CN114980723B - Fault prediction method and system for suction nozzle of cross-working-condition chip mounter - Google Patents
Fault prediction method and system for suction nozzle of cross-working-condition chip mounter Download PDFInfo
- Publication number
- CN114980723B CN114980723B CN202210571197.9A CN202210571197A CN114980723B CN 114980723 B CN114980723 B CN 114980723B CN 202210571197 A CN202210571197 A CN 202210571197A CN 114980723 B CN114980723 B CN 114980723B
- Authority
- CN
- China
- Prior art keywords
- data
- suction nozzle
- data set
- chip mounter
- prediction model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H05—ELECTRIC TECHNIQUES NOT OTHERWISE PROVIDED FOR
- H05K—PRINTED CIRCUITS; CASINGS OR CONSTRUCTIONAL DETAILS OF ELECTRIC APPARATUS; MANUFACTURE OF ASSEMBLAGES OF ELECTRICAL COMPONENTS
- H05K13/00—Apparatus or processes specially adapted for manufacturing or adjusting assemblages of electric components
- H05K13/08—Monitoring manufacture of assemblages
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/048—Activation functions
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Operations Research (AREA)
- Pure & Applied Mathematics (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Computational Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Manufacturing & Machinery (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Supply And Installment Of Electrical Components (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of chip mounters and discloses a fault prediction method and a fault prediction system for a chip mounter suction nozzle under a working condition II, wherein a data set B generated when the chip mounter suction nozzle runs to generate a fault is acquired, the data set B is input into a correction prediction model, and the residual service life of the chip mounter suction nozzle under the working condition II is predicted; the establishment process of the correction prediction model comprises the following steps: acquiring a data set A of a suction nozzle of the chip mounter in a first working condition; data enhancement is carried out on the data in the data set A through a TimeVAE network; carrying out Hilbert yellow transform preprocessing on the data in the data set A; inputting the preprocessed data set A into a prediction model, extracting high-dimensional characteristics, and predicting the residual service life of the suction nozzle of the chip mounter. Correcting the prediction model through RUL loss, and combining the corrected prediction model with transfer learning to obtain a corrected prediction model; and the MK-MMD similarity measurement method is adopted as a loss function of migration learning to correct the migration effect.
Description
Technical Field
The invention relates to the technical field of chip mounters, in particular to a fault prediction method and a fault prediction system for a suction nozzle of a cross-working-condition chip mounter.
Background
In the SMT production line, the chip mounter is the most important and complex equipment and is also the equipment most prone to failure, and at present, the chip mounter failure accounts for more than 60% of the whole SMT production line, and has a great influence on the normal production of the production line. In the chip mounter faults, the primary fault influencing factors are the suction nozzle components, and particularly under the condition of working condition change, how to solve the problem of predicting the cross-working condition faults of the suction nozzle of the chip mounter has a significant effect on the normal production of the SMT production line.
However, the main problems of the nozzle failure prediction are insufficient service life data, less abnormal data and poor cross-working condition prediction effect. In the past, the fault condition of the suction nozzle of the chip mounter is judged afterwards by the vacuum air pressure condition of the suction nozzle; the general deep learning method is generally suitable for fault prediction under a single working condition, and the cross-working condition application effect is poor. Therefore, there is a trend to solve these problems in combination with deep learning methods of data enhancement and transfer learning.
With the application of deep learning in the field of predictive maintenance, a large number of deep learning algorithms are proposed successively for solving the problem of fault prediction.
However, in an actual scene, abnormal data about the suction nozzle of the chip mounter is relatively less, the working condition is complex and changeable, the design of the prediction model aiming at the suction nozzle of the chip mounter under various specific working conditions is relatively labor-consuming, and once the working condition is changed, the model is easy to fail.
The invention is mainly used for solving the problems of insufficient service life data, less abnormal data and poor prediction effect of the cross-working condition in the fault prediction of the suction nozzle of the chip mounter.
