WO2022077901A1 - 一种面向小样本数据集的轴承故障模式诊断方法及系统 - Google Patents
一种面向小样本数据集的轴承故障模式诊断方法及系统 Download PDFInfo
- Publication number
- WO2022077901A1 WO2022077901A1 PCT/CN2021/093451 CN2021093451W WO2022077901A1 WO 2022077901 A1 WO2022077901 A1 WO 2022077901A1 CN 2021093451 W CN2021093451 W CN 2021093451W WO 2022077901 A1 WO2022077901 A1 WO 2022077901A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- bearing
- matching
- small sample
- failure mode
- dimensional
- 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.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Definitions
- the invention relates to the technical field of fault diagnosis of high-end equipment structures, in particular to a bearing fault mode diagnosis method and system for small sample data sets.
- rolling bearings As a key component in modern high-end equipment, rolling bearings have a weak ability to withstand shocks and are extremely prone to fatigue and damage. Once a failure occurs, it will have a huge negative impact on the entire production process, not only causing serious economic losses, but even endangering the lives of relevant personnel. Therefore, it is extremely necessary to carry out fault diagnosis technology research on rolling bearings, which is of great significance for the predictive maintenance of high-end equipment.
- Meta-learning is mainly used to solve the problem of learning to learn. Different from previous machine learning and deep learning methods, meta-learning pays more attention to how to use known knowledge to quickly adapt to the learning of new tasks, so it can effectively solve the above two problems. In recent years, meta-learning has emerged and has played a significant role in solving the learning problem of a small number of labeled samples or even unlabeled sample data. However, the application of this method in the field of high-end equipment fault diagnosis is very lacking at present.
- the purpose of the present invention is to provide a bearing failure mode diagnosis method and system for small sample data sets in order to overcome the above-mentioned defects of the prior art. In addition, it can also alleviate the performance degradation problem caused by the inconsistency of the sample distribution of the training set and the test set to a certain extent.
- a bearing failure mode diagnosis method for small sample data set includes the following steps:
- Step 1 Collect the vibration signal data of the bearing under different operating conditions of different equipment through the acceleration sensor, and store it in the server;
- Step 2 Preprocess the signal in the server, convert the original one-dimensional signal into a two-dimensional signal through the continuous wavelet transform algorithm, and store it in the database in the form of an image;
- Step 3 Construct a bearing fault diagnosis model framework based on convolutional neural network, including encoding module and matching module, and randomly sample from the image data in the database to construct a learning task of multiple small sample data sets to diagnose bearing faults model for training;
- Step 4 Collect the vibration signal of the target bearing, diagnose it according to the preprocessing method in the step 2 and the bearing fault diagnosis model framework in the step 3, and obtain the bearing failure mode.
- step 2 comprises the following sub-steps:
- Step 201 The vibration data collected by the acceleration sensor is a one-dimensional continuous time series signal, and the signal is preprocessed by a continuous wavelet transform algorithm to obtain a two-dimensional signal;
- Step 202 The two-dimensional signal is converted into image gray value and stored in the database in the form of image gray value conversion.
- CWT f (a, b) is a two-dimensional signal
- f(t) is a one-dimensional vibration signal
- ⁇ (t) is the complex conjugate of the wavelet function ⁇ (t)
- a and b represent the scaling and translation factors, respectively;
- the two-dimensional signal is converted into an image gray value, and the description formula is as follows:
- image(CWT f (a,b)) is the image data after the two-dimensional signal is converted into gray value
- max( ⁇ ) is the maximum value function
- min( ⁇ ) is the minimum value function
- step 3 comprises the following sub-steps:
- Step 301 randomly select l-type samples from all images as a small sample set for training, and construct a learning task
- Step 302 Pass the sampling set and the query set in each learning task through the coding module to obtain a high-dimensional coding expression, and after averaging the coding values of the obtained sampling sets of the same type, perform the query set on the feature dimension. Splicing processing to form matching pairs;
- Step 303 Pass each matching pair through the matching module to obtain a matching score
- Step 304 use the mean variance quantification model to predict the error between the matching score and the actual matching score;
- Step 305 Use the back-propagation algorithm in deep learning to optimize the model parameters until the final training is completed.
- Task i is the learning task, is the sampling set, is the query set, m is the number of samples in the sampling set, n is the number of samples in the query set, k and j are both natural numbers;
- Pair(l,k) is the matching pair
- Cat( ) is the splicing function in the feature dimension
- M i,l is the average value of the coded value of the l-th sample sample in the ith task
- r l,k ⁇ [0,1] is the matching score between the kth query set and the lth class
- g ⁇ ( ) is the function map of the matching module
- the matching score predicted by the model in the step 304 and the actual matching score error is:
- Loss is the error between the matching score predicted by the model and the actual matching score
- the back-propagation algorithm in deep learning is used to optimize the model parameters, and its description formula is:
- step 4 comprises the following sub-steps:
- Step 401 Pass all the samples of known categories in the database through the encoding module, and store the output high-dimensional encoded expression in the database;
- Step 402 After the vibration signal of the target bearing passes through the preprocessing and encoding module, it is then simultaneously used as the input of the matching module with the high-dimensional encoding of the known category in the database, so as to obtain a matching score with each category in the known category;
- Step 403 Take the maximum value among all matching scores of each category in the known categories, and its corresponding category is the failure mode of the target bearing.
