CN112528548A - Self-adaptive depth coupling convolution self-coding multi-mode data fusion method - Google Patents
Self-adaptive depth coupling convolution self-coding multi-mode data fusion method Download PDFInfo
- Publication number
- CN112528548A CN112528548A CN202011363611.4A CN202011363611A CN112528548A CN 112528548 A CN112528548 A CN 112528548A CN 202011363611 A CN202011363611 A CN 202011363611A CN 112528548 A CN112528548 A CN 112528548A
- Authority
- CN
- China
- Prior art keywords
- data
- self
- fusion
- coding
- coupling
- 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
Images
Classifications
-
- 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
- G06F30/25—Design optimisation, verification or simulation using particle-based methods
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The invention provides a self-adaptive depth coupling convolution self-coding multi-mode data fusion method, which comprises the following steps: s1, establishing a multi-modal data fusion detection framework; s2, collecting a data set, preprocessing the data S3, and fusing the data through a depth coupling convolution self-coding multi-modal data fusion model; s4, inputting the fused data into a Softmax network module for health detection of the packaging equipment; s5, carrying out self-adaptive optimization on the coupling parameters and the network through a wolf optimization algorithm (GWO). The invention designs a depth coupling convolution self-coding fusion model, and independent and combined features of synchronously extracted multi-modal data are fused and then used for health detection of equipment; meanwhile, the detection effect is used as an index, and adaptive adjustment is carried out on parameters of the coupling loss function and network parameters through an GWO algorithm, so that the problem of difficult data fusion caused by overlarge differences of different modal data characteristics is solved to a certain extent.
Description
Technical Field
The invention relates to a multi-modal data fusion method, in particular to a self-adaptive depth coupling convolution self-coding multi-modal data fusion method.
Background
With the development of intelligent technology, revolutionary changes are brought to equipment manufacturing industry, and intelligent equipment becomes the core of the leading edge of high-end equipment and the manufacturing industry and is an important mark for measuring the level of national technological innovation and high-end manufacturing industry. At present, the state has exported intelligent engineering of packaging equipment, organizes and implements important special items such as key technology of high-end packaging equipment and integrated technology, and vigorously promotes the intellectualization of the packaging equipment. The intelligent packaging equipment is electromechanical equipment with deep integration of information technology and artificial intelligence technology, and if a very small fault is not processed in time in the operation process, the coordinated operation state of the motor can be damaged, the shutdown is caused, even the equipment is damaged, and direct economic loss is caused for enterprises. Therefore, how to effectively detect the health of the equipment is the key to ensure the normal operation of the packaging equipment. Data monitoring is the basis for realizing fault diagnosis, and the packaging power equipment moving at high speed is exposed to complex operating environments with strong noise, large disturbance and the like, and the equipment state is not detected. However, due to the power of the packaging equipment, a large number of sensors are provided to monitor it. Therefore, how to apply the multi-sensor data fusion technology to the intelligent packaging equipment health detection is an important idea.
In the existing research, various multi-sensor fusion methods, such as the traditional methods of kalman filtering, adaptive weighting, neural network, etc., and deep learning techniques, have been proposed. However, the above methods have problems that multi-modal sensing data fusion, data correlation and excessive difference between data are not considered.
Disclosure of Invention
The invention aims at the problems that the correlation among data and the large difference of data characteristics are not fully considered in the existing multi-modal data fusion method. The method for self-adaptive depth coupling convolution self-coding multi-modal data fusion solves the problems that multi-modal sensing data fusion is difficult and data feature difference is large to a certain extent.
In order to achieve the purpose, the invention adopts the following technical scheme:
s1, establishing a multi-modal data fusion detection framework;
s2, performing data acquisition and preprocessing by the data acquisition and preprocessing module in the multi-modal data fusion detection framework in S1, and preprocessing the acquired data;
s3, designing a depth coupling convolution self-coding multi-mode data fusion model, and fusing the preprocessed data in S2 through the depth coupling convolution self-coding multi-mode data fusion model;
s4, inputting the fused data in the S3 into a Softmax network module for health detection of packaging equipment, and inputting a detection result;
s5, continuously optimizing key parameters in the multi-modal data fusion detection framework through a wolf optimization algorithm (GWO), so that the multi-modal data fusion detection framework achieves the optimal effect.
Further, in step S1, the multi-modal data fusion detection framework includes five modules, namely a data collection module, a preprocessing module, a deep-coupled convolution self-coding multi-modal data fusion module, a Softmax network classification evaluation module, and a GWO parameter optimization module.
Furthermore, the data acquisition module acquires current and vibration signals as a training data set and a test data set through the current sensor and the acceleration sensor.
Further, the data preprocessing in step S2 includes a data normalization process, which normalizes the original vibration and current data to eliminate the dimensional influence.
Furthermore, after the normalized vibration and current data of the packaging equipment are segmented and intercepted, the 1-D data are recombined into a 2-D grid matrix form in a segmentation mode.
Further, the fusion of the multi-modal data using the depth-coupled convolution self-coding fusion model in step S3 includes the following steps: (1) firstly, inputting preprocessed data into a deep coupling convolution self-encoder for training; (2) reserving an encoder part of the trained depth coupling convolution self-encoder, and splicing the output of the encoding part of the depth coupling convolution self-encoder; (3) inputting splicing data into the two-layer multi-channel complete convolution layer for interpretation and preliminary fusion; (4) and inputting the preliminarily fused data into two fully-connected layers for deep fusion and compression.
Further, in step S4, the fused data is input to the Softmax network to perform classified evaluation on the health status of the device.
Further, in step S5, the optimal coupling weight and the network parameter are searched for by GWO algorithm to obtain the optimal model with the goal of maximizing the classification accuracy.
The invention has the beneficial effects that: a multi-mode data fusion detection framework is established, a coupling loss function is designed to train a deep coupling convolution self-coding fusion network, so that the correlation among data can be considered when the data fusion is carried out, and fusion data with more comprehensive equipment information can be obtained; and meanwhile, optimizing the weight parameters of the coupling loss function through an GWO algorithm to obtain the weight parameters capable of balancing the problem of large characteristic difference among data, so as to obtain an optimal fusion model. Finally, the problems that multi-modal data fusion is difficult and the difference of data characteristics is large are solved to a certain extent.
Drawings
FIG. 1 is a block diagram of an overall process flow of a self-adaptive depth-coupled convolution self-coding multi-modal data fusion method;
FIG. 2 is a detailed block diagram of a depth-coupled convolution autoencoder fusion method;
FIG. 3 is a graph of the training effect of experiment I;
FIG. 4 is a graph of the training effect of experiment II;
FIG. 5 is a comparison graph of the effect of the adaptive optimization method;
FIG. 6 is a confusion matrix visualization of the health classifications of experiments I and II;
FIG. 7 is a graph comparing the effect of the method before and after fusion.
Detailed Description
The present invention will be further described with reference to the following embodiments.
A self-adaptive depth coupling convolution self-coding multi-mode data fusion method comprises the following steps:
and S1, establishing a multi-modal data fusion detection framework.
As shown in FIG. 1, the multi-modal data fusion detection framework comprises a vibration and current data collection and preprocessing module, a deep coupling convolution self-coding fusion network module and an GWO optimization parameter module. A training data set and a testing data set are obtained through an acceleration sensor and a current sensor in the intelligent packaging equipment data acquisition module.
And S2, acquiring data and preprocessing the acquired data.
The data preprocessing comprises data normalization and a data dimension-increasing process for converting 1-D data into 2-D grid matrixes. The specific flow is that firstly, input data is normalized to eliminate the influence caused by dimension difference among different data; and then carrying out data rearrangement on the normalized data from 1-D to 2-D to prepare for convolution input.
And S3, fusing the data through a depth coupling convolution self-coding multi-mode data fusion model.
The fusion step comprises the following procedures: (1) firstly, inputting preprocessed data into a deep coupling convolution self-encoder for training; (2) reserving an encoder part of the trained depth coupling convolution self-encoder, and splicing the output of the encoding part of the depth coupling convolution self-encoder; (3) inputting splicing data into the two-layer multi-channel complete convolution layer for interpretation and preliminary fusion; (4) and inputting the preliminarily fused data into two fully-connected layers for deep fusion and compression.
The deep coupling convolution self-encoder is formed by two same convolution self-encoders as shown in fig. 2, and is specifically set as follows:
deep coupling convolution self-coding network structure parameter
Wherein, Conv _1 and Conv _2 and pooling layers pool _1 and pool _2 constitute an encoding part of the convolutional auto-encoder, and Conv _3 and Conv _4 and F.interpolate operation are used to constitute a decoding part, wherein F.interpolate is used to replace the upsampling operation to realize the decoding function of the encoder.
The correlation between data is defined by a similarity measure, defined as follows:
S(xv,xc；θv,θc)＝||zv(xv,θv)-zc(xc,θc)||2
in the formula: z is a radical ofv、zcRepresenting the reconstructed output of vibration and current, respectively, thetavAnd thetacParameters representing two models, xvAnd xcAre input representations of vibration and current, respectively.
The coupling loss is defined as follows:
Lcoupling(xv,xc；θx,θc)＝εLv(xv；θv)+ηLc(xc；θc)+λS(xv,xc；θv,θc)
Lv(xv；θv)＝||xv-zv||2
Lc(xc；θc)＝||xc-zc||2
ε,η,λ>0
in the formula: l isvAnd LcRespectively corresponding to the reconstruction loss of the convolution self-encoder by the vibration data and the current data; ε, η and λ represent the sum of the reconstruction lossesThe parameters of the coupling model are controlled between the similarity loss functions. λ is a control parameter for similarity measurement between two types of data.
During training, parameters are effectively updated through an Adam optimization algorithm by using a back propagation algorithm as a standard network, so that the minimization of a coupling loss function is realized, wherein the gradient of the model loss function is calculated as follows:
after the deep coupling convolution self-encoder is trained, the outputs with independent characteristics and related characteristics output by the encoding part of the deep coupling convolution self-encoder are spliced and then input into a fusion compression network and output, and the specific structural parameters are as follows:
feature fusion compression network architecture parameters
The part of input data is formed by splicing the output data of the coding layers of the two convolution self-encoders, so that the characteristic information quantity in the data classes is too concentrated and the distance between the data classes is too absolute. In order to enable the network to better interpret the coded data characteristics and fuse the two types of data, the fusion part is designed into two multi-channel complete convolution layers to simultaneously interpret and fuse the data characteristics, so as to achieve a better fusion effect. Finally, in order to use the fusion data for performance verification and visualization, two full-connection layers are set as compression parts of the fusion network.
And S4, inputting the fused data into a Softmax network module for health detection of the packaging equipment.
In the step, the fused data is input into a Softmax module, the device state is classified and detected, and meanwhile the classification precision is used as a fusion effect evaluation standard. Assuming that there are K tags, the definition of the Softmax module is as follows:
in the formula, theta(1),θ(2),…θ(K)Is a parameter of the model, OjIs the evaluation result of DCCAE fusion model.
S5, carrying out self-adaptive optimization on the coupling parameters and the network through a wolf optimization algorithm (GWO) to obtain an optimal model.
In this step, the classification accuracy as the fusion effect evaluation parameter is used as an object, and with the classification accuracy maximization target, the optimal coupling weight and the network parameter are searched through the GWO algorithm to obtain the optimal model.
Simulation of experiment
In order to verify the detection precision and the detection effect of the model, the program experiment is realized by python codes, and the equipment of the simulation experiment is as follows: (1) a Processor (AMD Ryzen 52600X Six-Core Processor,3.60 GHz); (2) operating a memory (16G); (3) display card (NVIDIA GeForce GTX1660, 6G); (4) code operating environment (Pytorch 1.2.0, Python 3.7.9).
In order to verify the effectiveness of the model, motor bearing test data including outer ring single-point fatigue pitting damage (A), outer ring plastic deformation indentation damage (B), inner ring single-point fatigue pitting damage (C) and normal bearing data (D) are used for model test. A, B, D was set as the experiment I data and A, C, D was set as the experiment II data for the generalization ability of the test model to different types of lesion detection. Where each set of data contains 160000 discrete points. In addition, for effective training, the test network divides each set of data into a training set and a test set in a 3:1 ratio.