Disclosure of Invention
In order to solve the technical problems, the invention provides a fault prediction method and a fault prediction system for a suction nozzle of a cross-working-condition chip mounter.
In order to solve the technical problems, the invention adopts the following technical scheme:
the fault prediction method for the suction nozzle of the chip mounter under the working condition II comprises the steps of obtaining a data set B generated when the suction nozzle of the chip mounter under the working condition II runs to a fault, inputting the data set B into a correction prediction model, and predicting to obtain the residual service life of the suction nozzle of the chip mounter under the working condition II;
the establishment process of the correction prediction model comprises the following steps:
step one: acquiring a data set A of the whole life cycle of a suction nozzle of the chip mounter in the first working condition;
step two: data enhancement is carried out on the data in the data set A through a TimeVAE network;
step three: carrying out Hilbert yellow transform preprocessing on the data in the data set A;
step four: inputting the preprocessed data set A into a prediction model, extracting high-dimensional characteristics, and predicting the residual usable life of a suction nozzle of the chip mounter;
step five: correcting the prediction model through the residual available life loss function, and combining the corrected prediction model with transfer learning to obtain a corrected prediction model; and the MK-MMD similarity measurement method is adopted as a loss function of migration learning to correct the migration effect.
A failure prediction system for a cross-condition chip mounter suction nozzle, comprising:
the data set acquisition module is used for acquiring a data set A of the whole life cycle of the suction nozzle of the chip mounter in the first working condition;
the data enhancement module is used for enhancing the data in the data set A through the TimeVAE network;
the preprocessing module is used for carrying out Hilbert yellow conversion preprocessing on the data in the data set A;
the service life prediction module inputs the preprocessed data set A into a prediction model, extracts high-dimensional characteristics and predicts the residual service life of the suction nozzle of the chip mounter;
the model correction module corrects the prediction model through the residual service life loss function, and combines the corrected prediction model with transfer learning to obtain a corrected prediction model; the MK-MMD similarity measurement method is adopted as a loss function of migration learning to correct migration effects;
and the corrected service life prediction module is used for acquiring a data set B generated when the suction nozzle of the chip mounter in the second working condition runs to a fault, inputting the data set B into the corrected prediction model, and predicting to obtain the residual service life of the suction nozzle of the chip mounter in the second working condition.
Compared with the prior art, the invention has the beneficial technical effects that:
(1) The invention combines the data enhancement, transfer learning and deep learning methods, and further provides a solving method for predicting the suction nozzle faults of the chip mounter under the working conditions on the basis of solving the problems of insufficient service life data and less abnormal data in the suction nozzle fault prediction of the chip mounter;
(2) The original service life data is expanded by a data enhancement method, so that the robustness of the prediction model is effectively improved;
(3) The soft threshold function adopted in the prediction model has good anti-noise performance, and is suitable for a strong noise environment in a real industrial scene;
(4) The attention mechanism GAU adopted in the prediction model can be used for grabbing useful features in the original life data well.
(5) As shown in FIG. 1, the prediction model of the invention adopts a time sequence convolution network TCN, and has better parallel processing capability, more flexible receptive field, more stable gradient, lower memory and good long-time memory performance compared with the prior cyclic neural network RNN.
(6) The migration learning method adopted by the invention can effectively improve the failure prediction effect of the suction nozzle of the chip mounter under the condition of the cross working condition, and can effectively cope with the complex change working condition of the industrial field.
(7) Through adopting MK-MMD as the loss function of transfer learning, a better kernel function representation can be selected, and the transfer effect is effectively improved.
Drawings
FIG. 1 is a flow chart of a fault prediction method of the present invention;
FIG. 2 is a schematic diagram of data enhancement using a TimeVAE network in accordance with the present invention;
FIG. 3 is a schematic representation of a predictive model of the present invention.
Detailed Description
A preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.