- step 403 the maximum value of the matching scores of all and each category in the known categories is taken, and the description formula is:
- class is the class corresponding to the maximum matching score, that is, the failure mode of the target bearing.
- the convolutional neural network adopts a deep neural network with sparse connection and parameter sharing characteristics.
- the present invention also provides a system for the aforementioned method for diagnosing bearing failure modes oriented to small sample data sets, the system comprising:
- the preprocessing module is used to convert the one-dimensional vibration signal of the bearing into a two-dimensional signal through continuous wavelet transformation, and perform image gray value conversion on it;
- the task generation module is used to randomly sample from the preprocessed image data to construct a learning task of multiple small sample sets, wherein each small sample set includes a sampling set and a query set;
- an encoding module configured to perform function mapping on the samples of the sampling set and the query set to obtain an encoded expression in a higher dimensional space
- the matching module is used to match the coding of the query set samples with the coding of each category in the sampling set, so as to obtain the corresponding categories of the query set samples;
- the diagnosis module is used for collecting the vibration signal of the target bearing, and diagnosing the bearing failure mode according to the vibration signal of the target bearing and the bearing failure diagnosis model.
- the basic structures of the encoding module and the matching module are both convolutional neural networks, and the convolutional neural networks use deep neural networks with sparse connection and parameter sharing characteristics.
- the present invention has the following advantages:
- Both the coding module and the matching module in the present invention are based on the convolutional neural network and have two characteristics of sparse connection and parameter sharing, which are extremely suitable for deep feature expression mining of image data.
- the preprocessing module in the present invention can process non-stationary and nonlinear signals through the continuous wavelet transform method, so as to obtain a more robust feature expression.
- the task generation module in the present invention can construct learning tasks of multiple small sample sets through image data, so that the model can learn transferable knowledge between different tasks and help the model to quickly adapt to new tasks.
- FIG. 1 is a partial bearing signal preprocessing image of the present invention.
- FIG. 2 is a schematic diagram of the bearing failure mode diagnosis framework of the present invention.
- FIG. 3 is a schematic diagram of the model structure of the present invention.
- FIG. 4 is a flow chart of the method of the present invention.
- FIG. 5 is a schematic diagram of the fault diagnosis system of the present invention.
- the present invention provides a bearing failure mode diagnosis method oriented to small sample data sets, as shown in FIG. 4 , including the following steps:
- the bearing vibration timing signal collected in this example has four different working conditions. Each working condition includes 10 bearing failure modes. In addition to normal, there are 9 fault types, including three different faults. Locations: inner ring fault, ball fault, and outer ring fault, with three different fault sizes for each fault location.
- the bearing fault feature map is formed, part of which is shown in Figure 1, and finally stored in the database of the server.
- Step 2) specifically includes:
- Step 201 The vibration data collected by the acceleration sensor is a one-dimensional continuous time series signal, and the signal is preprocessed by a continuous wavelet transform algorithm to obtain a two-dimensional signal;
- Step 202 The two-dimensional signal is converted into image gray value and stored in the database in the form of image gray value conversion.
- the description formula of the two-dimensional signal in step 201 is:
- CWT f (a, b) is a two-dimensional signal
- f(t) is a one-dimensional vibration signal
- ⁇ (t) is the complex conjugate of the wavelet function ⁇ (t)
- a and b represent the scaling and translation factors, respectively;
- step 202 the two-dimensional signal is converted into an image gray value, and its description formula is:
- image(CWT f (a,b)) is the image data after the two-dimensional signal is converted into gray value
- max( ⁇ ) is the maximum value function
- min( ⁇ ) is the minimum value function
- Task i is the learning task, is the sampling set, is the query set, m is the number of samples in the sampling set, n is the number of samples in the query set, k and j are both natural numbers, and each sample in the sampling set and the query set is composed of a pair of fault features and fault modes;
- sampling samples and query samples of each task pass through the encoding module to obtain high-dimensional encoded representations. After the coding values of the samples of the same class are averaged, they are spliced with the query samples in the feature dimension to form matching pairs:
- Pair(l,k) is the matching pair
- Cat( ) is the splicing function in the feature dimension
- M i,l is the average value of the coded value of the l-th sample sample in the ith task, is the parameter of the encoding module
- Each matching pair passes through the matching module to obtain a matching score:
- r l,k ⁇ [0,1] is the matching score between the kth query set and the lth class
- g ⁇ ( ) is the function map of the matching module
- ⁇ is the parameter of the matching module
- Loss is the error between the matching score predicted by the model and the actual matching score
- the category corresponding to the maximum matching score is the failure mode of the target bearing:
- class is the class corresponding to the maximum matching score, that is, the failure mode of the target bearing.
