CN117708643A - Bridge monitoring abnormal data identification method and system based on fusion sequence characteristics - Google Patents
Bridge monitoring abnormal data identification method and system based on fusion sequence characteristics Download PDFInfo
- Publication number
- CN117708643A CN117708643A CN202311474400.1A CN202311474400A CN117708643A CN 117708643 A CN117708643 A CN 117708643A CN 202311474400 A CN202311474400 A CN 202311474400A CN 117708643 A CN117708643 A CN 117708643A
- Authority
- CN
- China
- Prior art keywords
- data
- anomaly
- monitoring
- initial
- calculating
- 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
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 70
- 238000012544 monitoring process Methods 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000004927 fusion Effects 0.000 title claims abstract description 23
- 230000008439 repair process Effects 0.000 claims abstract description 26
- 239000013598 vector Substances 0.000 claims abstract description 21
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 17
- 230000011218 segmentation Effects 0.000 claims abstract description 14
- 230000036541 health Effects 0.000 claims abstract description 13
- 230000015654 memory Effects 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims description 12
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000002156 mixing Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 239000010410 layer Substances 0.000 description 46
- 230000006870 function Effects 0.000 description 17
- 238000013528 artificial neural network Methods 0.000 description 14
- 238000012549 training Methods 0.000 description 9
- 238000011176 pooling Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 230000005856 abnormality Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 7
- 238000000605 extraction Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 6
- 230000004913 activation Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000010355 oscillation Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 239000012792 core layer Substances 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 238000005067 remediation Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a method and a system for identifying bridge monitoring abnormal data based on fusion sequence characteristics, wherein the method comprises the following steps: acquiring bridge health monitoring data, extracting feature vectors of the monitoring data as data points, and constructing a data anomaly identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to a comparison result; calculating initial anomaly scores of data according to the data anomaly identification tree, calculating total probability density of the initial anomaly scores of the data, setting a data anomaly model, calculating final anomaly scores of the data according to the total probability density, and identifying anomaly data in the monitoring data according to the final anomaly scores of the data; and inputting the identified abnormal data into a convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into a long-short-time memory network to finish the repair of the abnormal data.
Description
Technical Field
The invention belongs to the technical field of recognition of bridge monitoring abnormal data, and particularly relates to a method and a system for recognizing bridge monitoring abnormal data based on fusion sequence characteristics.
Background
With the acceleration of the urban process, the bridge is an important component of the infrastructure, and the health status monitoring of the bridge is particularly important. However, the conventional health monitoring method has problems of data distortion and abnormality, which is particularly prominent in complex bridge structures, and may lead to incorrect structural health evaluation, affecting safety judgment. In this context, neural networks, as a powerful pattern recognition, feature extraction and data modeling technique, bring new prospects for bridge health monitoring systems (Structure Health Monitoring system, SHMs).
At present, the technology for identifying and repairing the abnormal data of the SHMs is generally only capable of identifying without classification, so that valuable information in certain types of abnormal data is lost, or the processing method is too complex, so that the efficiency is low.
Disclosure of Invention
In order to solve the technical problems, the invention provides a method for identifying bridge monitoring abnormal data based on fusion sequence characteristics, which comprises the following steps:
acquiring bridge health monitoring data, extracting feature vectors of the monitoring data as data points, and constructing a data anomaly identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to comparison results;
calculating initial anomaly scores of data according to the data anomaly identification tree, calculating total probability density of the initial anomaly scores of the data, setting a data anomaly model, calculating final anomaly scores of the data according to the total probability density, and identifying anomaly data in the monitoring data according to the final anomaly scores of the data;
and inputting the identified abnormal data into a convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into a long-short-time-based memory network to finish the repair of the abnormal data.
Further, calculating the initial anomaly score for the data includes:
where S is the initial anomaly score of the data, n is the number of data points, c (n) is the normalization constant of the tree, and x is the average depth of the data points in the tree.
Further, calculating the probability density of the initial anomaly score S of the data includes:
wherein P (S) is the total probability density of the initial anomaly score S of the data, pi k Mixing coefficient of kth component, K is total number of components, mu k Is the mean value of the kth component, e k Is the co-party of the kth componentThe difference matrix is used to determine the difference between the two,is the probability density of the initial anomaly score S for the data given the mean and covariance.