As shown in fig. 1, an embodiment of the present invention provides a fault prediction method for a suction nozzle of a cross-working-condition chip mounter:
(1) Firstly, collecting a full life cycle data set A of the suction nozzle of the chip mounter running to a fault under a working condition and a data set B of the suction nozzle of the chip mounter running to the fault under a working condition through a sensor.
(2) The data in data set a and data set B are then data enhanced using the TimeVAE network, improving the robustness of the predictive model, as shown in fig. 2.
Specifically, in the TimeVAE network of fig. 2, the input X is a three-dimensional array comprising batch size, time step, feature dimension, and then X is passed into an encoder structure comprising 3 one-dimensional convolutional layers, flattening layers, fully-connected layers, with the output of the encoder structure parameterizing the multi-gaussian function. Next, Z is sampled from the multi-element gaussian function by a re-parameterization technique and input into the decoder structure, i.e. after Z is input into the fully-connected layer, the data is reshaped into a 3-dimensional matrix, then enters into 3 one-dimensional transposed convolutional layers, finally passes through the fully-connected layer of time distribution, the dimensions of which are such that the final output shape is the same as the original signal.
The TimeVAE network employed herein can be effectively extended to generate more high quality lifetime data, given the limited actual available lifetime data.
(3) The data in the data-enhanced data set a and data set B are subjected to hilbert yellow (HHT) transform preprocessing.
(4) And carrying out feature extraction on the data in the preprocessed data set A through a prediction model, and then predicting the residual service life RUL of the suction nozzle of the chip mounter in the first working condition. As shown in fig. 3, the data after preprocessing is input to a timing residual module Temporal Residual Block, which is mainly composed of two identical parts, each of which includes a discard Dropout, an activation function ReLu, a batch regularization BN, an expansion cause and effect convolution Dilated causal Convolution. Deep features of the original life data are obtained by time series residual convolution and then transferred into a gating attention unit GAU, input is subjected to twice full connection processing (Dense), and one of the full connection processing (Dense) is subjected to attention weighting by a Self-attention mechanism (Self-attention), wherein the relu is used 2 The (-) activation function corresponds to (relu (-)) 2 That is, the activation function relu is squared again, and finally, the full connection process (Dense) is performed again. After the GAU is used for characteristic processing, effective noise reduction is carried out by utilizing two times of subtraction operation and maximization operation in a soft threshold function (Soft thresholding function), and redundant information is removed. And then grabbing useful features of the result obtained by the sequential convolution block through multiplication operation. FinallyThrough residual connection and adding operation, the training effect of the deep neural network is improved.
The soft threshold function Softthreshold function in the predictive model is mainly used for noise reduction; the GAU is a gated attention unit that combines a gated linear unit Gated Linear Unit (GLU) and an attention mechanism to enable efficient extraction of useful features.
(5) The training effect of the prediction model under the working condition is improved by designing a residual available life loss function RUL loss, and the correction prediction model is obtained by combining migration learning.
Specifically, the RUL loss herein comprises a root mean square error RMSE, a mean absolute error MAE, a predictive Score function Score, wherein
Here er t Refers to the error between the real RUL and the predicted RUL in time step t, A t Refers to the weighted error between the real RUL and the predicted RUL in time step t, n is the total number of time steps, m is the percentage of the early stage, where m=n/2,w 1 ＝0.35，w 2 ＝0.65。
(6) The correction prediction model is applied to a second working condition, and residual service life prediction is carried out on a suction nozzle of the chip mounter to obtain residual service life RUL of the chip mounter under the second working condition; the migration effect of migration learning is corrected by an MK-MMD similarity measurement method.
The data set A is data of each sensor obtained in the process from the current state operation to failure of the suction nozzle of the chip mounter in the first working condition; the data set B is the data of each sensor obtained in the process from the current state operation to the failure of the suction nozzle of the chip mounter in the second working condition.