It should be noted that when the data preprocessing portion needs to normalize the data, the data needs to be mapped to the range of [ -1,1] in order to better conform to the characteristics of the vibration and current data. When training the depth-coupled auto-encoder, the learning rate is set to 0.001, the batch _ size is set to 100, the optimizer is set to Adam and the network is trained over 500 Epoch periods. The combination with the Softmax module is needed when the fusion compression layer is trained, and the network is trained according to the parameter settings with the learning rate of 0.0001 and the Epoch of 170, wherein the training results are shown in fig. 3 and fig. 4.
As can be seen from fig. 3 and 4, the deep coupling convolution self-encoder fusion network part in the invention can effectively fuse data and normally train the network. In addition, in order to solve the problem of large characteristic difference between data, the GWO algorithm is used for optimizing and adjusting the network with the aim of maximizing the test precision. In addition, in order to prove the superiority of GWO, an optimization algorithm based on Particle Swarm Optimization (PSO) was designed to perform the comparison as shown in fig. 5.
The complete network is used for testing, a confusion matrix is obtained as shown in fig. 6, and it can be seen that the adaptive deep coupling convolution self-coding multi-modal data fusion method provided by the invention can effectively perform classification detection on the health state of equipment.
Finally, as shown in fig. 7, the equipment health detection method using a single vibration and current model is compared with the method of the present invention, which proves that the multi-modal fusion model is more advantageous to the health detection effect of the packaging equipment compared with the single-modal model.
The above-mentioned embodiments are merely preferred embodiments for fully illustrating the present invention, and the scope of the present invention is not limited thereto. The equivalent substitution or change made by the technical personnel in the technical field on the basis of the invention is all within the protection scope of the invention.
Claims (8)
1. A self-adaptive depth coupling convolution self-coding multi-mode data fusion method comprises the following steps:
s1, establishing a multi-modal data fusion detection framework;
s2, data acquisition and preprocessing are carried out by the data acquisition and preprocessing module in the multi-modal data fusion detection framework in the S1, and the acquired data are preprocessed;
s3, fusing the preprocessed data in the S2 through a depth coupling convolution self-coding multi-mode data fusion module in the multi-mode data fusion detection framework;
s4, inputting the fusion data in S3 into a Softmax network classification evaluation module in the multi-modal data fusion detection framework for health detection of packaging equipment, and inputting a detection result;
s5, continuously optimizing key parameters in the multi-mode data fusion detection framework through a wolf optimization algorithm (GWO), and enabling the health state detection model of the packaging equipment to achieve the optimal effect.
2. The adaptive depth-coupled convolution self-coding multi-modal data fusion method of claim 1, wherein in step S1, the multi-modal data fusion detection framework includes five modules, namely a data acquisition module, a preprocessing module, a depth-coupled convolution self-coding fusion module, a Softmax network classification evaluation module, and a GWO parameter optimization module.
3. The adaptive deep-coupling convolution self-coding multi-modal data fusion method according to claim 2, characterized in that the data acquisition module acquires current and vibration signals as a training data set and a test data set through a current sensor and an acceleration sensor.
4. The adaptive deep-coupled convolution self-coding multi-mode data fusion method as claimed in claim 3, wherein the data preprocessing includes a data normalization procedure, in step S2, the raw vibration and current data are normalized to eliminate the dimensional effect.
5. The adaptive deep-coupling convolution self-coding multi-mode data fusion method according to claim 4, characterized in that after segmentation and interception are performed on normalized vibration and current data of the packaging equipment, 1-D data are recombined into a 2-D grid matrix form in a segmentation manner.
6. The method of claim 1, wherein the fusing the multi-modal data with the depth-coupled convolutional self-coding fusion model in step S3 comprises the following steps: (1) firstly, inputting preprocessed data into a depth coupling self-encoder for training; (2) reserving an encoder part of the trained depth coupling self-encoder, and splicing the output of the encoding part of the depth coupling self-encoder; (3) inputting splicing data into the two-layer multi-channel complete convolution layer for interpretation and preliminary fusion; (4) and inputting the preliminarily fused data into two fully-connected layers for deep fusion and compression.
7. The adaptive deep-coupled convolutional self-coding multimodal data fusion method as claimed in claim 1, wherein in step S4, the fused data is inputted to Softmax network for classification evaluation of device health status.
