WO2020156348A1 - 基于总体经验模态分解与卷积神经网络的结构损伤识别法 - Google Patents
基于总体经验模态分解与卷积神经网络的结构损伤识别法 Download PDFInfo
- Publication number
- WO2020156348A1 WO2020156348A1 PCT/CN2020/073357 CN2020073357W WO2020156348A1 WO 2020156348 A1 WO2020156348 A1 WO 2020156348A1 CN 2020073357 W CN2020073357 W CN 2020073357W WO 2020156348 A1 WO2020156348 A1 WO 2020156348A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- layer
- processing
- neural network
- training
- feature map
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/46—Processing the detected response signal, e.g. electronic circuits specially adapted therefor by spectral analysis, e.g. Fourier analysis or wavelet analysis
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
Definitions
- This application belongs to the technical field of structural health monitoring, and in particular relates to a structural damage identification method based on total empirical mode decomposition and convolutional neural network.
- CN109145446A discloses a structural damage identification method based on modal strain energy and convolutional neural network, including S1: building a structural model through software simulation, and dividing units according to the structural model; S2: simulating several types of structural models in different units Structural damage; S3: Extract the first-order modal strain energy of the structure under free vibration, and convert the obtained data into matrix data form as the input of the convolutional neural network; S4: Train the convolutional neural network; S5 : Perform actual measurement of the structure, and calculate the modal strain energy of different units in different modes according to the unit division method of S1; S6: Substitute the data matrix of the modal vector of S5 into the trained convolutional neural network of S4 , Get the damage result of the structure.
- CN109115879A discloses a structural damage recognition method based on modal shape and convolutional neural network, including S1: construct a structural model through software simulation, and divide the unit according to the structural model; S2: simulate several types of structural models in different units Structural damage; S3: extract the first-order mode shape of the structure under free vibration, and convert the obtained data into matrix data form as the input of the convolutional neural network; S4: train the convolutional neural network; S5 : Perform actual measurement of the structure, and measure the displacement signal of the cell centroid according to the unit division of S1 to form a matrix data form corresponding to S3; S6: Substitute the data matrix of the cell centroid of S5 into the trained convolutional nerve of S4 Network, get the damage result of the structure.
- This application proposes a structural damage recognition method based on total empirical mode decomposition and convolutional neural network, which can improve recognition accuracy and recognition efficiency; its technical solution is:
- the structure damage identification method based on total empirical mode decomposition and convolutional neural network includes the following steps:
- Step 1 Measure the structure acceleration response signal x(t);
- Step 2 Decompose the acceleration response signal x(t) using the overall empirical mode decomposition technique to obtain a series of intrinsic mode function IMF; specifically, a series of intrinsic mode function IMF component signals.
- Step 3 Perform Fourier transform on the intrinsic mode function IMF obtained in Step 2 to obtain its spectral characteristics, and select the intrinsic mode function IMF containing the natural frequency of the structure for reconstruction to obtain the reconstructed signal z(t);
- the IMF component signals of the intrinsic mode function obtained in step 2 are respectively Fourier transformed to obtain their respective spectral characteristics, and the intrinsic mode function IMF component signals containing the natural frequency of the structure are selected for reconstruction to obtain the reconstructed signal z(t);
- Step 4 Standardize the reconstructed signal z(t) obtained in Step 3 to obtain a standardized reconstructed signal Z(t); divide the standardized reconstructed signal Z(t) into multiple sub-signals with a certain length, and select the sub-signals The part in is used as training samples, and the remaining part is used as test samples;
- Step 5 Input the training samples obtained in step 4 into the one-dimensional convolutional neural network model for training.
- the cross-entropy loss function is used as the loss function
- the Adam optimization algorithm is used as the optimization algorithm
- Step 6 Input the test samples obtained in Step 4 into the one-dimensional convolutional neural network model trained in Step 5 to obtain the recognition result of the damage location.
- the reconstructed signal z(t) is standardized using a standardized processing formula, and the expression of the standardized processing formula is:
- Z(t) is the standardized reconstructed signal
- z(t) is the reconstructed signal
- ⁇ is the mean value of the reconstructed signal z(t)
- ⁇ is the variance of the reconstructed signal z(t).
- the one-dimensional convolutional neural network model includes an input layer, a first convolutional layer, a first batch normalization layer, a first pooling layer, and a second convolutional layer connected in sequence ,
- the training of the one-dimensional convolutional neural network model includes alternate forward propagation training and back propagation training until the network converges or the output result and setting of the forward propagation training The error value between the output results reaches the specified iteration termination condition.
- the forward propagation training step includes: inputting the training samples obtained in step 4 from the input layer, and the first convolutional layer processes the input training samples to obtain an initial processing feature map, and the initial processing feature map It is sent to the first batch normalization layer for normalization processing, and after the normalization processing, it is sent to the first pooling layer for pooling processing to obtain the first sampling feature map; the first sampling feature map is sent to The second convolutional layer is processed to obtain a secondary processing feature map, the secondary processing feature map is transmitted to the second batch normalization layer for normalization processing, and the normalization processing is transmitted to the The second pooling layer performs pooling processing to obtain a secondary sampling feature map; the secondary sampling feature map is sequentially sent to the first fully connected layer and the second fully connected layer for processing, and then sent to the output after processing Layer, the output layer classifies the processed sub-sampling feature map and outputs the classification vector of the damage location.
- the back propagation training step includes: returning the error value between the output result of the forward propagation training and the set output result according to the backward propagation of the forward propagation training to obtain the error value of each layer;
- the Adam optimization algorithm is used to adjust the network parameters of the one-dimensional convolutional neural network model.
- At least one embodiment of the present application provides a structural damage identification method based on total empirical mode decomposition and convolutional neural network, which selects the acceleration response signal of the structure as the analysis object, and uses the total empirical mode decomposition technology to perform the acceleration response signal Decomposition and reconstruction. Only the information related to structural damage is extracted during reconstruction, which can reduce the influence of noise unrelated to structural damage and improve the accuracy of damage recognition; at the same time, a one-dimensional convolutional neural network is used to reconstruct Signal feature extraction and damage classification, the efficiency of damage recognition is higher. Therefore, the structural damage identification method based on the overall empirical mode decomposition and convolutional neural network provided by this application has significant advantages in real-time structural health monitoring.
- FIG. 1 is a flowchart of a structural damage identification method based on total empirical mode decomposition and convolutional neural network provided by an embodiment of the application;
- Fig. 2 is a schematic structural diagram of a one-dimensional convolutional neural network provided by an embodiment of the application;
- an embodiment of the present application provides a structural damage identification method based on total empirical mode decomposition and convolutional neural network, including the following steps:
- Step 1 Measure the structure acceleration response signal x(t).
- the acceleration response signal x(t) is a time series composed of a set of data (for example, it can be 100 data), which can be regarded as a one-dimensional vector.
- Step 2 Decompose the acceleration response signal x(t) using the overall empirical mode decomposition technique to obtain a series of intrinsic mode functions IMF. Specifically, a series of intrinsic modal function component signals are obtained.
- h 1 (t) y(t)-m 1 (t) (2).