A failure prediction system for a cross-condition chip mounter suction nozzle, comprising:
the data set acquisition module acquires a data set A of a suction nozzle of the chip mounter in a first working condition;
the data enhancement module is used for enhancing the data in the data set A through the TimeVAE network;
the preprocessing module is used for carrying out HHT preprocessing on the data in the data set A;
the service life prediction module inputs the preprocessed data set A into a prediction model, extracts high-dimensional characteristics and predicts the residual service life of the suction nozzle of the chip mounter;
the model correction module corrects the prediction model through RUL loss, and combines the corrected prediction model with transfer learning to obtain a corrected prediction model; the MK-MMD similarity measurement method is adopted as a loss function of migration learning to correct migration effects;
and the corrected service life prediction module is used for acquiring a data set B generated when the suction nozzle of the chip mounter in the second working condition runs to a fault, inputting the data set B into the corrected prediction model, and predicting to obtain the residual service life of the suction nozzle of the chip mounter in the second working condition.
The fault prediction system corresponds to the fault prediction method, is used for a preferable scheme of the fault prediction method, and is also suitable for the fault prediction system.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a single embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to specific embodiments, and that the embodiments may be combined appropriately to form other embodiments that will be understood by those skilled in the art.
Claims (2)
1. The fault prediction method for the suction nozzle of the chip mounter under the cross-working condition is used for predicting the residual service life of the suction nozzle of the chip mounter and is characterized in that: acquiring a data set B generated when the suction nozzle of the chip mounter runs to a fault in the second working condition, inputting the data set B into a correction prediction model, and predicting to obtain the residual service life of the suction nozzle of the chip mounter in the second working condition;
the establishment process of the correction prediction model comprises the following steps:
step one: acquiring a data set A of the whole life cycle of a suction nozzle of the chip mounter in the first working condition;
step two: data enhancement is carried out on the data in the data set A through a TimeVAE network;
step three: carrying out Hilbert yellow transform preprocessing on the data in the data set A;
step four: inputting the preprocessed data set A into a prediction model, extracting features, and predicting the residual usable life of a suction nozzle of the chip mounter;
step five: correcting the prediction model through the residual available life loss function, and combining the corrected prediction model with transfer learning to obtain a corrected prediction model; the MK-MMD similarity measurement method is adopted as a loss function of migration learning to correct migration effects;
the data set A is data of each sensor obtained in the process from the current state operation to failure of the suction nozzle of the chip mounter in the first working condition; the data set B is the data of each sensor obtained in the process from the current state operation to the failure of the suction nozzle of the chip mounter in the second working condition;
in the second step, when data enhancement is carried out on the data in the data set A through the TimeVAE network, the input data X is a three-dimensional array comprising batch size, time step length and feature dimension, the data X is transmitted into an encoder structure comprising three one-dimensional convolution layers, flattening layers and full connection layers, and the output of the encoder structure is utilized to carry out parameterization on a multi-element Gaussian function; then sampling data Z from a multi-element Gaussian function by a re-parameterization method, inputting the data Z into a decoder structure, after inputting the data Z into a full-connection layer, remodelling the data into a three-dimensional matrix, then entering three one-dimensional transposed convolution layers, and finally obtaining data with the same shape as the input data X through the time-distributed full-connection layer;
in the fourth step, the process of extracting features of the data in the preprocessed data set a and predicting the remaining service life of the suction nozzle of the chip mounter through the prediction model includes: inputting the preprocessed data into a time sequence residual error module; then the input is transmitted into a gating attention unit GAU, the input is subjected to twice full connection processing, attention weighting is carried out on one of the full connection processing through a self-attention mechanism, and the full connection processing is carried out again; after the data is subjected to characteristic processing by a gating attention unit GAU, noise reduction and redundant information removal are performed by utilizing twice subtraction operation and maximization operation in a soft threshold function; then, carrying out characteristic grabbing on a result obtained by the time sequence residual error module through multiplication operation; finally, carrying out residual connection and adding operation; the time sequence residual error module consists of two identical parts, wherein each part comprises a discarding method, an activating function, batch regularization and expansion causal convolution;
in step five, the remaining useful life loss function RULloss includes a root mean square error RMSE, an average absolute error MAE, and a predictive Score function Score, wherein
Here er t Refers to the error between the true remaining usable life and the predicted remaining usable life in time step t, A t Refers to the weighted error between the real RUL and the predicted RUL in time step t, n is the total number of time steps, m is the percentage of the early stage, where m=n/2,w 1 ＝0.35，w 2 ＝0.65。
2. A fault prediction system for striding operating mode chip mounter suction nozzle, characterized by comprising:
the data set acquisition module is used for acquiring a data set A of the whole life cycle of the suction nozzle of the chip mounter in the first working condition;
the data enhancement module is used for enhancing the data in the data set A through the TimeVAE network;
the preprocessing module is used for carrying out Hilbert yellow conversion preprocessing on the data in the data set A;
the service life prediction module inputs the preprocessed data set A into a prediction model, performs feature extraction, and predicts the residual usable life of the suction nozzle of the chip mounter;
the model correction module corrects the prediction model through the residual service life loss function, and combines the corrected prediction model with transfer learning to obtain a corrected prediction model; the MK-MMD similarity measurement method is adopted as a loss function of migration learning to correct migration effects;
the corrected service life prediction module is used for obtaining a data set B generated when the suction nozzle of the chip mounter in the second working condition runs to a fault, inputting the data set B into the corrected prediction model, and predicting to obtain the residual service life of the suction nozzle of the chip mounter in the second working condition;
the data set A is data of each sensor obtained in the process from the current state operation to failure of the suction nozzle of the chip mounter in the first working condition; the data set B is the data of each sensor obtained in the process from the current state operation to the failure of the suction nozzle of the chip mounter in the second working condition;
when data enhancement is carried out on data in a data set A through a TimeVAE network, the input data X is a three-dimensional array comprising batch size, time step and feature dimension, the data X is transmitted into an encoder structure comprising three one-dimensional convolution layers, a flattening layer and a full connection layer, and the output of the encoder structure is utilized to carry out parameterization on a multi-element Gaussian function; then sampling data Z from a multi-element Gaussian function by a re-parameterization method, inputting the data Z into a decoder structure, after inputting the data Z into a full-connection layer, remodelling the data into a three-dimensional matrix, then entering three one-dimensional transposed convolution layers, and finally obtaining data with the same shape as the input data X through the time-distributed full-connection layer;
the process of extracting the characteristics of the data in the preprocessed data set A through the prediction model and predicting the residual service life of the suction nozzle of the chip mounter comprises the following steps: inputting the preprocessed data into a time sequence residual error module; then the input is transmitted into a gating attention unit GAU, the input is subjected to twice full connection processing, attention weighting is carried out on one of the full connection processing through a self-attention mechanism, and the full connection processing is carried out again; after the data is subjected to characteristic processing by a gating attention unit GAU, noise reduction and redundant information removal are performed by utilizing twice subtraction operation and maximization operation in a soft threshold function; then, carrying out characteristic grabbing on a result obtained by the time sequence residual error module through multiplication operation; finally, carrying out residual connection and adding operation; the time sequence residual error module consists of two identical parts, wherein each part comprises a discarding method, an activating function, batch regularization and expansion causal convolution;
