CN114048790A - Road base layer strain analysis method based on coupling bidirectional LSTM and convolution structure network - Google Patents
Road base layer strain analysis method based on coupling bidirectional LSTM and convolution structure network Download PDFInfo
- Publication number
- CN114048790A CN114048790A CN202110877678.8A CN202110877678A CN114048790A CN 114048790 A CN114048790 A CN 114048790A CN 202110877678 A CN202110877678 A CN 202110877678A CN 114048790 A CN114048790 A CN 114048790A
- Authority
- CN
- China
- Prior art keywords
- data
- layer
- road
- strain
- convolution
- 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.)
- Granted
Links
- 230000002457 bidirectional effect Effects 0.000 title claims abstract description 23
- 238000004458 analytical method Methods 0.000 title claims abstract description 20
- 230000008878 coupling Effects 0.000 title abstract description 6
- 238000010168 coupling process Methods 0.000 title abstract description 6
- 238000005859 coupling reaction Methods 0.000 title abstract description 6
- 238000012544 monitoring process Methods 0.000 claims abstract description 43
- 238000012549 training Methods 0.000 claims abstract description 22
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 239000010410 layer Substances 0.000 claims description 74
- 238000000034 method Methods 0.000 claims description 31
- 239000002689 soil Substances 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 10
- 238000010219 correlation analysis Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000012935 Averaging Methods 0.000 claims description 4
- 239000010426 asphalt Substances 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 4
- 230000003204 osmotic effect Effects 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 239000013589 supplement Substances 0.000 claims description 2
- 239000002344 surface layer Substances 0.000 claims description 2
- 230000007774 longterm Effects 0.000 abstract description 5
- 230000009467 reduction Effects 0.000 abstract description 5
- 238000012423 maintenance Methods 0.000 abstract description 4
- 238000013508 migration Methods 0.000 abstract description 2
- 238000010220 Pearson correlation analysis Methods 0.000 abstract 1
- 230000003446 memory effect Effects 0.000 abstract 1
- 238000011160 research Methods 0.000 abstract 1
- 238000013527 convolutional neural network Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000000875 corresponding effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a road base strain analysis method based on a coupling bidirectional LSTM and a convolution structure network, which comprises the steps of firstly preprocessing long-term monitoring data acquired by an automatic road monitoring system, on the basis, the Pearson correlation analysis is adopted to research the relationship among the data so as to achieve the purpose of dimension reduction of the input data, thereby reducing training errors, improving the prediction accuracy of the model, finally selecting data in a proper time period and inputting the data into the BilSTM-CNN network model for training, learning the physical characteristics among sensor data, adopting a training mode of weight self-migration aiming at the problem that each time period is not continuous, the historical information characteristics are relatively more obvious in the next time period, the model is easier to learn, the memory effect of the model is enhanced, overfitting is reduced, and the generalization capability of the model is improved. In addition, the invention can save labor and time cost and lay a foundation for subsequent road maintenance work.
Description
Technical Field
The invention belongs to the field of time sequence data analysis, and relates to a road base layer strain analysis method based on a coupled bidirectional LSTM and a convolution structure network. The method is applied to road base strain analysis by using the road surface structure monitoring data acquired by the road monitoring system with multi-sensor fusion.
Background
In recent years, as the road network of China has been basically formed, the road capital construction investment in the future will gradually point to the road maintenance management business. The safety and the comfort of driving are influenced to a certain extent by diseases such as cracks, ruts and the like, and the pavement needs to be maintained in time. Therefore, the accurate basic strain analysis result can avoid the road from serious structural damage in the future to the maximum extent, and provide reliable and effective technical support for subsequent road maintenance management work.
At present, the traditional basic layer strain prediction analysis method is generally used for analyzing and calculating stress and strain data based on finite element simulation, mechanics-experience models, mathematical statistics methods and the like, and the traditional methods usually carry out data comparison to correct model parameters and simplify a three-dimensional model, so that the analysis result is influenced to a certain extent, and the analysis process is complicated and time-consuming. The machine learning method is gradually applied to the field of monitoring big data analysis due to the development of the artificial intelligence technology, can relatively quickly and efficiently analyze and predict the obtained data, saves time, reduces cost, and obtains accurate results.