- h 1 (t) is the first IMF component c 1 (t) of y(t); if h 1 (t) does not satisfy For the intrinsic mode function IMF, use h 1 (t) instead of y(t) and repeat steps 1-3, then:
- k is the number of repetitions of steps 1-3
- m 1k (t) is the average value of the upper envelope and the lower envelope obtained when steps 1-2 are repeated for the kth time
- the intrinsic modal function IMF needs to meet two conditions: one is that the number of extreme points and the number of zero-crossing points are equal or different in the entire data range, and the other is that all the maximum and minimum points are at any point.
- the average value of the envelope formed at the point is equal to zero.
- r n (t) is the nth residual quantity
- r n-1 (t) is the n-1th residual quantity
- c n (t) is the nth IMF component of y(t) ;
- step (1) and step (2) add M white noise signal ⁇ (t) to x(t), and the white noise signal ⁇ (t) added each time is different, take M times of IMF component
- the average value of the set is taken as the final intrinsic mode function IMF, and the component expression of the final intrinsic mode function IMF is:
- c zi (t) is the final i-th IMF component
- c im (t) is the i-th IMF component obtained by adding the white noise signal for the m-th time
- M is the total number of additions of the white noise signal .
- step (3) the 5 groups of intrinsic mode functions IMF are averaged, then the final intrinsic mode function IMF can be obtained, which contains 10 averaged IMF component signals, that is, the series of intrinsic mode functions described (Component signal). Since this technology is an existing technology, it will not be repeated.
- the intrinsic mode function can refer to the intrinsic mode function containing multiple component signals on the one hand, and on the other hand, it can also refer to the component signal of the intrinsic mode function;
- the intrinsic modal functions containing multiple component signals are collectively referred to as intrinsic modal functions, and the component signals therein are called intrinsic modal function component signals to facilitate understanding of the technical solution.
- Step 3 Perform Fourier transform on the intrinsic mode function IMF obtained in step 2 to obtain their respective spectral characteristics, and select the intrinsic mode function IMF containing the natural frequency of the structure for reconstruction to obtain the reconstructed signal z(t).
- each component signal of a series of intrinsic mode functions obtained in step 2 Fourier transform is performed on each component signal of a series of intrinsic mode functions obtained in step 2 to obtain its respective (ie, each component signal) spectral characteristics; the spectral characteristics refer to the frequency contained in the signal Components; the frequency components in the spectral characteristics of a part of the component signal are the same as the natural frequency of the structure, which is the characteristic of the structure; then select this part of the component signal containing the natural mode function IMF of the structure’s natural frequency to reconstruct, get Reconstruct the signal z(t).
- the reconstruction described in this embodiment is mainly to perform superposition processing on selected component signals, and the superposition processing can be implemented by using conventional technical solutions in the field, which will not be repeated here.
- selecting the component signal containing the natural modal function of the natural frequency of the structure can reduce the influence of noise unrelated to structural damage and improve the accuracy of damage recognition.
- Step 4 Use a standardized processing formula to standardize the reconstructed signal z(t) obtained in step 3 to obtain a standardized reconstructed signal Z(t).
- the expression of the standardized processing formula is:
- Z(t) is the standardized reconstructed signal
- z(t) is the reconstructed signal
- ⁇ is the mean value of the reconstructed signal
- ⁇ is the variance of the reconstructed signal
- the standardized reconstructed signal Z(t) is divided into multiple sub-signals with a certain length, and some of the sub-signals are selected as training samples, and the remaining sub-signals are used as test samples.
- the standardized reconstructed signal Z(t) can be divided into multiple sub-signals of equal length; and the length of the sub-signal can be determined according to actual needs.
- Step 5 Input the training samples obtained in step 4 into the one-dimensional convolutional neural network model for training, and a trained one-dimensional convolutional neural network model can be obtained; among them, the cross-entropy loss function is used as the loss function in the training process.
- Adam optimization algorithm is used as the optimization algorithm.
- the one-dimensional convolutional neural network model includes the input layer 1, the first convolution layer 2, the first batch normalization layer 3, the first pooling layer 4, and the second Convolutional layer 5, second batch normalization layer 6, second pooling layer 7, first fully connected layer 8, second fully connected layer 9 and output layer 10; among them, the first convolutional layer 2 and the second
- the convolutional layer 5 contains 16 convolution kernels with a size of 10 ⁇ 1, and the pooling window sizes of the first pooling layer 4 and the second pooling layer 7 are both 2 ⁇ 1.
- the training of the one-dimensional convolutional neural network model includes alternate forward propagation training and back propagation training until the network converges or the error between the output result of forward propagation training and the set output result reaches the specified iteration termination condition.
- the forward propagation training step includes: input the training samples obtained in step 4 from the input layer 1, the first convolutional layer 2 processes the input training samples to obtain the initial processing feature map, and the initial processing feature map is transmitted to the first batch return
- the first layer 3 undergoes normalization processing, and after the normalization processing, it is sent to the first pooling layer 4 for pooling processing to obtain the first sampling feature map
- the first sampling feature map is sent to the second convolutional layer 5 for processing to obtain Secondary processing feature map
- the secondary processing feature map is sent to the second batch normalization layer 6 for normalization processing, and after the normalization processing is sent to the second pooling layer 7 for pooling processing, the secondary sampling feature is obtained Figure
- the sub-sampling feature map is sequentially sent to the first fully connected layer 8 and the second fully connected layer 9 for processing, and then sent to the output layer 10 after processing.
- the output layer 10 classifies and outputs the processed sub-sampling feature map Classification vector of damage location.
- the back propagation training step includes: the error value between the output result of the forward propagation training (ie, the classification vector) and the set output result is returned according to the reverse of the forward propagation training to obtain the error value of each layer; training;
- the cross entropy loss function is also used as the loss function.
- the Adam optimization algorithm is used to adjust the network parameters of the one-dimensional convolutional neural network model. Adam optimization algorithm is used in the entire one-dimensional neural network training process. It can iteratively update the neural network weights based on the training data so that the network converges or the error value reaches the specified iteration termination condition.
- Step 6 Input the test samples obtained in Step 4 into the one-dimensional convolutional neural network model trained in Step 5 to obtain the recognition result of the damage location.