- FIG. 5 is a schematic structural diagram of a bearing failure mode diagnosis system oriented to a small sample data set according to the present invention.
- the bearing failure mode diagnosis system 10 for small sample data sets includes: a preprocessing module 100 , a task generation module 200 , an encoding module 300 , a matching module 400 , and a diagnosis module 500 .
- the preprocessing module 100 converts the one-dimensional vibration signal of the bearing into a two-dimensional signal through continuous wavelet transformation, and performs image gray value conversion on it.
- the task generation module 200 randomly samples from the preprocessed image data, so as to construct a learning task of multiple small sample sets, wherein each small sample set includes a sampling set and a query set.
- the encoding module 300 performs functional mapping on the samples of the sample set and the query set to obtain the encoded expression in a higher dimensional space.
- the matching module 400 matches the codes of the query set samples with the codes of various categories in the sample set, so as to obtain the corresponding categories of the query set samples.
- the diagnosis model 500 is used to collect the vibration signal of the target bearing, and diagnose the bearing failure mode according to the vibration signal of the bearing and the bearing failure diagnosis model.
- the system 10 of the embodiment of the present invention combines deep learning and meta-learning algorithms to improve diagnostic accuracy in small sample data sets.
- the preprocessing module 100 can process non-stationary and nonlinear signals through the continuous wavelet transform method to obtain a more robust feature expression.
- the task generation module 200 can construct learning tasks of multiple small sample sets through image data, so that the model can learn transferable knowledge between different tasks and help the model quickly adapt to new tasks.
- both the encoding module 300 and the matching module 400 are based on a convolutional neural network, with two characteristics of sparse connection and parameter sharing, which are extremely suitable for deep feature expression mining of image data.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Library & Information Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Acoustics & Sound (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
Description
Claims (10)
- 一种面向小样本数据集的轴承故障模式诊断方法,其特征在于,该方法包括以下步骤:步骤1:通过加速度传感器收集轴承在不同设备不同工况运行下的振动信号数据,并将其存储至服务器中;步骤2:于服务器中对信号进行预处理,通过连续小波变换算法,将原始一维信号转化为二维信号,并以图像的形式存储于数据库中;步骤3:构建基于卷积神经网络的轴承故障诊断模型框架,包括编码模块和匹配模块,同时从数据库中的图像数据中随机取样,构建多个小样本数据集的学习任务,以对轴承故障诊断模型进行训练;步骤4:采集目标轴承的振动信号,并根据所述步骤2中的预处理方法和所述步骤3中的轴承故障诊断模型框架对其诊断并得出轴承故障模式。
- 根据权利要求1所述的一种面向小样本数据集的轴承故障模式诊断方法,其特征在于,所述的步骤2包括以下分步骤:步骤201:由加速度传感器收集到的振动数据为一维的连续时间序列信号,通过连续小波变换算法预处理该信号,得到二维信号;步骤202:将二维信号进行图像灰度值化以转换为图像的形式存储于数据库中。
- 根据权利要求1所述的一种面向小样本数据集的轴承故障模式诊断方法,其特征在于,所述的步骤3包括以下分步骤:步骤301:从所有图像中随机选取l类样本作为训练的小样本集,构建学习任务;步骤302:将每个学习任务中的采样集和查询集通过所述编码模块,获得高维编码表达,针对得到的同一类的采样集的编码值取平均后,在特征维度上与查询集进行拼接处理,形成匹配对;步骤303:将每个匹配对通过所述匹配模块,获得匹配分数;步骤304:使用平均方差量化模型预测的匹配分数与实际匹配分数误差;步骤305:使用深度学习中的反向传播算法优化模型参数,直至最终训练完毕。
- 根据权利要求4所述的一种面向小样本数据集的轴承故障模式诊断方法,其特征在于,所述的步骤301中的学习任务,其描述公式为:所述步骤302中的匹配对,其描述公式为:所述步骤303中的匹配分数,其描述公式为:r l,k=g φ(Pair(l,k))式中,r l,k∈[0,1],为第k个查询集与第l类的匹配分数,g φ(·)为匹配模块的函数映射;所述步骤304中的模型预测的匹配分数与实际匹配分数误差,其描述公式为:式中,Loss为模型预测的匹配分数与实际匹配分数误差;所述305中使用深度学习中的反向传播算法优化模型参数,其描述公式为:
- 根据权利要求1所述的一种面向小样本数据集的轴承故障模式诊断方法,其特征在于,所述的步骤4包括以下分步骤:步骤401:将数据库中所有已知类别的样本通过编码模块,并将输出的高维编码表达存储于数据库;步骤402:目标轴承的振动信号通过预处理、编码模块后,再与数据库中已知类别的高维编码同时作为匹配模块的输入,从而获得和已知类别中每个类别的匹配分数;步骤403:取所有和已知类别中每个类别的匹配分数中的最大值,其对应的类别即目标轴承的故障模式。
- 根据权利要求4所述的一种面向小样本数据集的轴承故障模式诊断方法,其特征在于,所述的卷积神经网络采用具有稀疏连接和参数共享特性的深层神经网络。
- 一种用于如权利要求1至8中任一项所述的一种面向小样本数据集的轴 承故障模式诊断方法的系统,其特征在于,该系统包括:预处理模块,用于通过连续小波变换将轴承一维振动信号转换为二维信号,并对其进行图像灰度值化;任务生成模块,用于从预处理后的图像数据中随机取样,从而构建多个小样本集的学习任务,其中每个小样本集包括采样集与查询集;编码模块,用于对所述采样集与查询集的样本进行函数映射,获得更高维空间中的编码表达;匹配模块,用于将查询集样本的编码与采样集中各类别的编码进行匹配,从而得到查询集样本的相对应类别;诊断模块,用于采集目标轴承的振动信号,并根据所述目标轴承的振动信号和所述轴承故障诊断模型诊断出所述轴承故障模式。
- 根据权利要求9所述的一种用于所述的一种面向小样本数据集的轴承故障模式诊断方法的系统,其特征在于,所述编码模块和所述匹配模块的基础架构均为卷积神经网络,所述卷积神经网络采用具有稀疏连接和参数共享特性的深层神经网络。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011091094.XA CN112417954B (zh) | 2020-10-13 | 2020-10-13 | 一种面向小样本数据集的轴承故障模式诊断方法及系统 |
| CN202011091094.X | 2020-10-13 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2022077901A1 true WO2022077901A1 (zh) | 2022-04-21 |
Family
ID=74854429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2021/093451 Ceased WO2022077901A1 (zh) | 2020-10-13 | 2021-05-12 | 一种面向小样本数据集的轴承故障模式诊断方法及系统 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112417954B (zh) |
| WO (1) | WO2022077901A1 (zh) |
Cited By (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114818826A (zh) * | 2022-05-19 | 2022-07-29 | 石家庄铁道大学 | 基于轻量级Vision Transformer模块的故障诊断方法 |
| CN114818811A (zh) * | 2022-05-02 | 2022-07-29 | 大连理工大学 | 一种基于孪生网络度量学习的航空发动机滚动轴承故障诊断方法 |
| CN114896733A (zh) * | 2022-05-20 | 2022-08-12 | 合肥工业大学 | 一种基于深度强化学习的滚动轴承故障类型识别方法 |
| CN114923691A (zh) * | 2022-06-24 | 2022-08-19 | 杭州星宸智联技术有限公司 | 一种高通用性、多场景智能化的滚动轴承故障诊断系统 |
| CN114925725A (zh) * | 2022-05-09 | 2022-08-19 | 频率探索智能科技江苏有限公司 | 基于多尺度lbp-cnn网络的滚动轴承故障诊断方法和系统 |
| CN114942140A (zh) * | 2022-05-31 | 2022-08-26 | 杭州电子科技大学 | 基于多输入并行图卷积神经网络的滚动轴承故障诊断方法 |
| CN114993677A (zh) * | 2022-05-11 | 2022-09-02 | 山东大学 | 不平衡小样本数据的滚动轴承故障诊断方法及系统 |
| CN115046766A (zh) * | 2022-06-15 | 2022-09-13 | 石家庄铁道大学 | 基于二维灰度图像自适应子空间的小样本轴承故障诊断方法 |
| CN115424053A (zh) * | 2022-07-25 | 2022-12-02 | 北京邮电大学 | 小样本图像识别方法、装置、设备及存储介质 |
| CN115511063A (zh) * | 2022-11-10 | 2022-12-23 | 东南大学 | 基于Transformer神经网络的小样本结构故障诊断方法 |
| CN115586005A (zh) * | 2022-10-21 | 2023-01-10 | 浙江理工大学 | 基于递归图和深度自适应Transformer的轴承故障检测方法及系统 |