8. The adaptive deep-coupled convolutional self-coding multi-modal data fusion method as claimed in claim 1, wherein in step S5, with the goal of maximizing classification accuracy, the optimal coupling weights and network parameters are searched by GWO algorithm to obtain the optimal model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011363611.4A CN112528548A (en) | 2020-11-27 | 2020-11-27 | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011363611.4A CN112528548A (en) | 2020-11-27 | 2020-11-27 | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112528548A true CN112528548A (en) | 2021-03-19 |
Family
ID=74994710
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011363611.4A Pending CN112528548A (en) | 2020-11-27 | 2020-11-27 | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112528548A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569928A (en) * | 2021-07-13 | 2021-10-29 | 湖南工业大学 | Train running state detection data missing processing model and reconstruction method |
| CN116303786A (en) * | 2023-03-18 | 2023-06-23 | 上海圈讯科技股份有限公司 | Block chain financial big data management system based on multidimensional data fusion algorithm |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170371329A1 (en) * | 2014-12-19 | 2017-12-28 | United Technologies Corporation | Multi-modal sensor data fusion for perception systems |
| CN110108456A (en) * | 2019-04-16 | 2019-08-09 | 东南大学 | A kind of rotating machinery health evaluating method of depth convolutional neural networks |
| CN110163131A (en) * | 2019-05-09 | 2019-08-23 | 南京邮电大学 | Mix the human action classification method of convolutional neural networks and the optimization of microhabitat grey wolf |
| CN110455537A (en) * | 2019-09-18 | 2019-11-15 | 合肥工业大学 | A kind of Method for Bearing Fault Diagnosis and system |
| CN110595775A (en) * | 2019-09-19 | 2019-12-20 | 电子科技大学 | Rolling bearing fault diagnosis method based on multi-branch multi-scale convolutional neural network |
| CN111024433A (en) * | 2019-12-30 | 2020-04-17 | 辽宁大学 | Industrial equipment health state detection method for optimizing support vector machine by improving wolf algorithm |
| CN111046945A (en) * | 2019-12-10 | 2020-04-21 | 北京化工大学 | Fault type and damage degree diagnosis method based on combined convolutional neural network |
| US20200294257A1 (en) * | 2019-03-16 | 2020-09-17 | Nvidia Corporation | Leveraging multidimensional sensor data for computationally efficient object detection for autonomous machine applications |
| CN111721535A (en) * | 2020-06-23 | 2020-09-29 | 中国人民解放军战略支援部队航天工程大学 | Bearing fault detection method based on convolution multi-head self-attention mechanism |
| CN111914883A (en) * | 2020-06-18 | 2020-11-10 | 西安交通大学 | Spindle bearing state evaluation method and device based on deep fusion network |
| CN111967486A (en) * | 2020-06-02 | 2020-11-20 | 安徽三禾一信息科技有限公司 | Complex equipment fault diagnosis method based on multi-sensor fusion |
-
2020
- 2020-11-27 CN CN202011363611.4A patent/CN112528548A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170371329A1 (en) * | 2014-12-19 | 2017-12-28 | United Technologies Corporation | Multi-modal sensor data fusion for perception systems |
| US20200294257A1 (en) * | 2019-03-16 | 2020-09-17 | Nvidia Corporation | Leveraging multidimensional sensor data for computationally efficient object detection for autonomous machine applications |
| CN110108456A (en) * | 2019-04-16 | 2019-08-09 | 东南大学 | A kind of rotating machinery health evaluating method of depth convolutional neural networks |
| CN110163131A (en) * | 2019-05-09 | 2019-08-23 | 南京邮电大学 | Mix the human action classification method of convolutional neural networks and the optimization of microhabitat grey wolf |
| CN110455537A (en) * | 2019-09-18 | 2019-11-15 | 合肥工业大学 | A kind of Method for Bearing Fault Diagnosis and system |
| CN110595775A (en) * | 2019-09-19 | 2019-12-20 | 电子科技大学 | Rolling bearing fault diagnosis method based on multi-branch multi-scale convolutional neural network |
| CN111046945A (en) * | 2019-12-10 | 2020-04-21 | 北京化工大学 | Fault type and damage degree diagnosis method based on combined convolutional neural network |
| CN111024433A (en) * | 2019-12-30 | 2020-04-17 | 辽宁大学 | Industrial equipment health state detection method for optimizing support vector machine by improving wolf algorithm |