- the above-mentioned structural damage identification method based on total empirical mode decomposition and convolutional neural network selects the acceleration response signal of the structure as the analysis object, and uses the total empirical mode decomposition technology to decompose and reconstruct the acceleration response signal, Only extracting information related to structural damage during reconstruction can reduce the impact of noise that is not related to structural damage, and improve the accuracy of damage recognition; at the same time, a one-dimensional convolutional neural network is used to extract features and reconstruct the reconstructed signal. Damage classification and damage identification are more efficient. Therefore, the above-mentioned structural damage identification method based on overall empirical mode decomposition and convolutional neural network provided by this application has significant advantages in real-time structural health monitoring.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Signal Processing (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Saccharide Compounds (AREA)
- Complex Calculations (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Image Analysis (AREA)
Abstract
Description
Claims (6)
- 基于总体经验模态分解与卷积神经网络的结构损伤识别法,其特征在于,包括以下步骤:步骤1:测得结构加速度响应信号x(t);步骤2:利用总体经验模态分解技术对加速度响应信号x(t)进行分解,得到一系列固有模态函数IMF;步骤3:对步骤2得到的固有模态函数IMF进行傅里叶变换得到其频谱特性,选择含有结构自振频率的固有模态函数IMF进行重构,得到重构信号z(t);步骤4:对步骤3得到的重构信号z(t)进行标准化处理,得到标准化重构信号Z(t);将标准化重构信号Z(t)分为多段长度一定的子信号,选取子信号中的部分信号作为训练样本,剩余部分信号作为测试样本;步骤5:将步骤4得到的训练样本输入一维卷积神经网络模型中进行训练,训练过程中采用交叉熵损失函数作为损失函数,采用Adam优化算法作为优化算法;步骤6:将步骤4得到的测试样本输入到步骤5训练好的一维卷积神经网络模型中,得到损伤位置的识别结果。
- 根据权利要求1所述的结构损伤识别法,其特征在于:所述步骤5中,所述一维卷积神经网络模型包括依次连接的输入层、第一卷积层、第一批量归一化层、第一池化层、第二卷积层、第二批量归一化层、第二池化层、第一全连接层、第二全连接层和输出层。
- 根据权利要求3所述的结构损伤识别法,其特征在于:所述第一卷积层和第二卷积层均含有16个大小为10×1的卷积核,所述第一池化层和第二池化层的池化窗口大小均为2×1。
- 根据权利要求3或4所述的结构损伤识别法,其特征在于,所述步骤5中,所述一维卷积神经网络模型的训练包括交替进行的前向传播训练和反向传播训练,直至网络收敛或所述前向传播训练的输出结果与设定的输出结果之间的误差值达到指定的迭代终止条件。
- 根据权利要求5所述的结构损伤识别法,其特征在于,所述前向传播训练步骤包括:将步骤4得到的训练样本从所述输入层中输入,所述第一卷积层对输入的训练样本进行处理得到初次处理特征图,所述初次处理特征图传送至所述第一批量归一化层进行归一化处理,归一化处理后传送至所述第一池化层进行池化处理,得到初次采样特征图;所述初次采样特征图传送至所述第二卷积层进行处理,得到二次处理特征图,所述二次处理特征图传送至所述第二批量归一化层进行归一化处理,归一化处理后传送至所述第二池化层进行池化处理,得到二次采样特征图;所述二次采样特征图依次传送至所述第一全连接层和第二全连接层进行处理,处理后传送至所述输出层,所述输出层对处理后的二次采样特征图进行分类并输出损伤位置的分类向量;所述反向传播训练步骤包括:将所述前向传播训练的输出结果与设定的输出结果的误差值按照所述前向传播训练的反向回传,得到每层的误差值,利用Adam优化算法调整所述一维卷积神经网络模型的网络参数。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2020214409A AU2020214409B2 (en) | 2019-01-31 | 2020-01-21 | Structural damage identification method based on ensemble empirical mode decomposition and convolution neural network |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910099951.1 | 2019-01-31 | ||
| CN201910099951.1A CN109946389B (zh) | 2019-01-31 | 2019-01-31 | 基于总体经验模态分解与卷积神经网络的结构损伤识别法 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020156348A1 true WO2020156348A1 (zh) | 2020-08-06 |
Family
ID=67007470
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2020/073357 Ceased WO2020156348A1 (zh) | 2019-01-31 | 2020-01-21 | 基于总体经验模态分解与卷积神经网络的结构损伤识别法 |
Country Status (3)
| Country | Link |
|---|---|
| CN (1) | CN109946389B (zh) |
| AU (1) | AU2020214409B2 (zh) |
| WO (1) | WO2020156348A1 (zh) |
Cited By (66)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111896625A (zh) * | 2020-08-17 | 2020-11-06 | 中南大学 | 钢轨伤损实时监测方法及其监测系统 |
| CN112284736A (zh) * | 2020-10-23 | 2021-01-29 | 天津大学 | 一种基于多通道注意力模块的卷积神经网络故障诊断方法 |
| CN112348255A (zh) * | 2020-11-06 | 2021-02-09 | 湖南大学 | 基于小波时频图像化的超短期风电功率预测方法 |
| CN112364710A (zh) * | 2020-10-20 | 2021-02-12 | 西安理工大学 | 一种基于深度学习算法的植物电信号分类识别方法 |
| CN112381770A (zh) * | 2020-11-03 | 2021-02-19 | 西安交通大学 | 一种基于融合卷积神经网络的磨损表面三维形貌测量方法 |
| CN112465027A (zh) * | 2020-11-27 | 2021-03-09 | 株洲时代电子技术有限公司 | 钢轨伤损检测方法 |
| CN112465225A (zh) * | 2020-11-27 | 2021-03-09 | 云南电网有限责任公司电力科学研究院 | 基于二次模态分解和级联深度学习的风电功率预测方法 |
| CN112528819A (zh) * | 2020-12-05 | 2021-03-19 | 西安电子科技大学 | 基于卷积神经网络的p300脑电信号分类方法 |
