detection system and method based on acoustic emission signals |
| CN116147548A (en) * | 2023-04-19 | 2023-05-23 | 西南林业大学 | Non-destructive testing method and system for steel fiber RPC cover plate thickness |
| WO2023158360A1 (en) * | 2022-02-18 | 2023-08-24 | Telefonaktiebolaget Lm Ericsson (Publ) | Evaluation of performance of an ae-encoder |
-
2024
- 2024-03-07 CN CN202410259815.5A patent/CN117849193A/en not_active Withdrawn
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113313059A (en) * | 2021-06-16 | 2021-08-27 | 燕山大学 | One-dimensional spectrum classification method and system |
| CN114487129A (en) * | 2022-01-24 | 2022-05-13 | 中北大学 | Damage identification method for flexible materials based on acoustic emission technology |
| WO2023158360A1 (en) * | 2022-02-18 | 2023-08-24 | Telefonaktiebolaget Lm Ericsson (Publ) | Evaluation of performance of an ae-encoder |
| CN115187832A (en) * | 2022-06-24 | 2022-10-14 | 同济大学 | Energy system fault diagnosis method based on deep learning and gram angular field image |
| CN115456012A (en) * | 2022-08-24 | 2022-12-09 | 华能新能源股份有限公司 | Wind power plant fan major component state monitoring system and method |
| CN115791969A (en) * | 2022-12-08 | 2023-03-14 | 深圳量云能源网络科技有限公司 | Jacket underwater crack detection system and method based on acoustic emission signals |
| CN116147548A (en) * | 2023-04-19 | 2023-05-23 | 西南林业大学 | Non-destructive testing method and system for steel fiber RPC cover plate thickness |
Non-Patent Citations (4)
| Title |
|---|
| CHEN, L等: "GAF and CBAM-ResNet:An Efficient Combination for Identifying the Degree of Safety Valve Leakage", 2023 CAA SYMPOSIUM ON FAULT DETECTION, SUPERVISON AND SAFETY FOR TECHNICAL PROCESSES, 23 November 2023 (2023-11-23), pages 1 - 5 * |
| 古莹奎等: "基于格拉姆角场与深度卷积生成对抗网络的行星齿轮箱故障诊断", 噪声与振动控制, vol. 44, no. 01, 18 February 2024 (2024-02-18), pages 111 - 118 * |
| 古莹奎等: "基于格拉姆角场和迁移深度残差神经网络的滚动轴承故障诊断", 振动与冲击, vol. 41, no. 21, 15 November 2022 (2022-11-15), pages 228 - 237 * |
| 银磊等: "基于声发射信号特征的海上风电场导管架裂纹检测系统", 中国新通信, vol. 25, no. 14, 20 July 2023 (2023-07-20), pages 39 - 41 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118035798A (en) * | 2024-04-11 | 2024-05-14 | 克拉玛依市城投油砂矿勘探有限责任公司 | Intelligent monitoring system and method for oil sand production |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Automatic fault diagnosis of infrared insulator images based on image instance segmentation and temperature analysis | |
| CN111860573B (en) | Model training method, image category detection method and device and electronic equipment | |
| CN110569901B (en) | A weakly supervised object detection method based on channel selection for adversarial elimination | |
| CN117784710B (en) | Remote state monitoring system and method for numerical control machine tool | |
| CN116310850B (en) | Remote sensing image target detection method based on improved RetinaNet | |
| CN116012709B (en) | High-resolution remote sensing image building extraction method and system | |
| CN117849193A (en) | Online crack damage monitoring method for neodymium iron boron sintering | |
| CN117274212A (en) | Bridge underwater structure crack detection method | |
| CN117951646A (en) | A data fusion method and system based on edge cloud | |
| CN115311553A (en) | Target detection method and device, electronic equipment and storage medium | |
| CN117705059B (en) | Positioning method and system for remote sensing mapping image of natural resource | |
| Sun et al. | Substation high-voltage switchgear detection based on improved EfficientNet-YOLOv5s model | |
| CN116205905B (en) | Method and system for image detection of distribution network construction safety and quality based on mobile terminal | |
| CN118294455A (en) | Intelligent detection system and method for neodymium-iron-boron electroplated blank | |
| CN118071714A (en) | Transmission Line Insulator Defect Detection Method Based on CNN-BiGRU | |
| CN118097415A (en) | Video automatic inspection early warning platform based on RPA and early warning method thereof | |
| CN117237930A (en) | Etching hardware SEM image identification method based on ResNet and transfer learning | |
| CN110728316A (en) | Classroom behavior detection method, system, device and storage medium | |
| CN112348060B (en) | Classification vector generation method, device, computer equipment and storage medium | |
| Da et al. | Remote sensing image ship detection based on improved YOLOv3 | |
| Zhou et al. | Target recognition and evaluation of typical transmission line equipment based on deep learning | |
| CN113971737A (en) | Object recognition methods, electronic devices, media and program products for use in robots | |
| CN112733670A (en) | Fingerprint feature extraction method and device, electronic equipment and storage medium | |
| CN117953589B (en) | Interactive action detection method, system, device and medium | |
| CN112446440B (en) | Multi-sensor target tracking method of robot based on MSR-CNN |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20240409 |