the residual useful life loss function RUL loss comprises a root mean square error RMSE, a mean absolute error MAE, a predictive Score function Score, wherein
Here er t Refers to the error between the true remaining usable life and the predicted remaining usable life in time step t, A t Refers to the weighted error between the real RUL and the predicted RUL in time step t, n is the total number of time steps, m is the percentage of the early stage, where m=n/2,w 1 ＝0.35，w 2 ＝0.65。/>
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210571197.9A CN114980723B (en) | 2022-05-24 | 2022-05-24 | Fault prediction method and system for suction nozzle of cross-working-condition chip mounter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210571197.9A CN114980723B (en) | 2022-05-24 | 2022-05-24 | Fault prediction method and system for suction nozzle of cross-working-condition chip mounter |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114980723A CN114980723A (en) | 2022-08-30 |
| CN114980723B true CN114980723B (en) | 2023-04-28 |
Family
ID=82955904
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210571197.9A Active CN114980723B (en) | 2022-05-24 | 2022-05-24 | Fault prediction method and system for suction nozzle of cross-working-condition chip mounter |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114980723B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116975604B (en) * | 2023-09-22 | 2023-12-08 | 张家港市得道电子有限公司 | Fault prediction method and system for chip mounter driving system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112381213A (en) * | 2020-12-01 | 2021-02-19 | 重庆邮电大学 | Industrial equipment residual life prediction method based on bidirectional long-term and short-term memory network |
| CN113656915B (en) * | 2021-08-19 | 2023-08-25 | 燕山大学 | Bearing residual life prediction method based on deep attention network |
| CN114118225A (en) * | 2021-11-02 | 2022-03-01 | 苏州热工研究院有限公司 | Method, system, electronic device and storage medium for predicting remaining life of generator |
-
2022
- 2022-05-24 CN CN202210571197.9A patent/CN114980723B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114980723A (en) | 2022-08-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112149316B (en) | Aero-engine residual life prediction method based on improved CNN model | |
| CN109035149B (en) | License plate image motion blur removing method based on deep learning | |
| CN110109015B (en) | Asynchronous motor fault monitoring and diagnosing method based on deep learning | |
| CN112200244B (en) | Intelligent detection method for anomaly of aerospace engine based on hierarchical countermeasure training | |
| CN111709292B (en) | Compressor vibration fault detection method based on recursion diagram and deep convolution network | |
| CN110726898B (en) | Power distribution network fault type identification method | |
| CN109389171B (en) | Medical image classification method based on multi-granularity convolution noise reduction automatic encoder technology | |
| CN112990553B (en) | Wind power ultra-short-term power prediction method using self-attention mechanism and bilinear fusion | |
| CN115017826B (en) | Method for predicting residual service life of equipment | |
| CN111242377A (en) | Short-term wind speed prediction method integrating deep learning and data denoising | |
| CN113449473B (en) | Wind power bearing degradation trend prediction method, system and device based on DFCAE-BiLSTM | |
| CN114332008B (en) | Unsupervised defect detection and positioning method based on multi-level feature reconstruction | |
| CN114980723B (en) | Fault prediction method and system for suction nozzle of cross-working-condition chip mounter | |
| CN113947114B (en) | Gas turbine rotor fault diagnosis method based on Transformer neural network and axis track | |
| CN112798956A (en) | Wind turbine generator fault diagnosis method based on multi-resolution time sequence cyclic neural network | |
| CN111815526B (en) | Rain image rainstrip removing method and system based on image filtering and CNN | |
| CN113033276A (en) | Behavior recognition method based on conversion module | |
| CN113343760A (en) | Human behavior recognition method based on multi-scale characteristic neural network | |
| CN115452376A (en) | Bearing fault diagnosis method based on improved lightweight deep convolution neural network | |
| CN115204272A (en) | Industrial system fault diagnosis method and equipment based on multi-sampling rate data | |
| CN114818826A (en) | Fault diagnosis method based on lightweight Vision Transformer module | |
| CN111275751B (en) | Unsupervised absolute scale calculation method and system | |
| CN116881639A (en) | Electricity larceny data synthesis method based on generation countermeasure network | |
| CN114048762B (en) | Double-attention-guided rotary machine health assessment method | |
| CN115375729A (en) | Single-target short-range tracking method and device based on deep pulse neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |