CN116703642A - Intelligent management system of product manufacturing production line based on digital twin technology - Google Patents
Intelligent management system of product manufacturing production line based on digital twin technology Download PDFInfo
- Publication number
- CN116703642A CN116703642A CN202310998683.3A CN202310998683A CN116703642A CN 116703642 A CN116703642 A CN 116703642A CN 202310998683 A CN202310998683 A CN 202310998683A CN 116703642 A CN116703642 A CN 116703642A
- Authority
- CN
- China
- Prior art keywords
- training
- preset time
- power
- vector
- product manufacturing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 68
- 238000005516 engineering process Methods 0.000 title claims abstract description 39
- 239000011159 matrix material Substances 0.000 claims abstract description 44
- 238000004458 analytical method Methods 0.000 claims abstract description 23
- 239000013598 vector Substances 0.000 claims description 118
- 238000012549 training Methods 0.000 claims description 87
- 238000007726 management method Methods 0.000 claims description 40
- 238000005070 sampling Methods 0.000 claims description 27
- 238000000034 method Methods 0.000 claims description 26
- 238000010586 diagram Methods 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 24
- 238000005457 optimization Methods 0.000 claims description 14
- 230000002708 enhancing effect Effects 0.000 claims description 11
- 230000003014 reinforcing effect Effects 0.000 claims 1
- 230000006870 function Effects 0.000 description 18
- 230000008569 process Effects 0.000 description 10
- 238000013528 artificial neural network Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 5
- 230000004913 activation Effects 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- 206010000117 Abnormal behaviour Diseases 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 108010001267 Protein Subunits Proteins 0.000 description 1
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- 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
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Manufacturing & Machinery (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
An intelligent management system for the production line of product based on digital twin technology is disclosed. Firstly, vibration signals of monitored equipment in a preset time period and power values of the monitored equipment at a plurality of preset time points in the preset time period, which are collected by cameras arranged on a product manufacturing production line, are acquired, then, joint analysis is carried out on the vibration signals and the power values of the preset time points to obtain a multi-mode equipment state characteristic matrix, and then, whether the working state of the monitored equipment is normal or not is determined based on the multi-mode equipment state characteristic matrix. Thus, equipment faults and production line stoppage can be avoided, and production loss is reduced.
Description
Technical Field
The present disclosure relates to the field of intelligent management, and more particularly, to an intelligent management system for a product manufacturing line based on digital twin technology.
Background
In a product manufacturing line, whether equipment is in a normal working state is critical to production efficiency and product quality. By monitoring the working state of the equipment, the problems of abrasion, fatigue, overload and the like of the equipment can be found in time, and the equipment is prevented from running under overload or abnormal working state, so that the service life of the equipment is prolonged, and the reliability and stability of the equipment are improved.
However, the conventional device monitoring and management methods have some limitations, such as difficulty in setting a threshold value, high cost of manual intervention, and the like. Thus, an optimized solution is desired.
Disclosure of Invention
In view of this, the disclosure provides a product manufacturing production line intelligent management system based on digital twin technology, which can realize analysis and evaluation of the working state of the monitored equipment, discover abnormal behavior and potential fault signs of the equipment in time, take maintenance and repair measures in advance, avoid equipment faults and production line shutdown, and reduce production loss.
According to an aspect of the present disclosure, there is provided a digital twin technology-based product manufacturing line intelligent management system, including:
the device comprises a data acquisition module, a control module and a control module, wherein the data acquisition module is used for acquiring vibration signals of monitored equipment in a preset time period and power values of the monitored equipment at a plurality of preset time points in the preset time period, wherein the vibration signals are acquired by cameras arranged on a product manufacturing production line;
the joint analysis module is used for carrying out joint analysis on the vibration signals and the power values of the plurality of preset time points to obtain a multi-mode equipment state characteristic matrix; and the working state judging module is used for determining whether the working state of the monitored equipment is normal or not based on the multi-mode equipment state feature matrix.
According to the embodiment of the disclosure, firstly, vibration signals of monitored equipment in a preset time period and power values of the monitored equipment at a plurality of preset time points in the preset time period, which are acquired by cameras arranged on a product manufacturing production line, are acquired, then, joint analysis is carried out on the vibration signals and the power values of the preset time points to obtain a multi-mode equipment state characteristic matrix, and then, whether the working state of the monitored equipment is normal or not is determined based on the multi-mode equipment state characteristic matrix. Thus, equipment faults and production line stoppage can be avoided, and production loss is reduced.
Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds 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 disclosure and together with the description, serve to explain the principles of the disclosure.
FIG. 1 illustrates a block diagram of a digital twinning technology based product manufacturing line intelligent management system, in accordance with an embodiment of the present disclosure.