| CN112528820A (zh) * | 2020-12-03 | 2021-03-19 | 重庆邮智机器人研究院有限公司 | 一种基于emd和dcnn的运动想象脑电信号识别方法 |
| CN112580239A (zh) * | 2020-12-15 | 2021-03-30 | 中南大学 | 一种子结构响应重构方法及系统、存储介质 |
| CN112613105A (zh) * | 2020-12-25 | 2021-04-06 | 扬州大学 | 一种基于声发射参数与遗传算法优化bp神经网络的钢筋混凝土梁损伤状态识别方法 |
| CN112668526A (zh) * | 2020-12-31 | 2021-04-16 | 武汉地震工程研究院有限公司 | 基于深度学习和压电主动传感的螺栓群松动定位监测方法 |
| CN112668527A (zh) * | 2020-12-31 | 2021-04-16 | 华南理工大学 | 一种超声导波半监督成像检测方法 |
| CN112699792A (zh) * | 2020-12-29 | 2021-04-23 | 长安大学 | 基于正余弦算法优化卷积神经网络的机床轴承状态识别方法 |
| CN113155973A (zh) * | 2021-05-05 | 2021-07-23 | 温州大学 | 一种基于自适应奇异值分解的梁损伤识别方法 |
| CN113159179A (zh) * | 2021-04-22 | 2021-07-23 | 中车株洲电力机车有限公司 | 地铁及地铁转向架运行状态识别方法、系统 |
| CN113158769A (zh) * | 2021-03-03 | 2021-07-23 | 安徽大学 | 基于CEEMDAN和FastICA的机电设备轴承振动信号除噪方法 |
| CN113378426A (zh) * | 2021-05-05 | 2021-09-10 | 温州大学 | 一种基于信息融合的机械叶片损伤识别方法 |
| CN113591604A (zh) * | 2021-07-09 | 2021-11-02 | 浙江工业大学 | 基于噪声标签重构的风力发电机电能质量扰动状态监测方法 |
| CN113743297A (zh) * | 2021-09-03 | 2021-12-03 | 重庆大学 | 基于深度学习的储罐穹顶位移数据修复方法及装置 |
| CN113742983A (zh) * | 2021-10-09 | 2021-12-03 | 福州大学 | 基于深度自编码器神经网络的长跨结构损伤识别方法 |
| CN113837141A (zh) * | 2021-10-12 | 2021-12-24 | 国网山东省电力公司电力科学研究院 | 一种抗驱鼠器干扰的信号提取方法及装置 |
| CN114048781A (zh) * | 2021-11-15 | 2022-02-15 | 西安航空学院 | 一种水中目标辐射噪声识别方法及系统 |
| CN114065815A (zh) * | 2021-11-17 | 2022-02-18 | 国网山东省电力公司电力科学研究院 | 基于eemd与神经网络的变压器局部放电识别方法及系统 |
| CN114077849A (zh) * | 2021-11-17 | 2022-02-22 | 西安交通大学 | 一种基于部件级融合的发动机健康状态识别方法 |
| CN114169374A (zh) * | 2021-12-10 | 2022-03-11 | 湖南工商大学 | 一种斜拉桥斜拉索损伤识别方法及电子设备 |
| CN114202015A (zh) * | 2021-11-26 | 2022-03-18 | 中国科学院上海微系统与信息技术研究所 | 一种信号数据融合方法和识别方法 |
| CN114266280A (zh) * | 2021-12-29 | 2022-04-01 | 西安交通大学 | 一种基于改进卷积神经网络的螺栓松动状态识别方法 |
| CN114330636A (zh) * | 2021-06-04 | 2022-04-12 | 北京蜜蜂工物科技有限公司 | 非介入式电力负荷分解与识别方法、装置、电子设备及存储介质 |
| CN114324580A (zh) * | 2021-12-03 | 2022-04-12 | 西安交通大学 | 一种结构缺陷的智能敲击检测方法及系统 |
| CN114386452A (zh) * | 2021-12-06 | 2022-04-22 | 西安交通大学 | 核电循环水泵太阳轮故障检测方法 |
| CN114417681A (zh) * | 2022-04-01 | 2022-04-29 | 中国工程物理研究院流体物理研究所 | 基于动态决策和神经网络的二维结构变形监测方法及装置 |
| CN114510762A (zh) * | 2022-01-26 | 2022-05-17 | 中铁第四勘察设计院集团有限公司 | 基于时间序列模型系数灵敏度的结构损伤识别方法及系统 |
| CN114548174A (zh) * | 2022-02-23 | 2022-05-27 | 盐城工学院 | 一种用于滚动轴承故障特征提取的自适应变分模式分解法 |
| CN114580115A (zh) * | 2022-03-15 | 2022-06-03 | 中铁十四局集团大盾构工程有限公司 | 一种盾构机刀具状态分析与诊断方法 |
| CN114611560A (zh) * | 2022-03-18 | 2022-06-10 | 西安电子科技大学 | 基于卷积神经网络的ssvep脑电信号分类方法 |
| CN114742101A (zh) * | 2022-03-29 | 2022-07-12 | 哈尔滨工业大学 | 一种基于机器学习的自适应稀疏模态分解方法 |
| CN114821415A (zh) * | 2022-04-24 | 2022-07-29 | 中国水利水电科学研究院 | 基于机器学习的视频微振动放大校准和实时提取方法 |
| CN114818839A (zh) * | 2022-07-01 | 2022-07-29 | 之江实验室 | 一种基于深度学习的光纤传感水声信号识别方法及装置 |
| CN115048993A (zh) * | 2022-06-06 | 2022-09-13 | 兰州交通大学 | 一种基于传感器、结构部件的多维数据集构建方法及系统 |
| CN115236455A (zh) * | 2022-07-20 | 2022-10-25 | 广西大学 | 一种量子模糊分解图神经网络的临时接地线定位检测方法 |
| CN115393640A (zh) * | 2022-08-17 | 2022-11-25 | 四川大学 | 一种波形自适应与自平衡的损伤诊断方法 |
| CN115438769A (zh) * | 2021-06-04 | 2022-12-06 | 苏州蛟视智能科技有限公司 | 一种构建注意力机制网络的方法 |
| CN115455764A (zh) * | 2022-08-29 | 2022-12-09 | 武汉理工大学 | 一种框架结构多节点损伤诊断方法 |
| CN115470818A (zh) * | 2022-09-09 | 2022-12-13 | 安徽白鹭电子科技有限公司 | 基于变分模态分解的高阶调制信号样式识别方法及系统 |
| CN115619025A (zh) * | 2022-10-26 | 2023-01-17 | 安徽大学 | 基于meemd与最优组合集成的电力负荷三角模糊预测方法 |
| CN115684799A (zh) * | 2022-11-03 | 2023-02-03 | 哈尔滨工业大学 | 一种基于数据驱动的三相pfc转换器间歇故障检测方法 |
| CN115982568A (zh) * | 2022-12-22 | 2023-04-18 | 北京航空航天大学 | 一种基于卷积神经网络与集成学习的ads-b辐射源识别方法 |
| CN116208356A (zh) * | 2022-10-27 | 2023-06-02 | 浙江大学 | 一种基于深度学习的虚拟货币挖矿流量检测方法 |
| CN116226620A (zh) * | 2023-02-14 | 2023-06-06 | 哈尔滨理工大学 | 一种基于svm和lstm的次同步振荡模态辨识方法 |
| CN116244663A (zh) * | 2023-03-08 | 2023-06-09 | 北京工业大学 | 面向认知状态检测的多生理信号融合的分类方法 |
| CN116484171A (zh) * | 2022-01-07 | 2023-07-25 | 河海大学 | 一种基于改进的卷积神经网络的雷达干扰信号识别方法 |
| CN116559809A (zh) * | 2023-03-28 | 2023-08-08 | 南京桂瑞得信息科技有限公司 | 一种基于多源融合网络的雷达辐射源个体识别方法 |
| CN116595441A (zh) * | 2023-05-29 | 2023-08-15 | 中煤新集能源股份有限公司 | 一种基于神经网络算法的煤矿微震事件智能识别方法 |
| CN116597504A (zh) * | 2023-04-07 | 2023-08-15 | 中国科学院宁波材料技术与工程研究所 | 一种人类运动信息识别系统和方法 |
| CN116738288A (zh) * | 2023-06-14 | 2023-09-12 | 北京航空航天大学 | 基于单点振动响应的悬臂梁在线损伤识别方法 |
| CN116756515A (zh) * | 2023-07-07 | 2023-09-15 | 广州大学 | 基于深度卷积生成对抗网络的人工地震波生成方法 |
| CN116773679A (zh) * | 2023-06-19 | 2023-09-19 | 哈尔滨工程大学 | 一种基于神经网络的激光超声成像方法 |
| CN116842367A (zh) * | 2023-05-25 | 2023-10-03 | 中国石油大学(华东) | 基于小波包变换的管道弯折处损伤程度识别方法及系统 |
| CN116908312A (zh) * | 2023-04-14 | 2023-10-20 | 大连理工大学 | 基于超声导波和强泛化性胶囊网络模型的变温环境损伤定位方法、系统和存储介质 |
| CN117907445A (zh) * | 2024-01-13 | 2024-04-19 | 大连理工大学 | 基于超声导波和时空混合网络的复合材料加筋板损伤识别方法 |
| CN118312836A (zh) * | 2024-06-07 | 2024-07-09 | 山东科技大学 | 一种刚性罐道间螺栓连接状态检测方法及系统 |
| CN119046764A (zh) * | 2024-09-07 | 2024-11-29 | 国网福建省电力有限公司 | 一种基于同步相量量测数据的电能质量扰动识别方法 |
| CN120067535A (zh) * | 2025-02-07 | 2025-05-30 | 南京邮电大学 | 一种基于参量自信息的管道泄露振动信号降噪方法 |
| CN120105039A (zh) * | 2025-05-06 | 2025-06-06 | 汉江国家实验室 | 一种载流管路复模态参数识别方法及系统 |
| CN121253681A (zh) * | 2025-12-05 | 2026-01-02 | 天津滨海职业学院 | 一种建筑结构振动损伤在线监测方法 |
Families Citing this family (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109946389B (zh) * | 2019-01-31 | 2020-12-25 | 青岛理工大学 | 基于总体经验模态分解与卷积神经网络的结构损伤识别法 |
| CN110472272B (zh) * | 2019-07-01 | 2023-04-07 | 广东工业大学 | 一种基于多参数和卷积神经网络的结构损伤识别方法 |
| CN110490249B (zh) * | 2019-08-16 | 2022-06-07 | 哈尔滨工业大学 | 基于属性类别关系和少样本元学习的结构损伤识别方法 |
| CN111241725B (zh) * | 2019-12-30 | 2022-08-23 | 浙江大学 | 基于条件生成对抗网络的结构响应重构方法 |
| CN111220921A (zh) * | 2020-01-08 | 2020-06-02 | 重庆邮电大学 | 基于改进卷积-长短时记忆神经网络的锂电池容量估算方法 |
| CN111595541A (zh) * | 2020-04-27 | 2020-08-28 | 河海大学 | 一种海量振动传递率数据卷积神经网络处理的多维结构损伤识别方法 |
| CN111753776B (zh) * | 2020-06-29 | 2022-05-10 | 重庆交通大学 | 基于回声状态与多尺度卷积联合模型的结构损伤识别方法 |
| CN111931983B (zh) * | 2020-07-07 | 2023-09-19 | 湖南大学 | 一种降水量预测方法及系统 |
| CN112182490B (zh) * | 2020-09-01 | 2024-02-02 | 华中科技大学 | 一种电抗器状态诊断方法及系统 |
| CN112525467B (zh) * | 2020-11-26 | 2022-09-13 | 山东科技大学 | 一种适用于悬臂梁的冲击损伤区域识别方法及装置 |
| CN112434624A (zh) * | 2020-11-30 | 2021-03-02 | 广州大学 | 结构健康监测数据失真检测方法、系统、装置及存储介质 |
| CN112712060A (zh) * | 2021-01-15 | 2021-04-27 | 国网浙江省电力有限公司电力科学研究院 | 一种高压支柱瓷绝缘子声信号自动识别方法 |
| CN113159282B (zh) * | 2021-03-29 | 2022-04-22 | 东南大学 | 一种基于交叉模态置信准则矩阵的桥梁模态参数智能更新方法 |
| CN114002334A (zh) * | 2021-09-29 | 2022-02-01 | 西安交通大学 | 一种结构损伤声发射信号识别方法、装置、存储介质 |
| CN114062511B (zh) * | 2021-10-24 | 2024-09-17 | 北京化工大学 | 一种基于单传感器的航空发动机早期损伤声发射智能识别方法 |
| CN114660180A (zh) * | 2022-02-24 | 2022-06-24 | 郑州大学 | 一种基于声发射与1D CNNs的中小桥梁轻量化健康监测方法与系统 |
| CN114358214B (zh) * | 2022-03-11 | 2022-07-22 | 中国科学院上海高等研究院 | 步态自适应识别方法及装置、存储介质和终端 |
| CN115034254B (zh) * | 2022-03-22 | 2025-09-16 | 四川轻化工大学 | 基于hht频带能量特征和卷积神经网络的核素识别方法 |
| CN114878684A (zh) * | 2022-04-01 | 2022-08-09 | 中铁二十四局集团安徽工程有限公司 | 基于压电陶瓷超声的钢管混凝土拱肋密实度检测方法 |
| CN114462457B (zh) * | 2022-04-11 | 2022-07-08 | 自然资源部第一海洋研究所 | 基于本征概率密度函数的舰船水下噪声深度学习识别方法 |
| CN115169409B (zh) * | 2022-07-18 | 2023-05-09 | 四川省公路规划勘察设计研究院有限公司 | 基于滑窗的桥梁结构自振频率识别、预警方法及设备 |
| CN116244803B (zh) | 2023-02-22 | 2023-09-12 | 合肥工业大学 | 基于物理和数据双驱动的桥梁内部响应重构方法 |
| CN118535855B (zh) * | 2024-05-15 | 2025-01-28 | 北京市生态环境保护科学研究院 | 多源异构噪声治理方法及系统 |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103543026A (zh) * | 2013-10-23 | 2014-01-29 | 青岛理工大学 | 基于振动传递率函数和支持向量机的结构损伤识别方法 |
| US20150293143A1 (en) * | 2013-07-21 | 2015-10-15 | National Chiao Tung University | Feature Extraction from Human Gaiting Patterns using Principal Component Analysis and Multivariate Empirical Mode Decomposition |
| CN106096605A (zh) * | 2016-06-02 | 2016-11-09 | 史方 | 一种基于深度学习的图像模糊区域检测方法及装置 |
| CN107328868A (zh) * | 2017-06-27 | 2017-11-07 | 福州大学 | 一种快速识别陶瓷涂层失效类型的声发射信号分析方法 |
| CN107329933A (zh) * | 2017-07-14 | 2017-11-07 | 北京知觉科技有限公司 | 基于光纤传感振动信号的故障检测方法及装置 |
| CN109145961A (zh) * | 2018-07-30 | 2019-01-04 | 上海交通大学 | 一种非结构化局部放电数据的模式识别方法及系统 |
| CN109166100A (zh) * | 2018-07-24 | 2019-01-08 | 中南大学 | 基于卷积神经网络的多任务学习细胞计数方法 |
| US20190017911A1 (en) * | 2017-07-11 | 2019-01-17 | Qatar University | Real-time structural damage detection by convolutional neural networks |
| CN109946389A (zh) * | 2019-01-31 | 2019-06-28 | 青岛理工大学 | 基于总体经验模态分解与卷积神经网络的结构损伤识别法 |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201521676A (zh) * | 2013-12-13 | 2015-06-16 | 國家中山科學研究院 | 一種使用類神經網路產生判斷麻醉意識清醒程度指標的方法 |
| CN104502126B (zh) * | 2014-12-28 | 2017-08-15 | 华东交通大学 | 一种基于模态区间的高速列车转向架故障诊断方法 |
| CN106874957A (zh) * | 2017-02-27 | 2017-06-20 | 苏州大学 | 一种滚动轴承故障诊断方法 |
| CN107123033A (zh) * | 2017-05-04 | 2017-09-01 | 北京科技大学 | 一种基于深度卷积神经网络的服装搭配方法 |
| CN107505396A (zh) * | 2017-09-12 | 2017-12-22 | 燕山大学 | 一种结构损伤在线实时监测方法及系统 |
| CN107826919B (zh) * | 2017-10-20 | 2019-09-13 | 中国矿业大学 | 一种提升系统关键部件多状态健康监测装置及监测方法 |
| CN108717869B (zh) * | 2018-05-03 | 2021-08-13 | 中国石油大学(华东) | 基于卷积神经网络的糖尿病视网膜并发症诊断辅助系统 |
-
2019
- 2019-01-31 CN CN201910099951.1A patent/CN109946389B/zh active Active
-
2020
- 2020-01-21 AU AU2020214409A patent/AU2020214409B2/en not_active Ceased
- 2020-01-21 WO PCT/CN2020/073357 patent/WO2020156348A1/zh not_active Ceased
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150293143A1 (en) * | 2013-07-21 | 2015-10-15 | National Chiao Tung University | Feature Extraction from Human Gaiting Patterns using Principal Component Analysis and Multivariate Empirical Mode Decomposition |
| CN103543026A (zh) * | 2013-10-23 | 2014-01-29 | 青岛理工大学 | 基于振动传递率函数和支持向量机的结构损伤识别方法 |
| CN106096605A (zh) * | 2016-06-02 | 2016-11-09 | 史方 | 一种基于深度学习的图像模糊区域检测方法及装置 |
| CN107328868A (zh) * | 2017-06-27 | 2017-11-07 | 福州大学 | 一种快速识别陶瓷涂层失效类型的声发射信号分析方法 |
| US20190017911A1 (en) * | 2017-07-11 | 2019-01-17 | Qatar University | Real-time structural damage detection by convolutional neural networks |
| CN107329933A (zh) * | 2017-07-14 | 2017-11-07 | 北京知觉科技有限公司 | 基于光纤传感振动信号的故障检测方法及装置 |
| CN109166100A (zh) * | 2018-07-24 | 2019-01-08 | 中南大学 | 基于卷积神经网络的多任务学习细胞计数方法 |
| CN109145961A (zh) * | 2018-07-30 | 2019-01-04 | 上海交通大学 | 一种非结构化局部放电数据的模式识别方法及系统 |
| CN109946389A (zh) * | 2019-01-31 | 2019-06-28 | 青岛理工大学 | 基于总体经验模态分解与卷积神经网络的结构损伤识别法 |
Cited By (90)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111896625B (zh) * | 2020-08-17 | 2023-07-14 | 中南大学 | 钢轨伤损实时监测方法及其监测系统 |
| CN111896625A (zh) * | 2020-08-17 | 2020-11-06 | 中南大学 | 钢轨伤损实时监测方法及其监测系统 |
| CN112364710A (zh) * | 2020-10-20 | 2021-02-12 | 西安理工大学 | 一种基于深度学习算法的植物电信号分类识别方法 |
| CN112364710B (zh) * | 2020-10-20 | 2024-04-05 | 西安理工大学 | 一种基于深度学习算法的植物电信号分类识别方法 |
| CN112284736A (zh) * | 2020-10-23 | 2021-01-29 | 天津大学 | 一种基于多通道注意力模块的卷积神经网络故障诊断方法 |
| CN112381770B (zh) * | 2020-11-03 | 2024-04-05 | 西安交通大学 | 一种基于融合卷积神经网络的磨损表面三维形貌测量方法 |
| CN112381770A (zh) * | 2020-11-03 | 2021-02-19 | 西安交通大学 | 一种基于融合卷积神经网络的磨损表面三维形貌测量方法 |
| CN112348255B (zh) * | 2020-11-06 | 2024-04-09 | 湖南大学 | 基于小波时频图像化的超短期风电功率预测方法 |
| CN112348255A (zh) * | 2020-11-06 | 2021-02-09 | 湖南大学 | 基于小波时频图像化的超短期风电功率预测方法 |
| CN112465225A (zh) * | 2020-11-27 | 2021-03-09 | 云南电网有限责任公司电力科学研究院 | 基于二次模态分解和级联深度学习的风电功率预测方法 |
| CN112465027A (zh) * | 2020-11-27 | 2021-03-09 | 株洲时代电子技术有限公司 | 钢轨伤损检测方法 |
| CN112528820A (zh) * | 2020-12-03 | 2021-03-19 | 重庆邮智机器人研究院有限公司 | 一种基于emd和dcnn的运动想象脑电信号识别方法 |
| CN112528819B (zh) * | 2020-12-05 | 2023-01-20 | 西安电子科技大学 | 基于卷积神经网络的p300脑电信号分类方法 |
| CN112528819A (zh) * | 2020-12-05 | 2021-03-19 | 西安电子科技大学 | 基于卷积神经网络的p300脑电信号分类方法 |
| CN112580239A (zh) * | 2020-12-15 | 2021-03-30 | 中南大学 | 一种子结构响应重构方法及系统、存储介质 |
| CN112580239B (zh) * | 2020-12-15 | 2022-07-05 | 中南大学 | 一种子结构响应重构方法及系统、存储介质 |
| CN112613105B (zh) * | 2020-12-25 | 2024-03-08 | 扬州大学 | 一种基于声发射参数与遗传算法优化bp神经网络的钢筋混凝土梁损伤状态识别方法 |
| CN112613105A (zh) * | 2020-12-25 | 2021-04-06 | 扬州大学 | 一种基于声发射参数与遗传算法优化bp神经网络的钢筋混凝土梁损伤状态识别方法 |
| CN112699792A (zh) * | 2020-12-29 | 2021-04-23 | 长安大学 | 基于正余弦算法优化卷积神经网络的机床轴承状态识别方法 |
| CN112699792B (zh) * | 2020-12-29 | 2024-04-05 | 长安大学 | 基于正余弦算法优化卷积神经网络的机床轴承状态识别方法 |
| CN112668526B (zh) * | 2020-12-31 | 2025-01-24 | 武汉地震工程研究院有限公司 | 基于深度学习和压电主动传感的螺栓群松动定位监测方法 |
| CN112668526A (zh) * | 2020-12-31 | 2021-04-16 | 武汉地震工程研究院有限公司 | 基于深度学习和压电主动传感的螺栓群松动定位监测方法 |
| CN112668527A (zh) * | 2020-12-31 | 2021-04-16 | 华南理工大学 | 一种超声导波半监督成像检测方法 |
| CN112668527B (zh) * | 2020-12-31 | 2023-04-25 | 华南理工大学 | 一种超声导波半监督成像检测方法 |
| CN113158769A (zh) * | 2021-03-03 | 2021-07-23 | 安徽大学 | 基于CEEMDAN和FastICA的机电设备轴承振动信号除噪方法 |
| CN113159179B (zh) * | 2021-04-22 | 2023-04-18 | 中车株洲电力机车有限公司 | 地铁及地铁转向架运行状态识别方法、系统 |
| CN113159179A (zh) * | 2021-04-22 | 2021-07-23 | 中车株洲电力机车有限公司 | 地铁及地铁转向架运行状态识别方法、系统 |
| CN113155973A (zh) * | 2021-05-05 | 2021-07-23 | 温州大学 | 一种基于自适应奇异值分解的梁损伤识别方法 |
| CN113378426B (zh) * | 2021-05-05 | 2023-07-14 | 温州大学 | 一种基于信息融合的机械叶片损伤识别方法 |
| CN113378426A (zh) * | 2021-05-05 | 2021-09-10 | 温州大学 | 一种基于信息融合的机械叶片损伤识别方法 |
| CN114330636A (zh) * | 2021-06-04 | 2022-04-12 | 北京蜜蜂工物科技有限公司 | 非介入式电力负荷分解与识别方法、装置、电子设备及存储介质 |
| CN115438769A (zh) * | 2021-06-04 | 2022-12-06 | 苏州蛟视智能科技有限公司 | 一种构建注意力机制网络的方法 |
| CN113591604A (zh) * | 2021-07-09 | 2021-11-02 | 浙江工业大学 | 基于噪声标签重构的风力发电机电能质量扰动状态监测方法 |
| CN113743297A (zh) * | 2021-09-03 | 2021-12-03 | 重庆大学 | 基于深度学习的储罐穹顶位移数据修复方法及装置 |
| CN113742983A (zh) * | 2021-10-09 | 2021-12-03 | 福州大学 | 基于深度自编码器神经网络的长跨结构损伤识别方法 |
| CN113837141B (zh) * | 2021-10-12 | 2023-10-27 | 国网山东省电力公司电力科学研究院 | 一种抗驱鼠器干扰的信号提取方法及装置 |
| CN113837141A (zh) * | 2021-10-12 | 2021-12-24 | 国网山东省电力公司电力科学研究院 | 一种抗驱鼠器干扰的信号提取方法及装置 |
| CN114048781A (zh) * | 2021-11-15 | 2022-02-15 | 西安航空学院 | 一种水中目标辐射噪声识别方法及系统 |
| CN114077849B (zh) * | 2021-11-17 | 2024-04-05 | 西安交通大学 | 一种基于部件级融合的发动机健康状态识别方法 |
| CN114065815A (zh) * | 2021-11-17 | 2022-02-18 | 国网山东省电力公司电力科学研究院 | 基于eemd与神经网络的变压器局部放电识别方法及系统 |
| CN114077849A (zh) * | 2021-11-17 | 2022-02-22 | 西安交通大学 | 一种基于部件级融合的发动机健康状态识别方法 |
| CN114202015A (zh) * | 2021-11-26 | 2022-03-18 | 中国科学院上海微系统与信息技术研究所 | 一种信号数据融合方法和识别方法 |
| CN114324580A (zh) * | 2021-12-03 | 2022-04-12 | 西安交通大学 | 一种结构缺陷的智能敲击检测方法及系统 |
| CN114386452A (zh) * | 2021-12-06 | 2022-04-22 | 西安交通大学 | 核电循环水泵太阳轮故障检测方法 |
| CN114386452B (zh) * | 2021-12-06 | 2024-04-05 | 西安交通大学 | 核电循环水泵太阳轮故障检测方法 |
| CN114169374B (zh) * | 2021-12-10 | 2024-02-20 | 湖南工商大学 | 一种斜拉桥斜拉索损伤识别方法及电子设备 |
| CN114169374A (zh) * | 2021-12-10 | 2022-03-11 | 湖南工商大学 | 一种斜拉桥斜拉索损伤识别方法及电子设备 |
| CN114266280A (zh) * | 2021-12-29 | 2022-04-01 | 西安交通大学 | 一种基于改进卷积神经网络的螺栓松动状态识别方法 |
| CN114266280B (zh) * | 2021-12-29 | 2023-10-13 | 西安交通大学 | 一种基于改进卷积神经网络的螺栓松动状态识别方法 |
| CN116484171A (zh) * | 2022-01-07 | 2023-07-25 | 河海大学 | 一种基于改进的卷积神经网络的雷达干扰信号识别方法 |
| CN114510762A (zh) * | 2022-01-26 | 2022-05-17 | 中铁第四勘察设计院集团有限公司 | 基于时间序列模型系数灵敏度的结构损伤识别方法及系统 |
| CN114548174B (zh) * | 2022-02-23 | 2025-05-02 | 盐城工学院 | 一种用于滚动轴承故障特征提取的自适应变分模式分解法 |
| CN114548174A (zh) * | 2022-02-23 | 2022-05-27 | 盐城工学院 | 一种用于滚动轴承故障特征提取的自适应变分模式分解法 |
| CN114580115A (zh) * | 2022-03-15 | 2022-06-03 | 中铁十四局集团大盾构工程有限公司 | 一种盾构机刀具状态分析与诊断方法 |
| CN114611560A (zh) * | 2022-03-18 | 2022-06-10 | 西安电子科技大学 | 基于卷积神经网络的ssvep脑电信号分类方法 |
| CN114742101A (zh) * | 2022-03-29 | 2022-07-12 | 哈尔滨工业大学 | 一种基于机器学习的自适应稀疏模态分解方法 |
| CN114417681A (zh) * | 2022-04-01 | 2022-04-29 | 中国工程物理研究院流体物理研究所 | 基于动态决策和神经网络的二维结构变形监测方法及装置 |
| CN114821415B (zh) * | 2022-04-24 | 2022-11-29 | 中国水利水电科学研究院 | 基于机器学习的视频微振动放大校准和实时提取方法 |
| CN114821415A (zh) * | 2022-04-24 | 2022-07-29 | 中国水利水电科学研究院 | 基于机器学习的视频微振动放大校准和实时提取方法 |
| CN115048993A (zh) * | 2022-06-06 | 2022-09-13 | 兰州交通大学 | 一种基于传感器、结构部件的多维数据集构建方法及系统 |
| CN114818839A (zh) * | 2022-07-01 | 2022-07-29 | 之江实验室 | 一种基于深度学习的光纤传感水声信号识别方法及装置 |
| CN114818839B (zh) * | 2022-07-01 | 2022-09-16 | 之江实验室 | 一种基于深度学习的光纤传感水声信号识别方法及装置 |
| CN115236455A (zh) * | 2022-07-20 | 2022-10-25 | 广西大学 | 一种量子模糊分解图神经网络的临时接地线定位检测方法 |
| CN115393640A (zh) * | 2022-08-17 | 2022-11-25 | 四川大学 | 一种波形自适应与自平衡的损伤诊断方法 |
| CN115455764A (zh) * | 2022-08-29 | 2022-12-09 | 武汉理工大学 | 一种框架结构多节点损伤诊断方法 |
| CN115470818A (zh) * | 2022-09-09 | 2022-12-13 | 安徽白鹭电子科技有限公司 | 基于变分模态分解的高阶调制信号样式识别方法及系统 |
| CN115619025A (zh) * | 2022-10-26 | 2023-01-17 | 安徽大学 | 基于meemd与最优组合集成的电力负荷三角模糊预测方法 |