| CN111967486A (en) * | 2020-06-02 | 2020-11-20 | 安徽三禾一信息科技有限公司 | Complex equipment fault diagnosis method based on multi-sensor fusion |
| CN111914883A (en) * | 2020-06-18 | 2020-11-10 | 西安交通大学 | Spindle bearing state evaluation method and device based on deep fusion network |
| CN111721535A (en) * | 2020-06-23 | 2020-09-29 | 中国人民解放军战略支援部队航天工程大学 | Bearing fault detection method based on convolution multi-head self-attention mechanism |
Non-Patent Citations (2)
| Title |
|---|
| 孔俊;: "基于双层特征融合的生物识别", 北华大学学报(自然科学版), no. 01, pages 115 - 122 * |
| 敬明?;: "基于深度神经网络的多模态特征自适应聚类方法", 计算机应用与软件, no. 10, pages 268 - 275 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113569928A (en) * | 2021-07-13 | 2021-10-29 | 湖南工业大学 | Train running state detection data missing processing model and reconstruction method |
| CN113569928B (en) * | 2021-07-13 | 2024-01-30 | 湖南工业大学 | Train running state detection data missing processing model and reconstruction method |
| CN116303786A (en) * | 2023-03-18 | 2023-06-23 | 上海圈讯科技股份有限公司 | Block chain financial big data management system based on multidimensional data fusion algorithm |
| CN116303786B (en) * | 2023-03-18 | 2023-10-27 | 上海圈讯科技股份有限公司 | Block chain financial big data management system based on multidimensional data fusion algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110322423B (en) | Multi-modal image target detection method based on image fusion | |
| CN110097528B (en) | Image fusion method based on joint convolution self-coding network | |
| CN111914883B (en) | Spindle bearing state evaluation method and device based on deep fusion network | |
| CN108319962A (en) | A kind of Tool Wear Monitoring method based on convolutional neural networks | |
| CN111967486A (en) | Complex equipment fault diagnosis method based on multi-sensor fusion | |
| CN109389171B (en) | Medical image classification method based on multi-granularity convolution noise reduction automatic encoder technology | |
| Zhao et al. | Fault Diagnosis of Motor in Frequency Domain Signal by Stacked De-noising Auto-encoder. | |
| CN112528548A (en) | Self-adaptive depth coupling convolution self-coding multi-mode data fusion method | |
| CN112926641A (en) | Three-stage feature fusion rotating machine fault diagnosis method based on multi-modal data | |
| CN114283120B (en) | Domain-adaptive-based end-to-end multisource heterogeneous remote sensing image change detection method | |
| CN113807444A (en) | Fault diagnosis method based on constraint-confrontation convolutional self-coding memory fusion network | |
| CN110083125A (en) | A kind of machine tool thermal error modeling method based on deep learning | |
| CN114494222A (en) | Vision transducer-based rolling bearing fault intelligent identification method | |
| CN116894180B (en) | Product manufacturing quality prediction method based on different composition attention network | |
| CN116502069B (en) | Haptic time sequence signal identification method based on deep learning | |
| CN116383757B (en) | Bearing fault diagnosis method based on multi-scale feature fusion and migration learning | |
| CN116776209A (en) | Method, system, equipment and medium for identifying operation state of gateway metering device | |
| Tian et al. | Multi-scale deep coupling convolutional neural network with heterogeneous sensor data for intelligent fault diagnosis | |
| CN113973403B (en) | Temperature-induced strain field redistribution intelligent sensing method based on structure discrete measurement point topology | |
| CN115824519A (en) | Valve leakage fault comprehensive diagnosis method based on multi-sensor information fusion | |
| CN114926702A (en) | Small sample image classification method based on depth attention measurement | |
| CN114528869A (en) | Fan blade damage identification method | |
| Balaji et al. | Plant Infirmity Detection Using Vgg-16 Convolutional Neural Network | |
| CN113935413A (en) | Distribution network wave recording file waveform identification method based on convolutional neural network | |
| CN112598115A (en) | Deep neural network hierarchical analysis method based on non-local neighbor relation learning |
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 |