| CN115674272A (zh) * | 2022-11-03 | 2023-02-03 | 中国科学技术大学先进技术研究院 | 机器人故障诊断方法、装置、设备以及存储介质 |
| CN115952408A (zh) * | 2023-01-05 | 2023-04-11 | 东北大学 | 多通道跨域少样本的冲压生产线轴承故障诊断方法 |
| CN115980560A (zh) * | 2022-11-29 | 2023-04-18 | 西安交通大学 | 一种基于cnn-gru的高压断路器机械故障诊断系统、方法、设备 |
| CN116070098A (zh) * | 2022-09-16 | 2023-05-05 | 河北工业大学 | 一种基于元学习的设备故障诊断方法 |
| CN116127279A (zh) * | 2023-02-24 | 2023-05-16 | 上海师范大学 | 一种小样本液压泵气蚀诊断方法及装置 |
| CN116222753A (zh) * | 2023-01-09 | 2023-06-06 | 中国石油大学(北京) | 一种转子系统故障敏感特征提取方法及系统 |
| CN116223009A (zh) * | 2023-03-07 | 2023-06-06 | 国网江苏省电力有限公司超高压分公司 | 一种换流阀中主循环泵的故障诊断方法、系统、装置及介质 |
| CN116304603A (zh) * | 2023-04-03 | 2023-06-23 | 中铁工程装备集团有限公司 | 一种基于双对比架构的模式识别方法 |
| CN116558828A (zh) * | 2023-07-10 | 2023-08-08 | 昆明理工大学 | 基于自相关系数稀疏度特征的滚动轴承健康状态评估方法 |
| CN116701912A (zh) * | 2023-06-14 | 2023-09-05 | 盐城工学院 | 基于一维卷积神经网络的轴承故障诊断方法及系统 |
| CN116728291A (zh) * | 2023-08-16 | 2023-09-12 | 湖南大学 | 基于边缘计算的机器人打磨系统状态监测方法和装置 |
| CN116821697A (zh) * | 2023-08-30 | 2023-09-29 | 聊城莱柯智能机器人有限公司 | 一种基于小样本学习的机械设备故障诊断方法 |
| CN116894190A (zh) * | 2023-09-11 | 2023-10-17 | 江西南昌济生制药有限责任公司 | 轴承故障诊断方法、装置、电子设备和存储介质 |
| CN116933030A (zh) * | 2023-07-26 | 2023-10-24 | 中冶赛迪技术研究中心有限公司 | 一种机械设备故障诊断方法及系统 |
| CN116992341A (zh) * | 2023-06-27 | 2023-11-03 | 合肥工业大学 | 一种轴承故障诊断的分类方法 |
| CN116994076A (zh) * | 2023-09-28 | 2023-11-03 | 中国海洋大学 | 一种基于双分支相互学习特征生成的小样本图像识别方法 |
| CN117113062A (zh) * | 2023-09-18 | 2023-11-24 | 昆明理工大学 | 一种基于脑电运动想象的空域特征强化方法 |
| CN117217307A (zh) * | 2023-10-11 | 2023-12-12 | 山东师范大学 | 基于知识图谱和卷积神经网络的机械故障预测方法及系统 |
| CN117407796A (zh) * | 2023-12-15 | 2024-01-16 | 合肥工业大学 | 一种跨部件小样本故障诊断方法、系统和存储介质 |
| WO2024045555A1 (zh) * | 2022-08-30 | 2024-03-07 | 山东建筑大学 | 一种基于标准自学习数据增强的故障诊断方法及系统 |
| US20240085274A1 (en) * | 2022-09-12 | 2024-03-14 | University Of South Carolina | Hybrid bearing fault prognosis with fault detection and multiple model fusion |
| CN117892194A (zh) * | 2024-01-10 | 2024-04-16 | 兰州理工大学 | 基于Meta-GraphSage网络的小样本故障诊断方法及系统 |
| CN118013289A (zh) * | 2024-04-09 | 2024-05-10 | 北京理工大学 | 一种基于信息融合元迁移学习的变工况小样本故障诊断方法、装置、介质及产品 |
| CN118533476A (zh) * | 2024-05-09 | 2024-08-23 | 哈尔滨理工大学 | 基于振动信号延拓和时频信息融合网络的轴承故障诊断方法 |
| CN118818205A (zh) * | 2024-06-06 | 2024-10-22 | 国网宁夏电力有限公司中卫供电公司 | 一种分布式电池储能电站谐波故障诊断方法、介质及系统 |
| CN118940126A (zh) * | 2024-10-15 | 2024-11-12 | 山东轴研精密轴承有限公司 | 一种基于人工智能的轴承质量检测系统 |
| CN119249313A (zh) * | 2024-09-03 | 2025-01-03 | 复旦大学 | 基于故障特征优化的度量网络少样本故障诊断方法 |
| CN121561610A (zh) * | 2026-01-20 | 2026-02-24 | 湖南科技大学三亚研究院 | 基于在线软标签高斯原型网络的滚动轴承小样本故障诊断方法及系统 |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112417954B (zh) * | 2020-10-13 | 2022-12-16 | 同济大学 | 一种面向小样本数据集的轴承故障模式诊断方法及系统 |
| CN113128561A (zh) * | 2021-03-22 | 2021-07-16 | 南京航空航天大学 | 一种机床轴承故障诊断方法 |
| CN113869339A (zh) * | 2021-05-18 | 2021-12-31 | 华能沁北发电有限责任公司 | 用于故障诊断的深度学习分类模型及故障诊断方法 |
| CN113758709A (zh) * | 2021-09-30 | 2021-12-07 | 河南科技大学 | 结合边缘计算和深度学习的滚动轴承故障诊断方法及系统 |
| CN113988120B (zh) * | 2021-10-15 | 2025-09-23 | 同济大学 | 一种基于双图网络的小样本轴承故障诊断方法 |
| CN116126569A (zh) * | 2022-11-01 | 2023-05-16 | 中国工商银行股份有限公司 | 智能运维方法及装置 |
| CN115830383A (zh) * | 2022-12-08 | 2023-03-21 | 昆明理工大学 | 一种用于小样本条件下电机轴承故障诊断模型训练方法、装置 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200050667A1 (en) * | 2018-08-09 | 2020-02-13 | CloudMinds Technology, Inc. | Intent Classification Method and System |
| CN110823574A (zh) * | 2019-09-30 | 2020-02-21 | 安徽富煌科技股份有限公司 | 一种基于半监督学习深度对抗网络的故障诊断方法 |
| CN111597948A (zh) * | 2020-05-11 | 2020-08-28 | 苏州求臻智能科技有限公司 | 一种轴承振动信号的故障检测和分类方法 |
| CN111721535A (zh) * | 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 | 一种基于卷积多头自注意力机制的轴承故障检测方法 |
| CN112417954A (zh) * | 2020-10-13 | 2021-02-26 | 同济大学 | 一种面向小样本数据集的轴承故障模式诊断方法及系统 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100468263C (zh) * | 2007-09-05 | 2009-03-11 | 东北大学 | 连采机远程实时故障预测及诊断方法与装置 |
| US20190374160A1 (en) * | 2017-01-05 | 2019-12-12 | The Trustees Of Princeton University | Hierarchical health decision support system and method |
| CN109271936B (zh) * | 2018-09-18 | 2021-09-24 | 哈尔滨工程大学 | 基于感知哈希算法的飞机振动故障数据库构建与检索方法 |
| CN109633369B (zh) * | 2018-12-08 | 2020-12-04 | 国网山东省电力公司德州供电公司 | 一种基于多维数据相似性匹配的电网故障诊断方法 |
| CN109682596B (zh) * | 2018-12-20 | 2020-11-13 | 南京航空航天大学 | 非均衡样本下高速重载输入级故障诊断方法 |
| CN110516305B (zh) * | 2019-07-26 | 2021-02-12 | 西安交通大学 | 基于注意机制元学习模型的小样本下故障智能诊断方法 |
-
2020
- 2020-10-13 CN CN202011091094.XA patent/CN112417954B/zh active Active
-
2021
- 2021-05-12 WO PCT/CN2021/093451 patent/WO2022077901A1/zh not_active Ceased
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200050667A1 (en) * | 2018-08-09 | 2020-02-13 | CloudMinds Technology, Inc. | Intent Classification Method and System |
| CN110823574A (zh) * | 2019-09-30 | 2020-02-21 | 安徽富煌科技股份有限公司 | 一种基于半监督学习深度对抗网络的故障诊断方法 |
| CN111597948A (zh) * | 2020-05-11 | 2020-08-28 | 苏州求臻智能科技有限公司 | 一种轴承振动信号的故障检测和分类方法 |
| CN111721535A (zh) * | 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 | 一种基于卷积多头自注意力机制的轴承故障检测方法 |
| CN112417954A (zh) * | 2020-10-13 | 2021-02-26 | 同济大学 | 一种面向小样本数据集的轴承故障模式诊断方法及系统 |
Cited By (48)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114818811A (zh) * | 2022-05-02 | 2022-07-29 | 大连理工大学 | 一种基于孪生网络度量学习的航空发动机滚动轴承故障诊断方法 |
| CN114925725A (zh) * | 2022-05-09 | 2022-08-19 | 频率探索智能科技江苏有限公司 | 基于多尺度lbp-cnn网络的滚动轴承故障诊断方法和系统 |
| CN114993677A (zh) * | 2022-05-11 | 2022-09-02 | 山东大学 | 不平衡小样本数据的滚动轴承故障诊断方法及系统 |
| CN114818826A (zh) * | 2022-05-19 | 2022-07-29 | 石家庄铁道大学 | 基于轻量级Vision Transformer模块的故障诊断方法 |
| CN114896733A (zh) * | 2022-05-20 | 2022-08-12 | 合肥工业大学 | 一种基于深度强化学习的滚动轴承故障类型识别方法 |
| CN114896733B (zh) * | 2022-05-20 | 2024-02-13 | 合肥工业大学 | 一种基于深度强化学习的滚动轴承故障类型识别方法 |
| CN114942140A (zh) * | 2022-05-31 | 2022-08-26 | 杭州电子科技大学 | 基于多输入并行图卷积神经网络的滚动轴承故障诊断方法 |
| CN115046766A (zh) * | 2022-06-15 | 2022-09-13 | 石家庄铁道大学 | 基于二维灰度图像自适应子空间的小样本轴承故障诊断方法 |
| CN114923691A (zh) * | 2022-06-24 | 2022-08-19 | 杭州星宸智联技术有限公司 | 一种高通用性、多场景智能化的滚动轴承故障诊断系统 |
| CN115424053A (zh) * | 2022-07-25 | 2022-12-02 | 北京邮电大学 | 小样本图像识别方法、装置、设备及存储介质 |
| WO2024045555A1 (zh) * | 2022-08-30 | 2024-03-07 | 山东建筑大学 | 一种基于标准自学习数据增强的故障诊断方法及系统 |
| US20240085274A1 (en) * | 2022-09-12 | 2024-03-14 | University Of South Carolina | Hybrid bearing fault prognosis with fault detection and multiple model fusion |
| CN116070098A (zh) * | 2022-09-16 | 2023-05-05 | 河北工业大学 | 一种基于元学习的设备故障诊断方法 |
| CN115586005A (zh) * | 2022-10-21 | 2023-01-10 | 浙江理工大学 | 基于递归图和深度自适应Transformer的轴承故障检测方法及系统 |
| CN115674272A (zh) * | 2022-11-03 | 2023-02-03 | 中国科学技术大学先进技术研究院 | 机器人故障诊断方法、装置、设备以及存储介质 |
| CN115511063A (zh) * | 2022-11-10 | 2022-12-23 | 东南大学 | 基于Transformer神经网络的小样本结构故障诊断方法 |
| CN115980560A (zh) * | 2022-11-29 | 2023-04-18 | 西安交通大学 | 一种基于cnn-gru的高压断路器机械故障诊断系统、方法、设备 |
| CN115952408A (zh) * | 2023-01-05 | 2023-04-11 | 东北大学 | 多通道跨域少样本的冲压生产线轴承故障诊断方法 |
| CN116222753A (zh) * | 2023-01-09 | 2023-06-06 | 中国石油大学(北京) | 一种转子系统故障敏感特征提取方法及系统 |
| CN116127279A (zh) * | 2023-02-24 | 2023-05-16 | 上海师范大学 | 一种小样本液压泵气蚀诊断方法及装置 |
| CN116127279B (zh) * | 2023-02-24 | 2024-04-19 | 上海师范大学 | 一种小样本液压泵气蚀诊断方法及装置 |
| CN116223009A (zh) * | 2023-03-07 | 2023-06-06 | 国网江苏省电力有限公司超高压分公司 | 一种换流阀中主循环泵的故障诊断方法、系统、装置及介质 |
| CN116304603A (zh) * | 2023-04-03 | 2023-06-23 | 中铁工程装备集团有限公司 | 一种基于双对比架构的模式识别方法 |
| CN116701912A (zh) * | 2023-06-14 | 2023-09-05 | 盐城工学院 | 基于一维卷积神经网络的轴承故障诊断方法及系统 |
| CN116701912B (zh) * | 2023-06-14 | 2023-11-14 | 盐城工学院 | 基于一维卷积神经网络的轴承故障诊断方法及系统 |
| CN116992341A (zh) * | 2023-06-27 | 2023-11-03 | 合肥工业大学 | 一种轴承故障诊断的分类方法 |
| CN116558828B (zh) * | 2023-07-10 | 2023-09-15 | 昆明理工大学 | 基于自相关系数稀疏度特征的滚动轴承健康状态评估方法 |
| CN116558828A (zh) * | 2023-07-10 | 2023-08-08 | 昆明理工大学 | 基于自相关系数稀疏度特征的滚动轴承健康状态评估方法 |
| CN116933030A (zh) * | 2023-07-26 | 2023-10-24 | 中冶赛迪技术研究中心有限公司 | 一种机械设备故障诊断方法及系统 |
| CN116728291B (zh) * | 2023-08-16 | 2023-10-31 | 湖南大学 | 基于边缘计算的机器人打磨系统状态监测方法和装置 |
| CN116728291A (zh) * | 2023-08-16 | 2023-09-12 | 湖南大学 | 基于边缘计算的机器人打磨系统状态监测方法和装置 |
| CN116821697A (zh) * | 2023-08-30 | 2023-09-29 | 聊城莱柯智能机器人有限公司 | 一种基于小样本学习的机械设备故障诊断方法 |
| CN116821697B (zh) * | 2023-08-30 | 2024-05-28 | 聊城莱柯智能机器人有限公司 | 一种基于小样本学习的机械设备故障诊断方法 |
| CN116894190A (zh) * | 2023-09-11 | 2023-10-17 | 江西南昌济生制药有限责任公司 | 轴承故障诊断方法、装置、电子设备和存储介质 |
| CN116894190B (zh) * | 2023-09-11 | 2023-11-28 | 江西南昌济生制药有限责任公司 | 轴承故障诊断方法、装置、电子设备和存储介质 |
| CN117113062A (zh) * | 2023-09-18 | 2023-11-24 | 昆明理工大学 | 一种基于脑电运动想象的空域特征强化方法 |
| CN116994076A (zh) * | 2023-09-28 | 2023-11-03 | 中国海洋大学 | 一种基于双分支相互学习特征生成的小样本图像识别方法 |
| CN116994076B (zh) * | 2023-09-28 | 2024-01-19 | 中国海洋大学 | 一种基于双分支相互学习特征生成的小样本图像识别方法 |
| CN117217307A (zh) * | 2023-10-11 | 2023-12-12 | 山东师范大学 | 基于知识图谱和卷积神经网络的机械故障预测方法及系统 |
| CN117407796B (zh) * | 2023-12-15 | 2024-03-01 | 合肥工业大学 | 一种跨部件小样本故障诊断方法、系统和存储介质 |
| CN117407796A (zh) * | 2023-12-15 | 2024-01-16 | 合肥工业大学 | 一种跨部件小样本故障诊断方法、系统和存储介质 |
| CN117892194A (zh) * | 2024-01-10 | 2024-04-16 | 兰州理工大学 | 基于Meta-GraphSage网络的小样本故障诊断方法及系统 |
| CN118013289A (zh) * | 2024-04-09 | 2024-05-10 | 北京理工大学 | 一种基于信息融合元迁移学习的变工况小样本故障诊断方法、装置、介质及产品 |
| CN118533476A (zh) * | 2024-05-09 | 2024-08-23 | 哈尔滨理工大学 | 基于振动信号延拓和时频信息融合网络的轴承故障诊断方法 |
| CN118818205A (zh) * | 2024-06-06 | 2024-10-22 | 国网宁夏电力有限公司中卫供电公司 | 一种分布式电池储能电站谐波故障诊断方法、介质及系统 |
| CN119249313A (zh) * | 2024-09-03 | 2025-01-03 | 复旦大学 | 基于故障特征优化的度量网络少样本故障诊断方法 |
| CN118940126A (zh) * | 2024-10-15 | 2024-11-12 | 山东轴研精密轴承有限公司 | 一种基于人工智能的轴承质量检测系统 |
| CN121561610A (zh) * | 2026-01-20 | 2026-02-24 | 湖南科技大学三亚研究院 | 基于在线软标签高斯原型网络的滚动轴承小样本故障诊断方法及系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112417954B (zh) | 2022-12-16 |
| CN112417954A (zh) | 2021-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112417954B (zh) | 一种面向小样本数据集的轴承故障模式诊断方法及系统 | |
| CN110555273B (zh) | 基于隐马尔科夫模型和迁移学习的轴承寿命预测方法 | |
| Wang et al. | A method for rapidly evaluating reliability and predicting remaining useful life using two-dimensional convolutional neural network with signal conversion | |
| CN114239402B (zh) | 基于优化胶囊网络的核动力循环水泵故障诊断方法及系统 | |
| WO2022037068A1 (zh) | 一种机床轴承故障诊断方法 | |
| CN115640531B (zh) | 一种基于残差学习与注意力机制融合的故障诊断方法 | |
| CN115290326A (zh) | 一种滚动轴承故障智能诊断方法 | |
| CN116593157A (zh) | 少样本下基于匹配元学习的复杂工况齿轮故障诊断方法 | |
| CN115587290A (zh) | 基于变分自编码生成对抗网络的航空发动机故障诊断方法 | |
| CN112488179A (zh) | 一种基于gru的旋转机械故障诊断方法 | |
| CN117556317A (zh) | 一种基于时频图像的转动轴承故障诊断方法 | |
| CN113962289A (zh) | 面向终身学习的旋转机械在线智能故障诊断方法及系统 | |
| CN113723592A (zh) | 一种基于风电齿轮箱监测系统的故障诊断方法 | |
| CN117493793A (zh) | 一种基于改进脉冲可分离卷积增强Transformer encoder的剩余寿命预测方法 | |
| CN118193935A (zh) | 基于注意力机制的双通道特征融合轴承设备故障诊断方法 | |
| CN116720117A (zh) | 一种基于多分支融合卷积神经网络的轴承故障诊断方法 | |
| CN118035766B (zh) | 基于相似度对抗与对比学习的变工况轴承故障诊断方法 | |
| CN118190412A (zh) | 一种基于深度学习的轴承故障诊断方法及装置 | |
| CN118626929A (zh) | 基于时频增强与生成式学习的小样本滚动轴承故障诊断方法 | |
| CN117874466A (zh) | 一种基于注意力机制的端对端轴承寿命预报方法 | |
| CN116012687B (zh) | 一种用于轮对踏面缺陷识别的图像交互融合方法 | |
| CN115902620B (zh) | 一种基于循环生成对抗网络的电机故障增强诊断方法及装置 | |
| CN115586005A (zh) | 基于递归图和深度自适应Transformer的轴承故障检测方法及系统 | |
| CN116610940A (zh) | 一种基于小波变换与深度神经网络的装备故障诊断系统 | |
| CN113111752B (zh) | 针对样本不均衡的增强扩展深度置信网络的轧机故障诊断方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21878955 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 21878955 Country of ref document: EP Kind code of ref document: A1 |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 21878955 Country of ref document: EP Kind code of ref document: A1 |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 120A DATED 18/12/2023) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 21878955 Country of ref document: EP Kind code of ref document: A1 |

