| CN115619025B (zh) * | 2022-10-26 | 2026-03-10 | 安徽大学 | 基于meemd与最优组合集成的电力负荷三角模糊预测方法 |
| CN116208356B (zh) * | 2022-10-27 | 2023-09-29 | 浙江大学 | 一种基于深度学习的虚拟货币挖矿流量检测方法 |
| CN116208356A (zh) * | 2022-10-27 | 2023-06-02 | 浙江大学 | 一种基于深度学习的虚拟货币挖矿流量检测方法 |
| CN115684799A (zh) * | 2022-11-03 | 2023-02-03 | 哈尔滨工业大学 | 一种基于数据驱动的三相pfc转换器间歇故障检测方法 |
| CN115982568A (zh) * | 2022-12-22 | 2023-04-18 | 北京航空航天大学 | 一种基于卷积神经网络与集成学习的ads-b辐射源识别方法 |
| CN116226620A (zh) * | 2023-02-14 | 2023-06-06 | 哈尔滨理工大学 | 一种基于svm和lstm的次同步振荡模态辨识方法 |
| CN116244663A (zh) * | 2023-03-08 | 2023-06-09 | 北京工业大学 | 面向认知状态检测的多生理信号融合的分类方法 |
| CN116559809A (zh) * | 2023-03-28 | 2023-08-08 | 南京桂瑞得信息科技有限公司 | 一种基于多源融合网络的雷达辐射源个体识别方法 |
| CN116597504A (zh) * | 2023-04-07 | 2023-08-15 | 中国科学院宁波材料技术与工程研究所 | 一种人类运动信息识别系统和方法 |
| CN116908312A (zh) * | 2023-04-14 | 2023-10-20 | 大连理工大学 | 基于超声导波和强泛化性胶囊网络模型的变温环境损伤定位方法、系统和存储介质 |
| CN116842367A (zh) * | 2023-05-25 | 2023-10-03 | 中国石油大学(华东) | 基于小波包变换的管道弯折处损伤程度识别方法及系统 |
| CN116595441A (zh) * | 2023-05-29 | 2023-08-15 | 中煤新集能源股份有限公司 | 一种基于神经网络算法的煤矿微震事件智能识别方法 |
| CN116738288A (zh) * | 2023-06-14 | 2023-09-12 | 北京航空航天大学 | 基于单点振动响应的悬臂梁在线损伤识别方法 |
| CN116773679A (zh) * | 2023-06-19 | 2023-09-19 | 哈尔滨工程大学 | 一种基于神经网络的激光超声成像方法 |
| CN116773679B (zh) * | 2023-06-19 | 2026-04-07 | 哈尔滨工程大学 | 一种基于神经网络的激光超声成像方法 |
| CN116756515B (zh) * | 2023-07-07 | 2024-02-23 | 广州大学 | 基于深度卷积生成对抗网络的人工地震波生成方法 |
| CN116756515A (zh) * | 2023-07-07 | 2023-09-15 | 广州大学 | 基于深度卷积生成对抗网络的人工地震波生成方法 |
| CN117907445A (zh) * | 2024-01-13 | 2024-04-19 | 大连理工大学 | 基于超声导波和时空混合网络的复合材料加筋板损伤识别方法 |
| CN118312836A (zh) * | 2024-06-07 | 2024-07-09 | 山东科技大学 | 一种刚性罐道间螺栓连接状态检测方法及系统 |
| CN119046764A (zh) * | 2024-09-07 | 2024-11-29 | 国网福建省电力有限公司 | 一种基于同步相量量测数据的电能质量扰动识别方法 |
| CN120067535A (zh) * | 2025-02-07 | 2025-05-30 | 南京邮电大学 | 一种基于参量自信息的管道泄露振动信号降噪方法 |
| CN120105039A (zh) * | 2025-05-06 | 2025-06-06 | 汉江国家实验室 | 一种载流管路复模态参数识别方法及系统 |
| CN121253681A (zh) * | 2025-12-05 | 2026-01-02 | 天津滨海职业学院 | 一种建筑结构振动损伤在线监测方法 |
Also Published As
| Publication number | Publication date |
|---|---|
| AU2020214409B2 (en) | 2022-07-28 |
| CN109946389A (zh) | 2019-06-28 |
| CN109946389B (zh) | 2020-12-25 |
| AU2020214409A1 (en) | 2021-03-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020156348A1 (zh) | 基于总体经验模态分解与卷积神经网络的结构损伤识别法 | |
| CN112149316A (zh) | 基于改进的cnn模型的航空发动机剩余寿命预测方法 | |
| GB2625165A (en) | Method and system for fault diagnosis of rolling bearing | |
| CN112946600B (zh) | 基于wgan-gp的雷达hrrp数据库构建方法 | |
| CN114547795B (zh) | 一种基于数据驱动的滚动轴承剩余寿命预测方法 | |
| CN111860982A (zh) | 一种基于vmd-fcm-gru的风电场短期风电功率预测方法 | |
| CN107421741A (zh) | 一种基于卷积神经网络的滚动轴承故障诊断方法 | |
| CN111914703A (zh) | 一种基于小波变换和迁移学习GoogLeNet的机械旋转部件故障诊断方法 | |
| CN111257934A (zh) | 基于二阶神经元深度神经网络的地震动峰值加速度预测方法 | |
| CN109633270B (zh) | 故障能量区域边界识别及特征提取方法 | |
| CN111783236B (zh) | 基于自适应模型和子集模拟的涡轮机匣灵敏度分析方法 | |
| CN110825583B (zh) | 一种针对云数据中心多指标融合的能效定性评估技术 | |
| CN117647586B (zh) | 一种煤岩破坏动态监测与识别方法和装置 | |
| CN111580151A (zh) | 一种基于SSNet模型的地震事件到时识别方法 | |
| CN113761777A (zh) | 一种基于hp-ovmd的超短期光伏功率预测方法 | |
| CN111753776B (zh) | 基于回声状态与多尺度卷积联合模型的结构损伤识别方法 | |
| CN114239759A (zh) | 电动汽车的故障诊断方法、装置、电子设备及存储介质 | |
| CN120086668A (zh) | 一种基于时频增强CNN-Transformer的轴承故障诊断方法 | |
| CN118349889A (zh) | 一种风电齿轮箱智能故障诊断方法及设备 | |
| CN114330439A (zh) | 一种基于卷积神经网络的轴承诊断方法 | |
| CN113591625B (zh) | 一种不平衡样本重要性加权的齿轮箱故障诊断方法 | |
| CN114912674A (zh) | 基于全日面光球磁图的太阳射电爆发预报方法及系统 | |
| CN115932946A (zh) | 震级估算模型的训练方法、震级估算方法及震级估算模型 | |
| CN115204031A (zh) | 航空发动机装配过程的振动值预测方法 | |
| CN118817310A (zh) | 一种基于at-stnn的涡扇发动机轴承故障识别方法 |
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: 20748478 Country of ref document: EP Kind code of ref document: A1 |
|
| ENP | Entry into the national phase |
Ref document number: 2020214409 Country of ref document: AU Date of ref document: 20200121 Kind code of ref document: A |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 20748478 Country of ref document: EP Kind code of ref document: A1 |