At present, the road base layer strain prediction analysis based on deep learning often faces the problems of large dimensionality of input data, discontinuous time of monitoring data and the like, data correlation analysis is carried out to carry out data dimensionality reduction on an input end, considerable performance improvement can be brought by weight self-migration in the training process, the accuracy and the stability are improved, and overfitting is reduced.
Therefore, the invention provides a road base layer strain analysis method based on coupled bidirectional LSTM and convolution structure network (BiLSTM-CNN). The method comprises the steps of firstly preprocessing sensor data collected by a road monitoring system to obtain a data set which can be input into a network for training, then analyzing the characteristic correlation of the data set to obtain a Pearson correlation coefficient, removing the characteristic data with smaller correlation coefficient to achieve the purpose of data dimension reduction, and finally training the BiLSTM-CNN to the final data set, thereby carrying out predictive analysis on the strain data of a road base layer to know the strain change trend of the road base layer in the future in advance and provide help for the subsequent road maintenance decision.
Disclosure of Invention
The invention aims to analyze pavement monitoring data which are acquired by a sensor and are up to eight years from 2012 to 2020 through a road base layer strain prediction analysis method based on a coupled bidirectional LSTM and a convolution structure network, wherein the long-term monitoring data comprise asphalt strain, base layer three-dimensional strain, soil layer strain, soil pressure, temperature, osmotic pressure and soil moisture, so that the base layer strain is predicted according to proper influence factors, and the base layer strain change trend in a period of time in the future is acquired.
One, coupling two-way LSTM and convolution structure network
The coupling Bidirectional LSTM and convolution structure network (BiLSTM-CNN) adopted by the invention is composed of a Bidirectional LSTM layer and four 1D convolution layers, as shown in figure 1. The method comprises the steps that fixed features on a data time dimension acquired by a CNN extraction sensor are utilized to realize feature extraction of monitoring signals of various sensors on a road; meanwhile, a bidirectional LSTM neural network is selected to process the characteristic sequence output by the CNN part, time sequence information in data is mined, the signal variation trend and the characteristics of the monitoring sensor are extracted in a time dimension, and the prediction of roadbed base layer strain is realized.
The technical scheme adopted by the invention is a road base layer strain prediction analysis method based on a coupled bidirectional LSTM and convolution structure network, which comprises four parts of original monitoring data preprocessing, characteristic correlation analysis, data set manufacturing and BiLSTM-CNN network base layer strain prediction, as shown in figure 2, and comprises the following specific steps:
the method comprises the following steps: preprocessing original monitoring data;
first, the road monitoring data acquired by various sensors are integrated.
And secondly, according to the time span of the data acquired by the road monitoring system, deleting the data repetition value and the jump value after selecting the data in a proper time period, and completing the supplement of the missing value to prevent poor training effect of the monitored data.
And finally, according to the characteristics of the time sequence data, unifying the time intervals among all the data points to be 1 hour.
Step two: analyzing the characteristic correlation;
and step one, carrying out correlation analysis on the first three data points acquired by each sensor after processing, and measuring the correlation between the characteristics according to a Pearson correlation coefficient. In the second step, soil layer strain is not much correlated and removed.
Step three: making a data set;
and averaging all monitoring data points of asphalt strain, osmotic pressure, soil moisture and base layer three-dimensional strain, and averaging the three parts of the surface layer, the base layer and the soil foundation according to the embedded position of the temperature sensor. And finally, the ratio of the monitoring data set to the training set to the testing set is about 8: and 2, dividing.
Step four: predicting strain of a BiLSTM-CNN network base layer;
the BilSTM-CNN network is composed of 4 layers of one-dimensional convolution layers, one layer of bidirectional LSTM and two layers of full connection layers. The convolution kernel size of the one-dimensional convolution layer is 2, the number of the convolution kernels is 18, 36, 72 and 144 in sequence, each convolution layer is followed by a pooling layer with the size of 2 and the step length of 1, the number of the bidirectional LSTM units is 256, each convolution layer uses a ReLU activation function, and the number of the units of the last full-connection layer is 95. And taking the preprocessed and dimensionality-reduced data set as an input of a time sequence convolution network, and setting the time step length to be 8.