FIG. 2 illustrates a block diagram of the joint analysis module in a digital twinning technology based product manufacturing line intelligent management system, in accordance with an embodiment of the present disclosure.
FIG. 3 illustrates a block diagram of the structuring and enhancing processing unit in a product manufacturing line intelligent management system based on digital twinning techniques in accordance with an embodiment of the present disclosure.
Fig. 4 shows a block diagram of the operating state determination module in the intelligent management system of the product manufacturing line based on the digital twin technology according to an embodiment of the present disclosure.
FIG. 5 illustrates a block diagram of a training module further included in a digital twinning technology based product manufacturing line intelligent management system, in accordance with an embodiment of the present disclosure.
Fig. 6 shows a flowchart of a method of intelligent management of a product manufacturing line based on digital twinning technology in accordance with an embodiment of the present disclosure.
Fig. 7 shows an architectural diagram of a product manufacturing line intelligent management method based on digital twinning techniques in accordance with an embodiment of the present disclosure.
Fig. 8 illustrates an application scenario diagram of a digital twinning technology-based product manufacturing line intelligent management system, according to an embodiment of the present disclosure.
Detailed Description
The following description of the embodiments of the present disclosure 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 disclosure. All other embodiments, which can be made by one of ordinary skill in the art without undue burden based on the embodiments of the present disclosure, are also within the scope of the present disclosure.
As used in this disclosure and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, 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 disclosure will be described in detail below with reference to the 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 disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art have not been described in detail in order not to obscure the present disclosure.
Aiming at the technical problems, the technical conception of the present disclosure is as follows: the method utilizes a digital twin technology and an artificial intelligence technology based on deep learning, comprehensively utilizes vibration signals and power values of monitored equipment to analyze and evaluate the working state of the monitored equipment, timely discovers abnormal behavior and potential fault signs of the equipment, adopts maintenance and repair measures in advance, avoids equipment faults and production line shutdown, and reduces production loss.
Based on this, fig. 1 shows a block diagram schematic of a digital twinning technology based product manufacturing line intelligent management system according to an embodiment of the present disclosure. As shown in fig. 1, a digital twin technology-based product manufacturing line intelligent management system 100 according to an embodiment of the present disclosure includes: a data acquisition module 110, configured to acquire a vibration signal of a monitored device acquired by a camera disposed in a product manufacturing line during a predetermined time period and power values of the monitored device at a plurality of predetermined time points during the predetermined time period; the joint analysis module 120 is configured to perform joint analysis on the vibration signal and the power values at the plurality of predetermined time points to obtain a multi-mode device state feature matrix; and a working state judging module 130, configured to determine whether the working state of the monitored device is normal based on the multi-mode device state feature matrix.
Accordingly, in the technical scheme of the present disclosure, first, a vibration signal of a monitored device in a predetermined time period collected by a camera disposed in a product manufacturing line and power values of the monitored device at a plurality of predetermined time points in the predetermined time period are acquired.
And then, carrying out data structuring and data enhancement processing on power values of a plurality of preset time points of the monitored equipment in the preset time period to obtain an up-sampling power time sequence input vector. That is, the power values as time-series discrete data are converted into structured vector representations, and then the up-sampling power time-series input vectors have denser data points through a data enhancement means. Since it is very difficult to capture dynamic change characteristics about power from a small number of power values, the problem of weak feature expression capability due to data sparsity can be greatly reduced by a processing means of data enhancement such as up-sampling.
In a specific example of the present disclosure, the encoding process of performing data structuring and data enhancing processing on power values of the monitored device at a plurality of predetermined time points within the predetermined time period to obtain an up-sampling power timing input vector includes: firstly, arranging power values of a plurality of preset time points of the monitored equipment in the preset time period into power time sequence input vectors according to a time dimension; the power timing input vector is then up-sampled based on linear interpolation to obtain an up-sampled power timing input vector.
Where linear interpolation is often used to estimate the value of an unknown point between known data points. In a product manufacturing line intelligent management system, linear interpolation is used to up-sample the power timing input vector, resulting in denser data points. Here, the power timing input vector is a power value recorded at a plurality of predetermined time points within a predetermined period. The power timing input vector is up-sampled using a linear interpolation method, and the power values at these time points can be estimated by linear interpolation between known time points. Specifically, linear interpolation takes a straight line segment between two known data points as an interpolation function. And calculating the slope between two points according to the two adjacent power values, and interpolating on the straight line according to the time point of the required interpolation, thereby obtaining the corresponding power value.
And then, passing the waveform diagram of the vibration signal and the up-sampling power time sequence input vector through a CLIP model comprising an image encoder and a sequence encoder to obtain the multi-mode device state characteristic matrix.