Further, the data anomaly model includes:
AnomalyScore＝-log(P(S))
where AnomalyScare is the final anomaly score of the data.
Further, the method further comprises the following steps: and preprocessing the monitoring data.
The invention also provides a system for identifying bridge monitoring abnormal data based on the fusion sequence characteristics, which comprises the following steps:
the system comprises a building tree module, a data exception identification tree generation module and a data exception identification tree generation module, wherein the building tree module is used for acquiring bridge health monitoring data, extracting feature vectors of the monitoring data as data points, and building a data exception identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to a comparison result;
the identification module is used for calculating initial anomaly scores of the data according to the data anomaly identification tree, calculating total probability density of the initial anomaly scores of the data, setting a data anomaly model, calculating final anomaly scores of the data according to the total probability density, and identifying anomaly data in the monitoring data according to the final anomaly scores of the data;
and the repair module is used for inputting the identified abnormal data into the convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into the long-short-time-based memory network to finish the repair of the abnormal data.
Further, calculating the initial anomaly score for the data includes:
where S is the initial anomaly score of the data, n is the number of data points, c (n) is the normalization constant of the tree, and x is the average depth of the data points in the tree.
Further, calculating the probability density of the initial anomaly score S of the data includes:
wherein P (S) is the total probability density of the initial anomaly score S of the data, pi k Mixing coefficient of kth component, K is total number of components, mu k Is the mean value of the kth component, e k For the covariance matrix of the kth component,is the probability density of the initial anomaly score S for the data given the mean and covariance.
Further, the data anomaly model includes:
AnomalyScore＝-log(P(S))
where AnomalyScare is the final anomaly score of the data.
Further, the method further comprises the following steps: and preprocessing the monitoring data.
Compared with the prior art, the technical scheme of the invention has the following beneficial effects:
the method can realize rapid anomaly identification and repair based on one-dimensional data through a neural network technology; the method comprises the steps of identifying abnormal data by constructing a data abnormal identification tree, repairing the abnormal data with serious original data information loss based on a Long Short-Term Memory (LSTM) network, and rapidly capturing Long-Term dependency and characteristics in a time sequence; the training cost is reduced, the manual intervention operation is reduced, meaningless repeated labor is avoided, and the productivity is liberated.
Drawings
FIG. 1 is a flow chart of embodiment 1 of the present invention;
FIG. 2 is a block diagram of the system of embodiment 2 of the present invention;
FIG. 3 is a schematic diagram of data classification and repair of the present invention;
FIG. 4 is a diagram of a data anomaly classification network framework of the present invention;
FIG. 5 is a diagram of an anomaly data location network framework of the present invention;
FIG. 6 is a flow chart of anomaly repair based on the LSTM method of the present invention.
Detailed Description
In order to better understand the above technical solutions, the following detailed description will be given with reference to the accompanying drawings and specific embodiments.
The method provided by the invention can be implemented in a terminal environment, wherein the terminal can comprise one or more of the following components: processor, storage medium, and display screen. Wherein the storage medium has stored therein at least one instruction that is loaded and executed by the processor to implement the method described in the embodiments below.
The processor may include one or more processing cores. The processor connects various parts within the overall terminal using various interfaces and lines, performs various functions of the terminal and processes data by executing or executing instructions, programs, code sets, or instruction sets stored in the storage medium, and invoking data stored in the storage medium.
The storage medium may include a random access Memory (Random Access Memory, RAM) or a Read-Only Memory (ROM). The storage medium may be used to store instructions, programs, code sets, or instructions.
The display screen is used for displaying a user interface of each application program.
All subscripts in the formula of the invention are only used for distinguishing parameters and have no practical meaning.
In addition, it will be appreciated by those skilled in the art that the structure of the terminal described above is not limiting and that the terminal may include more or fewer components, or may combine certain components, or a different arrangement of components. For example, the terminal further includes components such as a radio frequency circuit, an input unit, a sensor, an audio circuit, a power supply, and the like, which are not described herein.