According to the invention, long-term and multidimensional road monitoring data can be analyzed by using the road sensor monitoring data acquired by an automatic road monitoring system for a long time through a deep learning method, so that the prediction analysis is carried out on the strain of a road base layer, and meanwhile, the learning weight after each time period is trained is transferred to the training of the next time period, so that the accuracy and the stability of a prediction model are improved, and overfitting is reduced. In addition, the invention can effectively reduce labor cost and time loss, and the processed data can lay a foundation for the subsequent analysis and mining of monitoring data and the training of prediction models.
Drawings
FIG. 1 is a schematic diagram of a time-series convolutional network structure
FIG. 2 is a diagram of the steps of a method
FIG. 3 is a correlation analysis heatmap
FIG. 4 is a depiction of an input data set
FIG. 5 is a diagram of the predicted effect of coupled bidirectional LSTM and CNN networks
Detailed Description
The original pavement sensor monitoring data set adopted by the invention is data acquired by an automatic road monitoring system. The specific implementation steps are as follows:
(1) raw monitoring data preprocessing
Due to the problems of data loss during monitoring data transmission, temporary faults of acquired equipment and other systematic errors, the problems of repetition, loss, jump and the like of original data can cause great influence on the network prediction precision. Therefore, the original data acquired by the sensor is deleted from the duplicate value and the jump value, and the missing value is supplemented. In order to ensure the time continuity and consistency of the monitoring data, the data of nine proper time periods are selected, and the time intervals are unified into one hour, so that the characteristics of time sequence data are better met, and the model training is facilitated.
(2) Feature correlation analysis
The data monitored by the three-way strain gauge positioned on the base layer part is an actual measurement value of predicted target data, namely a base layer strain value, all data obtained by the monitoring system comprise a plurality of characteristics, and the exploration of the relation between the characteristics and the correlation degree with the predicted target is critical. The method is primarily conceived to predict the strain of the base layer through a plurality of factors of asphalt layer strain, soil pressure, osmotic pressure, water content, temperature and soil layer strain, and the correct reduction of some characteristic quantities can provide certain help for model operation, such as improvement of precision, reduction of risk of overfitting and the like. Therefore, correlation analysis of these data characteristics is necessary between the input data fed into the model. Firstly, because the dimension of monitoring data acquired by each sensor is huge, only the data of the first three monitoring positions acquired by each sensor is selected for correlation analysis; the correlation between features was then measured using pearson correlation coefficients to generate a correlation analysis heatmap, as shown in fig. 3. The more the pearson correlation coefficient value approaches 1 or-1, the more positive and negative correlations are present. Finally, according to the analysis result, the soil layer strain is removed from the input end, and the correlation between the soil layer strain and other characteristics is small.
(3) Data set production
In order to meet the requirements of supervised learning, the proportion of a road surface monitoring data set to a training set and a testing set is about 8: and 2, dividing. The input data set description is shown in fig. 4. In order to monitor the training effect of the model during the training process, 20% of the training set is divided into verification sets.
(5) BiLSTM-CNN network base layer strain prediction analysis
The BiLSTM-CNN network adopted by the invention utilizes a one-dimensional convolution structure thereof to extract invariant characteristics of monitoring data in a time dimension; the bidirectional LSTM structure is used for capturing bidirectional semantic dependence of historical data and learning mapping relation among features, the data can be inverted, and the hidden layer synthesizes forward and backward information, so that cells in a network can obtain context information at the same time. The invention combines the convolution layer with the extracted fixed property and the bidirectional LSTM, thereby better retaining the long-term dependence relationship and the historical characteristic information, improving the model training effect and improving the robustness and the precision of the prediction model.
The BilSTM-CNN network is composed of 4 layers of one-dimensional convolution layers, one layer of bidirectional LSTM and two layers of full connection layers, and has 7 layers in total. The sizes of the one-dimensional convolution kernels are all 2, the number of the one-dimensional convolution kernels is 18, 36, 72 and 144 in sequence, each layer uses a ReLU activation function, each one-dimensional convolution layer is connected with a maximum pooling layer with the size of 2 and the step length of 1, and a bidirectional LSTM layer with the unit number of 256 is connected with the four one-dimensional convolution layers. At this time, the input data is subjected to one-dimensional convolution of each layer, then the internal fixed features of the input data are extracted by the network, each layer is subjected to a ReLU activation function, then is subjected to an UpSampling layer (UpSampling) with the step of 2, and then the long-term dependency and the historical information of the data context are learned in the bidirectional LSTM structure. And finally, acquiring the prediction information at the full connection layer, and taking the prediction information sequentially acquired as output at the output end. The final predicted effect is shown in fig. 5.
The ReLU function used in the present invention is represented as follows:
in the process of training the road surface monitoring data by the coupling bidirectional LSTM and the convolution structure network, input multidimensional data and target data are trained together, the physical relation between data characteristics and data is continuously learned from the training, and the target is optimized to minimize the error between a predicted value and an actually measured value. An adaptive Learning Rate adjusting algorithm adadelta (adaptive Learning Rate method) is taken as an optimization algorithm for gradient descent in the back propagation process, and the algorithm has the advantage of being capable of adaptively adjusting the Learning Rate in the gradient descent without manual setting.
For the learning method, an Adam method is used for optimizing the model parameters, and the Adam method is a simple and high-calculation-efficiency random objective function gradient optimization algorithm. The method has two advantages in the aspects of processing sparse gradients and processing non-stationary targets. Adam is used in the present invention because it can be well adapted to a wide range of non-convex optimization problems.
Adam maintains the past trend of the mean squared gradient vt to decay exponentially. It also has an average of past gradients mt with an exponential decay trend and a flat minimum preference in the error plane. Then, the past attenuation mean and the past squared gradient m are calculatedtAnd vtThe corresponding is as follows:
mt＝β1mt-1+(1-β1)gt (2)
wherein m istAnd vtAre estimates of the first moment (mean) and the second moment (no central variance) of the gradient, respectively. The algorithm keeps the random gradient decline of multidimensional data to keep a single learning rate, and updates all weights in the time sequence convolution network.
Due to mtAnd vtVectors initialized to 0, which are biased toward 0, these biases can be calculated as:
these t are then used and the parameters are updated as:
β1default value is 0.9, beta2Default value of (2) is 0.999, and the default value of epsilon is 10-8. Each epoch is the entire process of neural network training through the entire data set, including forward and backward. The learning rate in the present invention is 0.002.
According to observation, the loss value is continuously reduced along with the increase of the iteration times, and after 100 generations of training, the predicted value of the used BilSTM-CNN network can be well fitted with the measured value.
Claims (3)
1. A road base layer strain analysis method based on a coupled bidirectional LSTM and a convolution structure network is characterized by comprising the following steps: the method comprises the following steps of acquiring historical monitoring data of a road surface by using an automatic road monitoring system, and predicting key data by combining a deep learning method, wherein the method comprises the following specific steps:
the method comprises the following steps: preprocessing original monitoring data;
firstly, integrating road monitoring data acquired by various sensors;
secondly, according to the time span of the data acquired by the road monitoring system, deleting the data repetition value and the jump value after selecting the data of the time period, and completing the supplement of the missing value;
finally, unifying the time intervals among the data points of all the time sequence data into 1 hour;
step two: analyzing the characteristic correlation;
firstly, carrying out correlation analysis on the first three data points acquired by each sensor after processing, and measuring the correlation between the characteristics according to a Pearson correlation coefficient; secondly, removing the soil layer strain due to extremely small relativity;
step three: making a data set;
averaging all monitoring data points of asphalt strain, osmotic pressure, soil moisture and base layer three-dimensional strain, and averaging the three parts of a surface layer, a base layer and a soil foundation according to the embedding position of a temperature sensor; and finally, the ratio of the monitoring data set to the training set to the testing set is about 8: 2, dividing;
step four: predicting the strain of a time sequence convolution network base layer;
the BilSTM-CNN network consists of 4 layers of one-dimensional convolution layers, one layer of bidirectional LSTM and two layers of full connection layers; the convolution kernel size of the one-dimensional convolution layer is 2, and the number of the convolution kernels is 18, 36, 72 and 144 in sequence; the method comprises the following steps that a maximum pooling layer with the size of 2 and the step length of 1 is arranged behind the device, and a ReLU activation function is used for each layer; and taking the preprocessed and dimensionality-reduced data set as the input of the BilSTM-CNN network, and setting the time step length to be 8.
2. The method of claim 1, wherein in the process of training the BiLSTM-CNN network on the road surface monitoring data, the input multidimensional data and the target data are trained together, from which the physical relationship between the data characteristics and the data is continuously learned, and the optimization target is to minimize the error between the predicted value and the measured value; and taking an adaptive learning rate adjustment algorithm Adadelta as an optimization algorithm of gradient descent in the back propagation process.
3. The strain analysis and prediction method based on deep learning of the automatic road monitoring system according to claim 1, characterized in that a BilSTM-CNN network is used to process the monitoring data: (1) acquiring context information of data by adopting a bidirectional LSTM; (2) adding a one-dimensional convolutional layer to extract invariant features in the monitoring data; (3) the one-dimensional convolution layer is followed by a maximum pooling layer; (4) all layers use the ReLU activation function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110877678.8A CN114048790B (en) | 2021-08-01 | Road base strain analysis method based on coupling bidirectional LSTM and convolution structure network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110877678.8A CN114048790B (en) | 2021-08-01 | Road base strain analysis method based on coupling bidirectional LSTM and convolution structure network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114048790A true CN114048790A (en) | 2022-02-15 |
| CN114048790B CN114048790B (en) | 2024-05-24 |
Family
ID=
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114413832A (en) * | 2021-12-02 | 2022-04-29 | 中国通信建设第三工程局有限公司 | Road monitoring system and method based on optical fiber sensing |
| CN115081706A (en) * | 2022-06-16 | 2022-09-20 | 中国安能集团第三工程局有限公司 | Loess collapse prediction method and device based on bidirectional LSTM network |
| CN115077474A (en) * | 2022-06-14 | 2022-09-20 | 天津华北地质勘查总院 | Ground settlement trend prediction method and system based on machine learning |
| CN115329812A (en) * | 2022-08-10 | 2022-11-11 | 贵州桥梁建设集团有限责任公司 | Road infrastructure abnormity monitoring method based on artificial intelligence |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355330A (en) * | 2016-08-31 | 2017-01-25 | 郑州航空工业管理学院 | Multi-response parameter optimization method based on radial basis function neural network prediction model |
| CN111209620A (en) * | 2019-12-30 | 2020-05-29 | 浙江大学 | LSTM-cGAN-based prediction method for residual bearing capacity and crack propagation path of crack-containing structure |
| CN112001177A (en) * | 2020-08-24 | 2020-11-27 | 浪潮云信息技术股份公司 | Electronic medical record named entity identification method and system integrating deep learning and rules |
| CN112069831A (en) * | 2020-08-21 | 2020-12-11 | 三峡大学 | Unreal information detection method based on BERT model and enhanced hybrid neural network |
| CN112733442A (en) * | 2020-12-31 | 2021-04-30 | 交通运输部公路科学研究所 | Road surface long-term performance prediction model based on deep learning and construction method thereof |
| CN112747779A (en) * | 2020-12-23 | 2021-05-04 | 武汉理工光科股份有限公司 | Asphalt pavement intelligent monitoring system based on fiber bragg grating sensing |
| CN112819136A (en) * | 2021-01-20 | 2021-05-18 | 南京邮电大学 | Time sequence prediction method and system based on CNN-LSTM neural network model and ARIMA model |
| CN112953862A (en) * | 2021-02-07 | 2021-06-11 | 哈尔滨工程大学 | High-speed railway scene channel estimation method based on deep learning |
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355330A (en) * | 2016-08-31 | 2017-01-25 | 郑州航空工业管理学院 | Multi-response parameter optimization method based on radial basis function neural network prediction model |
| CN111209620A (en) * | 2019-12-30 | 2020-05-29 | 浙江大学 | LSTM-cGAN-based prediction method for residual bearing capacity and crack propagation path of crack-containing structure |
| CN112069831A (en) * | 2020-08-21 | 2020-12-11 | 三峡大学 | Unreal information detection method based on BERT model and enhanced hybrid neural network |
| CN112001177A (en) * | 2020-08-24 | 2020-11-27 | 浪潮云信息技术股份公司 | Electronic medical record named entity identification method and system integrating deep learning and rules |