Accordingly, as shown in fig. 2, the joint analysis module 120 includes: a structuring and enhancing processing unit 121, configured to perform data structuring and data enhancing processing on power values of the monitored device at a plurality of predetermined time points in the predetermined time period to obtain an up-sampling power timing input vector; and an encoding unit 122, configured to pass the waveform diagram of the vibration signal and the up-sampling power timing input vector through a CLIP model including an image encoder and a sequence encoder to obtain the multi-mode device state feature matrix. It should be understood that the joint analysis module 120 is a functional module, and includes two sub-units, namely a structuring and enhancement processing unit 121 and an encoding unit 122. The structuring and enhancing processing unit 121 is configured to perform data structuring and data enhancing processing on power values of the monitored device at a plurality of predetermined time points within a predetermined period, and its main function is to process original power data, so that the original power data has a certain structure and enhanced characteristics, so as to be better used for subsequent analysis and modeling. This unit may sort and organize the raw power data into a form and structure, such as chronological or segmented power values; the structured power data may be subjected to enhancement processing to extract more features or increase the diversity of the data, which may include techniques such as data interpolation, filtering, noise reduction, smoothing, etc., to improve the quality and expressive power of the data; the processed power data may be converted into an up-sampled power timing input vector, which is a high-dimensional vector containing power value information at multiple points in time. Encoding unit 122: the main function of (a) is to input the waveform diagram of the vibration signal and the up-sampled power timing input vector into a multi-modal model called CLIP model and generate a multi-modal device state feature matrix. Specifically, the encoding unit 122 converts the waveform diagram of the vibration signal into an image representation so as to be able to be processed in the CLIP model, which generally involves feature extraction and encoding of the image using a Convolutional Neural Network (CNN); the upsampled power timing input vector is converted to a sequence representation to enable processing in the CLIP model, which may involve learning timing relationships and features in the sequence using a Recurrent Neural Network (RNN) or other sequence modeling technique. The CLIP model is a multi-modal model that processes both image and text data and inputs the outputs of the image encoder and the sequence encoder into a shared neural network to learn the characteristic representation of the multi-modal data. Finally, the CLIP model generates a multi-modal device state characterization matrix that includes a fused representation of the vibration signal and the power timing data. In other words, the structuring and enhancing processing unit 121 is responsible for processing and enhancing the power data, and the encoding unit 122 converts the processed power data and vibration signals into a multi-modal device state feature matrix for subsequent analysis and application.
Among others, it is worth mentioning that an image encoder is a module for converting an image into a representation vector, which is generally based on a deep learning technique, in particular a convolutional neural network (Convolutional Neural Network, CNN), for extracting features of the image and encoding it into a fixed length vector representation. The main function of the image encoder is to convert the image data into a more expressive and processable form for use in subsequent tasks. The image encoder can extract rich features from the original image through convolution, pooling and other operations, and the features can capture visual information such as edges, textures, shapes and the like in the image, so that the content of the image can be expressed better. The image encoder can reduce the high-dimensional image data into a low-dimensional vector representation, the process of reducing and compressing can reduce the storage and calculation cost of the data, and redundant information can be removed, so that main visual characteristics are reserved. By encoding images as vector representations, the distance or similarity measure between the vectors can be used to compare similarity between images, which is very useful for tasks such as image searching, clustering and classification, and similar images can be quickly found or grouped into different categories. The image encoder can be used in combination with encoders of other modal data (such as text, audio and the like) to realize fusion and joint analysis of the multi-modal data, and the data of different modalities can be jointly processed in a shared neural network by encoding the data into vector representations, so that richer multi-modal characteristic representations are obtained. That is, an image encoder is a key component that converts image data into a vector representation, which is capable of extracting features of an image and converting it into a form that is more convenient to process and analyze, providing a basis for various image-related tasks.
A sequence encoder is a module or algorithm for converting sequence data (e.g., time series, text series, etc.) into a fixed length vector representation, which is typically based on a recurrent neural network (Recurrent Neural Network, RNN) or other sequence modeling technique. The main function of the sequence encoder is to capture the time sequence relation and characteristics in the sequence data and encode the time sequence relation and characteristics into a vector representation with fixed length, and the vector representation can better represent the semantics and the context information of the sequence data, thereby facilitating the subsequent processing and analysis. The sequence encoder is used for converting a text sequence into a vector representation, for example, converting sentences into a semantic vector representation or converting documents into document embedding vectors, and can also be applied to other fields, such as time sequence analysis, audio processing, etc., for extracting important features of sequence data. By means of the sequence encoder, the sequence data can be converted into a vector representation of a fixed length, facilitating subsequent machine learning, deep learning or other analysis tasks, such as classification, clustering, generation, etc. The vector representation can better capture the semantic and contextual information of the sequence data, and improve the expression capacity and performance of the model.
More specifically, as shown in fig. 3, the structuring and enhancement processing unit 121 includes: an input vector arrangement subunit 1211, configured to arrange power values of the monitored device at a plurality of predetermined time points in the predetermined time period into a power timing input vector according to a time dimension; and an upsampling subunit 1212 configured to upsample the power timing input vector based on linear interpolation to obtain the upsampled power timing input vector. It will be appreciated that the input vector arrangement sub-unit 1211 functions to arrange the power values of the monitored device at a plurality of predetermined points in time within a predetermined period of time in a time dimension as a power time series input vector, and in particular, it arranges the plurality of power values in a time series to form a time series power vector, with the purpose of incorporating time dimension information into subsequent processing and analysis to better understand and predict the state change of the device. The upsampling subunit 1212 performs upsampling based on linear interpolation on the power timing input vector to obtain an upsampled power timing input vector, where the upsampling is to increase the sampling rate of the original data, that is, increase the number of data points, and by upsampling, more data points can be increased in the time dimension, so that the power timing input vector is finer and smoother, and linear interpolation is a common upsampling method, which generates new data points by performing linear interpolation calculation between the original data points, and the upsampled power timing input vector can provide more detailed and accurate time information, which is helpful for subsequent analysis and processing tasks. In other words, the input vector arrangement subunit is responsible for arranging the power values into the power timing input vectors according to the time dimension, and the up-sampling subunit performs up-sampling processing on the power timing input vectors to obtain more detailed and smooth up-sampled power timing input vectors, and the two subunits function to provide the input data with the more time dimension for the subsequent analysis module so as to better understand and process the power variation of the monitored device.
Further, the multi-mode equipment state feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the monitored equipment is normal or not; and finally, displaying the classification result on a screen.
Accordingly, as shown in fig. 4, the working state determining module 130 includes: a classification unit 131, configured to pass the multi-mode device state feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate whether the working state of the monitored device is normal; and a display unit 132 for displaying the classification result on a screen.
That is, in the technical solution of the present disclosure, the label of the classifier includes that the operation state of the monitored device is normal (first label) and that the operation state of the monitored device is abnormal (second label), where the classifier determines, through a soft maximum function, to which classification label the multi-mode device state feature matrix belongs. It should be noted that the first tag p1 and the second tag p2 do not include a concept set by human, and in fact, during the training process, the computer model does not have a concept of "whether the operation state of the monitored device is normal", which is simply that there are two kinds of classification tags and the probability that the output feature is under the two kinds of classification tags, that is, the sum of p1 and p2 is one. Therefore, the classification result of whether the working state of the monitored equipment is normal is actually converted into the classification probability distribution conforming to the natural rule through classifying the labels, and the physical meaning of the natural probability distribution of the labels is essentially used instead of the language text meaning of whether the working state of the monitored equipment is normal.
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.
More specifically, the classifying unit 131 is further configured to: expanding the multi-mode equipment state feature matrix into classification feature vectors according to row vectors or column vectors; performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
It should be noted that the full-connection coding refers to a process of coding the input feature vector through the full-connection layer. Fully connected layers are a common layer type in neural networks, where each input neuron is connected to all neurons of the next layer. In full-connection coding, the input feature vector is subjected to linear transformation and nonlinear activation through a series of full-connection layers, and finally the coded feature vector is obtained. The purpose of full-connection coding is to convert the input feature vector into a coding feature vector with more expressive power and discriminant through the learning capability of a neural network. Through the stacking of multiple fully connected layers and the action of nonlinear activation functions, the network can learn complex relationships between input features and higher-order feature representations. Such encoded feature vectors typically have better representation capabilities, which better distinguish differences between different classes, thereby improving the performance of the classifier. Full-join coding may be trained by back-propagation algorithms so that the network can automatically learn a representation of features appropriate for a particular task. In the training process, the network performs parameter updating according to the loss function of the classification task, so as to optimize the representation capability of the coding feature vector. After training, the fully-connected coding layer can map the input feature vector to a low-dimensional coding space, wherein samples of different categories have larger separation degree in the coding space. In summary, full-concatenated coding converts an input feature vector into a coded feature vector that is more expressive and discriminant by the stacking of multiple full-concatenated layers and the action of a nonlinear activation function. Such encoded feature vectors may improve the performance of the classifier, enabling classification tasks to more accurately classify input samples.
Further, the intelligent management system of the product manufacturing production line based on the digital twin technology further comprises a training module for training the CLIP model comprising the image encoder and the sequence encoder and the classifier; as shown in fig. 5, the training module 200 includes: a training data obtaining unit 210, configured to obtain training data, where the training data includes a training vibration signal of a monitored device in a predetermined period of time and training power values of the monitored device at a plurality of predetermined time points in the predetermined period of time, and a true value of whether an operating state of the monitored device is normal; a training input vector arrangement unit 220, configured to arrange training power values of the monitored device at a plurality of predetermined time points in the predetermined time period into training power time sequence input vectors according to a time dimension; a training up-sampling unit 230, configured to perform up-sampling based on linear interpolation on the training power timing input vector to obtain a training up-sampling power timing input vector; a training encoding unit 240, configured to pass the waveform diagram of the training vibration signal and the training up-sampling power timing input vector through the CLIP model including the image encoder and the sequence encoder to obtain a training multi-mode device state feature matrix; a training classification unit 250, configured to pass the training multi-mode device state feature matrix through a classifier to obtain a classification loss function value; and a training optimization iteration unit 260, configured to train the CLIP model including the image encoder and the sequence encoder and the classifier with the classification loss function value, where in each iteration of the training, fine-grained density prediction search optimization iteration of a weight space is performed on the training multi-mode device state feature vector obtained after the training multi-mode device state feature matrix is expanded.
In the technical solution of the present disclosure, when the waveform diagram of the training vibration signal and the training up-sampling power timing input vector pass through a CLIP model including an image encoder and a sequence encoder, each feature value of the waveform diagram of the training vibration signal expresses a local associated image semantic feature extracted via a two-dimensional convolution kernel of the image encoder, and each feature value of the training up-sampling power timing input vector passes through a training sequence feature vector obtained by the sequence encoder expresses a local time-sequence associated power feature extracted via a one-dimensional convolution kernel of the sequence encoder, so that, for the training multi-mode device state feature matrix obtained by associating the training image feature vector with the training sequence feature vector position by position, each feature value corresponds to an association value of a feature value of a corresponding position of the training image feature vector and the training sequence feature vector, thereby having a super-resolution expression characteristic based on the feature value, and considering that the training image feature vector and the training sequence feature vector have a super-resolution characteristic under the condition that the overall multi-mode device state feature matrix has an influence on the overall-resolution.
Therefore, when the training multi-modal device state feature matrix is trained by the classifier, in each iteration, the training multi-modal device state feature vector obtained after the training multi-modal device state feature matrix is expanded is, for example, recorded asAnd carrying out fine granularity density prediction search optimization of the weight space.
Accordingly, in a specific example, the training optimization iteration unit 260 is further configured to: in each round of iteration of training, carrying out fine granularity density prediction search optimization iteration of a weight space on the training multi-mode equipment state feature vector obtained after the training multi-mode equipment state feature matrix is unfolded by using the following optimization iteration formula so as to obtain an optimized training multi-mode equipment state feature vector; wherein, the optimization iterative formula is:wherein (1)>And->The weight matrix of the last iteration and the current iteration are respectively adopted, wherein, during the first iteration, different initialization strategies are adopted to set +.>And->(e.g.)>Set as a unitary matrix->Set as the diagonal matrix of the mean value of the feature vector to be classified),>is the training multi-modal device state feature vector, < >>And->Respectively represent feature vector +>And->Global mean of (2), and->Is a bias vector, e.g. initially set as a unit vector,/->Representing multiplication by location +.>Representing vector addition, ++>Representing matrix multiplication +.>Representing a minimum value +.>Is the optimized training multi-mode device state feature vector.
Here, the multi-modal device state feature vector is trained on the trainingSuper-resolution representation characteristics in multi-dimensional context, fine-grained density predictive search optimization of the weight space may be optimized by the training multi-modal device state feature vector ∈ ->Feed-forward serialization mapping of vector space of (2) while providing a corresponding fine-grained weight search strategy for dense prediction tasks within a weight search space, reducing the training multi-modal device state feature vector ∈within the weight search space>The overall sequence complexity (overall sequential complexity) of the representation of (c), thereby improving training efficiency.
In summary, the intelligent management system 100 of a product manufacturing line based on a digital twin technology according to an embodiment of the present disclosure is illustrated, which can determine whether the image quality of a medical image to be uploaded meets an uploading standard through a text recognition probability value, so as to ensure the quality and reliability of the uploaded medical data.
As described above, the digital twin technology-based product manufacturing line intelligent management system 100 according to the embodiment of the present disclosure may be implemented in various terminal devices, for example, a server or the like having a digital twin technology-based product manufacturing line intelligent management algorithm. In one example, the digital twinning technology based product manufacturing line intelligent management system 100 may be integrated into the terminal device as a software module and/or hardware module. For example, the digital twinning technology based product manufacturing line intelligent management system 100 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the intelligent management system 100 of the product manufacturing line based on the digital twin technology may also be one of numerous hardware modules of the terminal device.
Alternatively, in another example, the digital twinning technology based product manufacturing line intelligent management system 100 and the terminal device may also be separate devices, and the digital twinning technology based product manufacturing line intelligent management system 100 may be connected to the terminal device through a wired and/or wireless network and transmit interactive information in a agreed data format.
Fig. 6 shows a flowchart of a method of intelligent management of a product manufacturing line based on digital twinning technology in accordance with an embodiment of the present disclosure. Fig. 7 shows a schematic diagram of a system architecture of a product manufacturing line intelligent management method based on digital twinning technology according to an embodiment of the present disclosure. As shown in fig. 6 and 7, a digital twin technology-based product manufacturing line intelligent management method according to an embodiment of the present disclosure includes: s110, acquiring vibration signals of monitored equipment in a preset time period and power values of the monitored equipment at a plurality of preset time points in the preset time period, wherein the vibration signals are acquired by cameras arranged on a product manufacturing production line; s120, carrying out joint analysis on the vibration signals and the power values of the plurality of preset time points to obtain a multi-mode equipment state feature matrix; and S130, determining whether the working state of the monitored equipment is normal or not based on the multi-mode equipment state feature matrix.
In one possible implementation, performing joint analysis on the vibration signal and the power values at the plurality of predetermined time points to obtain a multi-mode device state feature matrix includes: performing data structuring and data enhancement processing on power values of a plurality of preset time points of the monitored equipment in the preset time period to obtain an up-sampling power time sequence input vector; and passing the waveform diagram of the vibration signal and the up-sampling power time sequence input vector through a CLIP model comprising an image encoder and a sequence encoder to obtain the multi-mode device state feature matrix.
In one possible implementation, the data structuring and data enhancing process is performed on power values of the monitored device at a plurality of predetermined time points in the predetermined time period to obtain an up-sampling power timing input vector, including: arranging power values of a plurality of preset time points of the monitored equipment in the preset time period into power time sequence input vectors according to a time dimension; and upsampling the power timing input vector based on linear interpolation to obtain the upsampled power timing input vector.
In one possible implementation manner, based on the multi-mode device state feature matrix, determining whether the working state of the monitored device is normal includes: the multi-mode equipment state feature matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the working state of the monitored equipment is normal or not; and displaying the classification result on a screen.
Here, it will be understood by those skilled in the art that the specific operations of the respective steps in the above-described intelligent management method for a product manufacturing line based on the digital twin technology have been described in detail in the above description of the intelligent management system for a product manufacturing line based on the digital twin technology with reference to fig. 1 to 5, and thus, repetitive descriptions thereof will be omitted.
Fig. 8 illustrates an application scenario diagram of a digital twinning technology-based product manufacturing line intelligent management system, according to an embodiment of the present disclosure. As shown in fig. 8, in this application scenario, first, a vibration signal (e.g., D1 illustrated in fig. 8) of a monitored device acquired by a camera disposed on a product manufacturing line for a predetermined period of time and power values (e.g., D2 illustrated in fig. 8) of the monitored device at a plurality of predetermined time points within the predetermined period of time are acquired, and then the vibration signal and the power values at the plurality of predetermined time points are input to a server (e.g., S illustrated in fig. 8) disposed with a digital twinning technology-based product manufacturing line intelligent management algorithm, wherein the server is capable of processing the vibration signal and the power values at the plurality of predetermined time points using the digital twinning technology-based product manufacturing line intelligent management algorithm to obtain a classification result indicating whether an operation state of the monitored device is normal.
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 disclosure. 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 foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or 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 (6)
1. Product manufacturing production line intelligent management system based on digital twin technique, characterized by comprising:
the device comprises a data acquisition module, a control module and a control module, wherein the data acquisition module is used for acquiring vibration signals of monitored equipment in a preset time period and power values of the monitored equipment at a plurality of preset time points in the preset time period, wherein the vibration signals are acquired by cameras arranged on a product manufacturing production line;
the joint analysis module is used for carrying out joint analysis on the vibration signals and the power values of the plurality of preset time points to obtain a multi-mode equipment state characteristic matrix; and the working state judging module is used for determining whether the working state of the monitored equipment is normal or not based on the multi-mode equipment state feature matrix.
2. The intelligent management system of a product manufacturing line based on digital twinning technology according to claim 1, wherein the joint analysis module comprises:
the structuring and enhancing processing unit is used for carrying out data structuring and data enhancing processing on power values of a plurality of preset time points of the monitored equipment in the preset time period so as to obtain an up-sampling power time sequence input vector; and the encoding unit is used for passing the waveform diagram of the vibration signal and the up-sampling power time sequence input vector through a CLIP model comprising an image encoder and a sequence encoder to obtain the multi-mode equipment state characteristic matrix.
3. The intelligent management system of a product manufacturing line based on digital twinning technology according to claim 2, characterized in that the structuring and reinforcing processing unit comprises:
an input vector arrangement subunit, configured to arrange power values of the monitored device at a plurality of predetermined time points in the predetermined time period into a power timing input vector according to a time dimension; and an up-sampling subunit, configured to perform up-sampling based on linear interpolation on the power timing input vector to obtain the up-sampled power timing input vector.
4. The intelligent management system of a product manufacturing line based on digital twin technology according to claim 3, wherein the working state judging module comprises:
the classifying unit is used for passing the state characteristic matrix of the multi-mode equipment through a classifier to obtain a classifying result, wherein the classifying result is used for indicating whether the working state of the monitored equipment is normal or not; and a display unit for displaying the classification result on a screen.
5. The intelligent management system of a product manufacturing line based on digital twinning technology according to claim 4, further comprising a training module for training the CLIP model including an image encoder and a sequence encoder and the classifier;
wherein, training module includes:
the device comprises a training data acquisition unit, a control unit and a control unit, wherein the training data acquisition unit is used for acquiring training data, the training data comprises training vibration signals of monitored equipment in a preset time period and training power values of the monitored equipment at a plurality of preset time points in the preset time period, and a true value of whether the working state of the monitored equipment is normal or not;
the training input vector arrangement unit is used for arranging training power values of the monitored equipment at a plurality of preset time points in the preset time period into training power time sequence input vectors according to a time dimension;
the training up-sampling unit is used for up-sampling the training power time sequence input vector based on linear interpolation to obtain a training up-sampling power time sequence input vector;
the training coding unit is used for enabling the waveform diagram of the training vibration signal and the training up-sampling power time sequence input vector to pass through the CLIP model comprising the image encoder and the sequence encoder so as to obtain a training multi-mode equipment state feature matrix;
the training classification unit is used for passing the training multi-mode equipment state feature matrix through a classifier to obtain a classification loss function value; and a training optimization iteration unit, configured to train the CLIP model including the image encoder and the sequence encoder and the classifier with the classification loss function value, where in each iteration of the training, fine granularity density prediction search optimization iteration of a weight space is performed on a training multi-mode device state feature vector obtained after the training multi-mode device state feature matrix is expanded.
6. The intelligent management system of a product manufacturing line based on digital twinning technology according to claim 5, wherein the training optimization iteration unit is further configured to:
in each round of iteration of training, carrying out fine granularity density prediction search optimization iteration of a weight space on the training multi-mode equipment state feature vector obtained after the training multi-mode equipment state feature matrix is unfolded by using the following optimization iteration formula so as to obtain an optimized training multi-mode equipment state feature vector;
wherein, the optimization iterative formula is:wherein (1)>And->The weight matrix of the last iteration and the current iteration are respectively adopted, wherein, during the first iteration, different initialization strategies are adopted to set +.>And->，/>Is the training multi-modal device state feature vector, < >>And->Respectively represent feature vector +>And->Global mean of (2), and->Is a bias vector, ++>Representing multiplication by location +.>Representing vector addition, ++>Representing matrix multiplication +.>Representing a minimum value +.>Is the optimized training multi-mode device state feature vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310998683.3A CN116703642A (en) | 2023-08-09 | 2023-08-09 | Intelligent management system of product manufacturing production line based on digital twin technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310998683.3A CN116703642A (en) | 2023-08-09 | 2023-08-09 | Intelligent management system of product manufacturing production line based on digital twin technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116703642A true CN116703642A (en) | 2023-09-05 |
Family
ID=87834335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310998683.3A Pending CN116703642A (en) | 2023-08-09 | 2023-08-09 | Intelligent management system of product manufacturing production line based on digital twin technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116703642A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117189071A (en) * | 2023-11-07 | 2023-12-08 | 克拉玛依市远山石油科技有限公司 | Automatic control method for core drilling rig operation |
| CN117202430A (en) * | 2023-09-20 | 2023-12-08 | 浙江炯达能源科技有限公司 | Energy-saving control method and system for intelligent lamp post |
| CN117375237A (en) * | 2023-10-20 | 2024-01-09 | 浙江日新电气有限公司 | Substation operation and maintenance method and system based on digital twin technology |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112629851A (en) * | 2020-12-11 | 2021-04-09 | 南方海上风电联合开发有限公司 | Offshore wind turbine gearbox fault diagnosis method based on data enhancement method and image recognition |
| CN115510962A (en) * | 2022-09-19 | 2022-12-23 | 浙江佳雪微特电机有限公司 | Water pump permanent magnet synchronous motor with fault self-monitoring function and method thereof |
| CN115546577A (en) * | 2022-09-19 | 2022-12-30 | 鼎富智能科技有限公司 | Data enhancement method and device for multi-modal data set |
| CN115586757A (en) * | 2022-11-09 | 2023-01-10 | 广东金柳信息科技有限公司 | Intelligent control system and method for mechanical equipment |
| CN115693918A (en) * | 2022-09-07 | 2023-02-03 | 浙江心友机电设备安装有限公司 | Comprehensive intelligent power utilization system and method for building |
| CN116111906A (en) * | 2022-11-17 | 2023-05-12 | 浙江精盾科技股份有限公司 | Special motor with hydraulic brake for turning and milling and control method thereof |
-
2023
- 2023-08-09 CN CN202310998683.3A patent/CN116703642A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112629851A (en) * | 2020-12-11 | 2021-04-09 | 南方海上风电联合开发有限公司 | Offshore wind turbine gearbox fault diagnosis method based on data enhancement method and image recognition |
| CN115693918A (en) * | 2022-09-07 | 2023-02-03 | 浙江心友机电设备安装有限公司 | Comprehensive intelligent power utilization system and method for building |
| CN115510962A (en) * | 2022-09-19 | 2022-12-23 | 浙江佳雪微特电机有限公司 | Water pump permanent magnet synchronous motor with fault self-monitoring function and method thereof |
| CN115546577A (en) * | 2022-09-19 | 2022-12-30 | 鼎富智能科技有限公司 | Data enhancement method and device for multi-modal data set |
| CN115586757A (en) * | 2022-11-09 | 2023-01-10 | 广东金柳信息科技有限公司 | Intelligent control system and method for mechanical equipment |
| CN116111906A (en) * | 2022-11-17 | 2023-05-12 | 浙江精盾科技股份有限公司 | Special motor with hydraulic brake for turning and milling and control method thereof |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117202430A (en) * | 2023-09-20 | 2023-12-08 | 浙江炯达能源科技有限公司 | Energy-saving control method and system for intelligent lamp post |
| CN117202430B (en) * | 2023-09-20 | 2024-03-19 | 浙江炯达能源科技有限公司 | Energy-saving control method and system for intelligent lamp post |
| CN117375237A (en) * | 2023-10-20 | 2024-01-09 | 浙江日新电气有限公司 | Substation operation and maintenance method and system based on digital twin technology |
| CN117375237B (en) * | 2023-10-20 | 2024-05-24 | 浙江日新电气有限公司 | Substation operation and maintenance method and system based on digital twin technology |
| CN117189071A (en) * | 2023-11-07 | 2023-12-08 | 克拉玛依市远山石油科技有限公司 | Automatic control method for core drilling rig operation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116703642A (en) | Intelligent management system of product manufacturing production line based on digital twin technology | |
| CN112381788B (en) | Part surface defect increment detection method based on double-branch matching network | |
| CN110795585B (en) | Zero sample image classification system and method based on generation countermeasure network | |
| CN112964469A (en) | Online fault diagnosis method for rolling bearing under variable load of transfer learning | |
| CN110188827A (en) | A kind of scene recognition method based on convolutional neural networks and recurrence autocoder model | |
| CN117034123B (en) | Fault monitoring system and method for fitness equipment | |
| CN117784710B (en) | Remote state monitoring system and method for numerical control machine tool | |
| CN117158904B (en) | Old people cognitive disorder detection system and method based on behavior analysis | |
| CN116929815A (en) | Equipment working state monitoring system and method based on Internet of things | |
| CN117689278B (en) | Construction quality intelligent management system and method | |
| CN117202430B (en) | Energy-saving control method and system for intelligent lamp post | |
| CN117123131B (en) | Petroleum aid production equipment and method thereof | |
| CN117676099B (en) | Security early warning method and system based on Internet of things | |
| CN117308077A (en) | Water supply control system for evaporator of nuclear power unit of reactor | |
| CN114492581A (en) | Method for classifying small sample pictures based on transfer learning and attention mechanism element learning application | |
| CN116630816B (en) | SAR target recognition method, device, equipment and medium based on prototype comparison learning | |
| CN117237095A (en) | Small micro financial industry risk early warning and monitoring platform based on big data analysis | |
| Li | Parallel two-class 3D-CNN classifiers for video classification | |
| CN117154256A (en) | Electrochemical repair method for lithium battery | |
| CN117333717A (en) | Security monitoring method and system based on network information technology | |
| CN117078007A (en) | Multi-scale wind control system integrating scale labels and method thereof | |
| CN116843345A (en) | Intelligent wind control system and method for trading clients based on artificial intelligence technology | |
| Ho et al. | Document classification in a non-stationary environment: A one-class svm approach | |
| Harshitha et al. | Hand Sign Classification Techniques Using Supervised And Unsupervised Learning Algorithms | |
| CN109002832B (en) | Image identification method based on hierarchical feature extraction |
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 |