Example 1
As shown in fig. 1, an embodiment of the present invention provides a method for identifying bridge monitoring abnormal data based on fusion sequence features, including:
step 101, acquiring bridge health monitoring data, preprocessing the monitoring data, extracting feature vectors of the monitoring data as data points, and constructing a data anomaly identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to a comparison result;
102, calculating initial anomaly scores of data according to the data anomaly identification tree, calculating total probability density of the initial anomaly scores of the data, setting a data anomaly model, calculating final anomaly scores of the data according to the total probability density, and identifying anomaly data in the monitoring data according to the final anomaly scores of the data;
specifically, calculating the initial anomaly score for the data includes:
where S is the initial anomaly score of the data, n is the number of data points, c (n) is the normalization constant of the tree, and x is the average depth of the data points in the tree.
Specifically, calculating the probability density of the initial anomaly score S of the data includes:
wherein P (S) is the total probability density of the initial anomaly score S of the data, pi k Mixing coefficient of kth component, K is total number of components, mu k Is the mean value of the kth component, e k For the covariance matrix of the kth component,is the probability density of the initial anomaly score S for the data given the mean and covariance.
Specifically, the data anomaly model includes:
AnomalyScare= -log (P (S)) where AnomalyScare is the final anomaly score of the data.
And step 103, inputting the identified abnormal data into a convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into a long-short-time-based memory network to finish the repair of the abnormal data.
Specifically, in combination with CNN and LSTM technologies, an abnormal data repair framework for a health monitoring system is built, as shown in fig. 3, to recover the integrity and accuracy of data through multi-stage processing. The framework not only considers the classification and the positioning of the anomalies, but also combines the repairing strategies of different types of abnormal data, and covers the whole process from data preprocessing to repairing, and the main steps are as follows:
(1) Pretreatment: the method mainly completes the functions of data preparation, data enhancement, sample balancing, data partitioning and the like, and provides a more reliable data basis for subsequent abnormality classification and repair through data preprocessing.
(2) Abnormality classification and positioning: and dividing different abnormal categories of the data by combining with a CNN method. These categories include missing, next-smallest, trend, outlier, overscan oscillation, drift, and normal data. And then, accurately positioning the abnormal data by using the abnormal classification result.
(3) Abnormal data repair: at this stage, different repair strategies are employed to restore the integrity of the data, depending on the type of anomaly.
Outlier, trend, and drift type anomaly data remediation: for these types of anomaly data, we employ an anomaly repair method based on the original data drive. The method involves the technologies of data interpolation, integral proportional movement and the like so as to restore abnormal data into normal data, thereby better reflecting the actual condition of the structure.
Repairing abnormal data of missing, next-smallest and overscan oscillation: for these types of abnormal data, a data repair method based on an LSTM neural network is introduced. By training the LSTM model, missing data points are predicted, thereby restoring data integrity.
The frames of the abnormal data classification network and the abnormal data positioning network are respectively shown in fig. 4 and 5, and the network frames are mainly divided into the following parts:
(1) Input layer: the input layer is the first layer of the neural network, which receives input data and passes it on to the next layer.
(2) And a feature extraction layer. Different functions correspond to different forms of frame details. Wherein the convolution layer is a core layer of the feature extraction layer, and local features are extracted by using different convolution kernels. In addition, the feature extraction layer includes other types of layers to optimize the training process and implement functions, as follows:
convolution layer: the convolutional layer is essentially an operation dedicated to linear transformation. In CNN, each convolution layer is made up of a plurality of convolution units, the parameters of each convolution unit being optimized by a back-propagation algorithm. The purpose of the convolution operation is to extract different features from the input.
Activation function ReLU: the ReLU can carry out nonlinear mapping on the output of the CNN layer, and has the advantages of high calculation speed and information filtering effect.
BatchNorm layer: the BatchNorm layer may prevent the gradient from disappearing or exploding, thereby reducing training time.
Maxpool layer: pooling is generally divided into maximum pooling and average pooling. The method is used for reducing the size of the feature map in the CNN or other types of neural networks, so that the calculation amount is reduced, the complexity of the model is reduced, and the robustness of the model is enhanced. In maximum pooling, the output value is obtained by selecting the maximum value from the local area, thereby preserving the most significant features.
(3) And a characteristic information fusion layer. The two forms of neural networks adopt the same characteristic information fusion layer:
full tie layer: the feature information extracted from the previous layer may be combined.
The activation function selects the Softmax function: the Softmax function may match the output range to the predicted demand, i.e., all limited to the 0-1 range.
(4) And an output layer. The output layer is the last layer of the neural network and is used for outputting the processing result of the neural network. The output size of the classification network is 1 x 7, corresponding to seven data types including normal data. The output size of the positioning network is 1 x 2, corresponding to the starting and ending values of the normalized anomaly range of the data.
The data exception repairing flow based on the LSTM method is shown in fig. 6, and the specific steps are as follows:
(1) After successfully identifying the data anomalies, data preprocessing is needed, and specific operations are to delete and set the data paragraphs with the data anomalies as 0 value, so that various data anomalies (drift, suboptimal value and the like) are converted into 'missing'.
(2) And inputting a part of data of the normal section in the data into the LSTM to obtain a data anomaly repair model.
(3) The remaining part of normal data (normal data which does not participate in training) is used for predicting the 'missing' part of data, so that the data abnormality is repaired.
Example 2
As shown in fig. 2, the embodiment of the invention further provides a system for identifying bridge monitoring abnormal data based on fusion sequence features, which includes:
the sample processing module is used for acquiring bridge health monitoring data, preprocessing the monitoring data, extracting feature vectors of the monitoring data as data points, and constructing a data anomaly identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to a comparison result;
the data anomaly model is set, the final anomaly score of the data is calculated according to the total probability density, and the anomaly data in the monitoring data is identified according to the final anomaly score of the data;
specifically, calculating the initial anomaly score for the data includes:
where S is the initial anomaly score of the data, n is the number of data points, c (n) is the normalization constant of the tree, and x is the average depth of the data points in the tree.
Specifically, calculating the probability density of the initial anomaly score S of the data includes:
wherein P (S) is the total probability density of the initial anomaly score S of the data, pi k Mixing coefficient of kth component, K is total number of components, mu k Is the mean value of the kth component, e k For the covariance matrix of the kth component,is the probability density of the initial anomaly score S for the data given the mean and covariance.
Specifically, the data anomaly model includes:
AnomalyScore＝-log(P(S))
where AnomalyScare is the final anomaly score of the data.
The detection module is used for inputting the identified abnormal data into the convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into the long-short-time-based memory network so as to finish the repair of the abnormal data.
Specifically, in combination with CNN and LSTM technologies, an abnormal data repair framework for a health monitoring system is built, as shown in fig. 3, to recover the integrity and accuracy of data through multi-stage processing. The framework not only considers the classification and the positioning of the anomalies, but also combines the repairing strategies of different types of abnormal data, and covers the whole process from data preprocessing to repairing, and the main steps are as follows:
(1) Pretreatment: the method mainly completes the functions of data preparation, data enhancement, sample balancing, data partitioning and the like, and provides a more reliable data basis for subsequent abnormality classification and repair through data preprocessing.
(2) Abnormality classification and positioning: and dividing different abnormal categories of the data by combining with a CNN method. These categories include missing, next-smallest, trend, outlier, overscan oscillation, drift, and normal data. And then, accurately positioning the abnormal data by using the abnormal classification result.
(3) Abnormal data repair: at this stage, different repair strategies are employed to restore the integrity of the data, depending on the type of anomaly.
Outlier, trend, and drift type anomaly data remediation: for these types of anomaly data, we employ an anomaly repair method based on the original data drive. The method involves the technologies of data interpolation, integral proportional movement and the like so as to restore abnormal data into normal data, thereby better reflecting the actual condition of the structure.
Repairing abnormal data of missing, next-smallest and overscan oscillation: for these types of abnormal data, a data repair method based on an LSTM neural network is introduced. By training the LSTM model, missing data points are predicted, thereby restoring data integrity.
The frames of the abnormal data classification network and the abnormal data positioning network are respectively shown in fig. 4 and 5, and the network frames are mainly divided into the following parts:
(1) Input layer: the input layer is the first layer of the neural network, which receives input data and passes it on to the next layer.
(2) And a feature extraction layer. Different functions correspond to different forms of frame details. Wherein the convolution layer is a core layer of the feature extraction layer, and local features are extracted by using different convolution kernels. In addition, the feature extraction layer includes other types of layers to optimize the training process and implement functions, as follows:
convolution layer: the convolutional layer is essentially an operation dedicated to linear transformation. In CNN, each convolution layer is made up of a plurality of convolution units, the parameters of each convolution unit being optimized by a back-propagation algorithm. The purpose of the convolution operation is to extract different features from the input.
Activation function ReLU: the ReLU can carry out nonlinear mapping on the output of the CNN layer, and has the advantages of high calculation speed and information filtering effect.
BatchNorm layer: the BatchNorm layer may prevent the gradient from disappearing or exploding, thereby reducing training time.
Maxpool layer: pooling is generally divided into maximum pooling and average pooling. The method is used for reducing the size of the feature map in the CNN or other types of neural networks, so that the calculation amount is reduced, the complexity of the model is reduced, and the robustness of the model is enhanced. In maximum pooling, the output value is obtained by selecting the maximum value from the local area, thereby preserving the most significant features.
(3) And a characteristic information fusion layer. The two forms of neural networks adopt the same characteristic information fusion layer:
full tie layer: the feature information extracted from the previous layer may be combined.
The activation function selects the Softmax function: the Softmax function may match the output range to the predicted demand, i.e., all limited to the 0-1 range.
(4) And an output layer. The output layer is the last layer of the neural network and is used for outputting the processing result of the neural network. The output size of the classification network is 1 x 7, corresponding to seven data types including normal data. The output size of the positioning network is 1 x 2, corresponding to the starting and ending values of the normalized anomaly range of the data.
The data exception repairing flow based on the LSTM method is shown in fig. 6, and the specific steps are as follows:
(1) After successfully identifying the data anomalies, data preprocessing is needed, and specific operations are to delete and set the data paragraphs with the data anomalies as 0 value, so that various data anomalies (drift, suboptimal value and the like) are converted into 'missing'.
(2) And inputting a part of data of the normal section in the data into the LSTM to obtain a data anomaly repair model.
(3) The remaining part of normal data (normal data which does not participate in training) is used for predicting the 'missing' part of data, so that the data abnormality is repaired.
Example 3
The embodiment of the invention also provides a storage medium which stores a plurality of instructions for realizing the method for identifying the bridge monitoring abnormal data based on the fusion sequence characteristics.
Alternatively, in this embodiment, the storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, or in any one of the mobile terminals in the mobile terminal group.
Alternatively, in the present embodiment, a storage medium is provided to store program codes for performing the steps of embodiment 1.
Example 4
The embodiment of the invention also provides electronic equipment, which comprises a processor and a storage medium connected with the processor, wherein the storage medium stores a plurality of instructions, and the instructions can be loaded and executed by the processor so that the processor can execute a method for identifying bridge monitoring abnormal data based on fusion sequence characteristics.
Specifically, the electronic device of the present embodiment may be a computer terminal, and the computer terminal may include: one or more processors, and a storage medium.
The storage medium may be used to store a software program and a module, for example, in an embodiment of the present invention, a method for identifying bridge monitoring abnormal data based on a fusion sequence feature, corresponding to a program instruction/module, and the processor executes various function applications and data processing by running the software program and the module stored in the storage medium, that is, the method for identifying bridge monitoring abnormal data based on the fusion sequence feature is implemented. The storage medium may include a high-speed random access storage medium, and may also include a non-volatile storage medium, such as one or more magnetic storage systems, flash memory, or other non-volatile solid-state storage medium. In some examples, the storage medium may further include a storage medium remotely located with respect to the processor, and the remote storage medium may be connected to the terminal through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The processor may invoke the information stored in the storage medium and the application program through the transmission system to perform the steps of embodiment 1.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present invention, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the embodiments provided in the present invention, it should be understood that the disclosed technology may be implemented in other manners. The system embodiments described above are merely exemplary, and for example, the division of the units is merely a logic function division, and there may be another division manner in actual implementation, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or partly in the form of a software product or all or part of the technical solution, which is stored in a storage medium, and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random-access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or the like, which can store program codes.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (10)
1. The method for identifying the bridge monitoring abnormal data based on the fusion sequence features is characterized by comprising the following steps of:
acquiring bridge health monitoring data, extracting feature vectors of the monitoring data as data points, and constructing a data anomaly identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to comparison results;
calculating initial anomaly scores of data according to the data anomaly identification tree, calculating total probability density of the initial anomaly scores of the data, setting a data anomaly model, calculating final anomaly scores of the data according to the total probability density, and identifying anomaly data in the monitoring data according to the final anomaly scores of the data;
and inputting the identified abnormal data into a convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into a long-short-time memory network to finish the repair of the abnormal data.
2. The method for identifying bridge monitoring anomaly data based on fusion sequence features of claim 1, wherein calculating an initial anomaly score for the data comprises:
where S is the initial anomaly score of the data, n is the number of data points, c (n) is the normalization constant of the tree, and x is the average depth of the data points in the tree.
3. The method for identifying bridge monitoring anomaly data based on fused sequence features as claimed in claim 2, wherein calculating the probability density of the initial anomaly score S of the data comprises:
wherein P (S) is the total probability density of the initial anomaly score S of the data, pi k Mixing coefficient of kth component, K is total number of components, mu k Is the mean value of the kth component, e k For the covariance matrix of the kth component,is the probability density of the initial anomaly score S for the data given the mean and covariance.
4. A method for identifying bridge monitoring anomaly data based on fusion sequence features as claimed in claim 3, wherein the data anomaly model comprises:
AnomalyScore＝-log(P(S))
where AnomalyScare is the final anomaly score of the data.
5. The method for identifying bridge monitoring anomaly data based on fusion sequence features as defined in claim 1, further comprising: and preprocessing the monitoring data.
6. The utility model provides a recognition system of bridge monitoring abnormal data based on fusion sequence characteristic which characterized in that includes:
the system comprises a building tree module, a data exception identification tree generation module and a data exception identification tree generation module, wherein the building tree module is used for acquiring bridge health monitoring data, extracting feature vectors of the monitoring data as data points, and building a data exception identification tree, wherein a segmentation threshold is set, the feature vectors are compared with the segmentation threshold, and the feature vectors are respectively segmented into a first subtree and a second subtree according to a comparison result;
the identification module is used for calculating initial anomaly scores of the data according to the data anomaly identification tree, calculating total probability density of the initial anomaly scores of the data, setting a data anomaly model, calculating final anomaly scores of the data according to the total probability density, and identifying anomaly data in the monitoring data according to the final anomaly scores of the data;
and the repair module is used for inputting the identified abnormal data into the convolutional neural network, classifying the abnormal data, and inputting the classified abnormal data into the long-short-time memory network so as to finish the repair of the abnormal data.
7. The system for identifying bridge monitoring anomaly data based on fused sequence features of claim 6, wherein calculating an initial anomaly score for the data comprises:
where S is the initial anomaly score of the data, n is the number of data points, c (n) is the normalization constant of the tree, and x is the average depth of the data points in the tree.
8. The system for identifying bridge monitoring anomaly data based on fused sequence features of claim 7, wherein calculating the probability density of the initial anomaly score S of the data comprises:
wherein P (S) is the total probability density of the initial anomaly score S of the data, pi k Mixing coefficient of kth component, K is total number of components, mu k Is the mean value of the kth component, e k For the covariance matrix of the kth component,is the probability density of the initial anomaly score S for the data given the mean and covariance.
9. The system for identifying bridge monitoring anomaly data based on fusion sequence features of claim 8, wherein the data anomaly model comprises:
AnomalyScore＝-log(P(S))
where AnomalyScare is the final anomaly score of the data.
10. The system for identifying bridge monitoring anomaly data based on fusion sequence features of claim 6, further comprising: and preprocessing the monitoring data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311474400.1A CN117708643B (en) | 2023-11-07 | Bridge monitoring abnormal data identification method and system based on fusion sequence characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311474400.1A CN117708643B (en) | 2023-11-07 | Bridge monitoring abnormal data identification method and system based on fusion sequence characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117708643A true CN117708643A (en) | 2024-03-15 |
| CN117708643B CN117708643B (en) | 2024-07-09 |
Family
ID=
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115130540A (en) * | 2022-04-24 | 2022-09-30 | 山西省交通建设工程质量检测中心（有限公司） | Bridge abnormal point identification method and device based on multi-granularity time window |
| CN115683504A (en) * | 2022-08-29 | 2023-02-03 | 山东高速集团有限公司创新研究院 | Bridge acceleration monitoring data anomaly identification method and system based on multi-label classification |
| CN116204770A (en) * | 2022-12-12 | 2023-06-02 | 中国公路工程咨询集团有限公司 | Training method and device for detecting abnormality of bridge health monitoring data |
| US20230228618A1 (en) * | 2021-10-15 | 2023-07-20 | Southeast University | A dynamic identification method of bridge scour based on health monitoring data |
| CN116502163A (en) * | 2023-04-20 | 2023-07-28 | 哈尔滨工业大学 | Vibration monitoring data anomaly detection method based on multi-feature fusion and deep learning |
| CN116776267A (en) * | 2023-06-13 | 2023-09-19 | 西南交通大学 | Unsupervised data processing method and system for bridge construction control |
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230228618A1 (en) * | 2021-10-15 | 2023-07-20 | Southeast University | A dynamic identification method of bridge scour based on health monitoring data |
| CN115130540A (en) * | 2022-04-24 | 2022-09-30 | 山西省交通建设工程质量检测中心（有限公司） | Bridge abnormal point identification method and device based on multi-granularity time window |
| CN115683504A (en) * | 2022-08-29 | 2023-02-03 | 山东高速集团有限公司创新研究院 | Bridge acceleration monitoring data anomaly identification method and system based on multi-label classification |
| CN116204770A (en) * | 2022-12-12 | 2023-06-02 | 中国公路工程咨询集团有限公司 | Training method and device for detecting abnormality of bridge health monitoring data |
| CN116502163A (en) * | 2023-04-20 | 2023-07-28 | 哈尔滨工业大学 | Vibration monitoring data anomaly detection method based on multi-feature fusion and deep learning |
| CN116776267A (en) * | 2023-06-13 | 2023-09-19 | 西南交通大学 | Unsupervised data processing method and system for bridge construction control |
Non-Patent Citations (1)
| Title |
|---|
| 杨建喜;张利凯;李韧;何盈盈;蒋仕新;邹俊志;: "联合卷积与长短记忆神经网络的桥梁结构损伤识别研究", 铁道科学与工程学报, no. 08, 15 August 2020 (2020-08-15) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020140386A1 (en) | Textcnn-based knowledge extraction method and apparatus, and computer device and storage medium | |
| CN109272500B (en) | Fabric classification method based on adaptive convolutional neural network | |
| CN110704842A (en) | Malicious code family classification detection method | |
| CN113627266B (en) | Video pedestrian re-recognition method based on transform space-time modeling | |
| CN111091839B (en) | Voice awakening method and device, storage medium and intelligent device | |
| CN111401149B (en) | Lightweight video behavior identification method based on long-short-term time domain modeling algorithm | |
| CN111104855B (en) | Workflow identification method based on time sequence behavior detection | |
| CN111275694B (en) | Attention mechanism guided progressive human body division analysis system and method | |
| CN113762265A (en) | Pneumonia classification and segmentation method and system | |
| CN113298817A (en) | High-accuracy semantic segmentation method for remote sensing image | |
| CN113420289B (en) | Hidden poisoning attack defense method and device for deep learning model | |
| CN114581789A (en) | Hyperspectral image classification method and system | |
| CN112766134B (en) | Expression recognition method for strengthening distinction between classes | |
| CN114037893A (en) | High-resolution remote sensing image building extraction method based on convolutional neural network | |
| CN112182568A (en) | Malicious code classification based on graph convolution network and topic model | |
| CN111783688B (en) | Remote sensing image scene classification method based on convolutional neural network | |
| CN117708643B (en) | Bridge monitoring abnormal data identification method and system based on fusion sequence characteristics | |
| CN117095247A (en) | Numerical control machining-based machining gesture operation optimization method, system and medium | |
| CN117076862A (en) | Electric power Internet of things network anomaly detection method and system based on attribute map | |
| CN117708643A (en) | Bridge monitoring abnormal data identification method and system based on fusion sequence characteristics | |
| CN113688715A (en) | Facial expression recognition method and system | |
| CN114359786A (en) | Lip language identification method based on improved space-time convolutional network | |
| CN114694080A (en) | Detection method, system and device for monitoring violent behavior and readable storage medium | |
| CN115019057A (en) | Image feature extraction model determining method and device and image identification method and device | |
| CN112329013A (en) | Malicious code classification method based on graph convolution network and topic model |
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 |