| CN112747779A (en) * | 2020-12-23 | 2021-05-04 | 武汉理工光科股份有限公司 | Asphalt pavement intelligent monitoring system based on fiber bragg grating sensing |
| CN112733442A (en) * | 2020-12-31 | 2021-04-30 | 交通运输部公路科学研究所 | Road surface long-term performance prediction model based on deep learning and construction method thereof |
| CN112819136A (en) * | 2021-01-20 | 2021-05-18 | 南京邮电大学 | Time sequence prediction method and system based on CNN-LSTM neural network model and ARIMA model |
| CN112953862A (en) * | 2021-02-07 | 2021-06-11 | 哈尔滨工程大学 | High-speed railway scene channel estimation method based on deep learning |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114413832A (en) * | 2021-12-02 | 2022-04-29 | 中国通信建设第三工程局有限公司 | Road monitoring system and method based on optical fiber sensing |
| CN114413832B (en) * | 2021-12-02 | 2023-12-15 | 中国通信建设第三工程局有限公司 | Road monitoring system and method based on optical fiber sensing |
| CN115077474A (en) * | 2022-06-14 | 2022-09-20 | 天津华北地质勘查总院 | Ground settlement trend prediction method and system based on machine learning |
| CN115081706A (en) * | 2022-06-16 | 2022-09-20 | 中国安能集团第三工程局有限公司 | Loess collapse prediction method and device based on bidirectional LSTM network |
| CN115329812A (en) * | 2022-08-10 | 2022-11-11 | 贵州桥梁建设集团有限责任公司 | Road infrastructure abnormity monitoring method based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113777496B (en) | Lithium ion battery residual life prediction method based on time convolution neural network | |
| CN115758212B (en) | Mechanical equipment fault diagnosis method based on parallel network and transfer learning | |
| CN112070322B (en) | High-voltage cable line running state prediction method based on long-short term memory network | |
| CN111784070A (en) | Intelligent landslide short-term early warning method based on XGboost algorithm | |
| CN110083125B (en) | Machine tool thermal error modeling method based on deep learning | |
| CN113128113B (en) | Lean information building load prediction method based on deep learning and transfer learning | |
| CN113743016B (en) | Engine residual life prediction method based on self-encoder and echo state network | |
| CN115987295A (en) | Crop monitoring data efficient processing method based on Internet of things | |
| CN115271186B (en) | Reservoir water level prediction and early warning method based on delay factor and PSO RNN Attention model | |
| CN110851897B (en) | Aqueduct stress-strain prediction method under multi-factor association | |
| CN114266278A (en) | Dual-attention-network-based method for predicting residual service life of equipment | |
| CN115730684A (en) | Air quality detection system based on LSTM-CNN model | |
| CN115374903A (en) | Long-term pavement monitoring data enhancement method based on expressway sensor network layout | |
| CN110222825B (en) | Cement product specific surface area prediction method and system | |
| CN115982141A (en) | Characteristic optimization method for time series data prediction | |
| CN115860197A (en) | Data-driven coal bed gas yield prediction method and system | |
| CN110260914B (en) | Engineering safety monitoring system region division method based on measuring point space-time characteristics | |
| CN116662920B (en) | Abnormal data identification method, system, equipment and medium for drilling and blasting method construction equipment | |
| CN114048790B (en) | Road base strain analysis method based on coupling bidirectional LSTM and convolution structure network | |
| CN114048790A (en) | Road base layer strain analysis method based on coupling bidirectional LSTM and convolution structure network | |
| CN116364203A (en) | Water quality prediction method, system and device based on deep learning | |
| CN114048835A (en) | Road base strain analysis method based on time sequence convolution network | |
| CN113973403B (en) | Temperature-induced strain field redistribution intelligent sensing method based on structure discrete measurement point topology | |
| CN115510748A (en) | Landslide displacement prediction method based on variational modal decomposition and CNN-GRU | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |