CN114563150A - Bridge health online detection module generation method, detection method, tool box and device - Google Patents
Bridge health online detection module generation method, detection method, tool box and device Download PDFInfo
- Publication number
- CN114563150A CN114563150A CN202111592343.8A CN202111592343A CN114563150A CN 114563150 A CN114563150 A CN 114563150A CN 202111592343 A CN202111592343 A CN 202111592343A CN 114563150 A CN114563150 A CN 114563150A
- Authority
- CN
- China
- Prior art keywords
- bridge
- damage
- detection module
- damage identification
- data
- 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
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M7/00—Vibration-testing of structures; Shock-testing of structures
- G01M7/02—Vibration-testing by means of a shake table
- G01M7/025—Measuring arrangements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M7/00—Vibration-testing of structures; Shock-testing of structures
- G01M7/02—Vibration-testing by means of a shake table
- G01M7/022—Vibration control arrangements, e.g. for generating random vibrations
-
- 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
- G01N29/04—Analysing solids
-
- 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
-
- 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
- G01N29/4481—Neural networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2291/00—Indexing codes associated with group G01N29/00
- G01N2291/02—Indexing codes associated with the analysed material
- G01N2291/023—Solids
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Pathology (AREA)
- Signal Processing (AREA)
- Acoustics & Sound (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a bridge health online detection module generation method, a bridge health online detection module detection method, a tool box and a bridge health online detection module device. And carrying out extreme value statistics on the original data of the structural response data, further carrying out extreme value statistics on the structural response data after noise reduction and fast Fourier transform, combining the two statistical results to obtain data characteristics, and accurately capturing the damage characteristics of the bridge. The bridge damage identification classification model capable of completing bridge damage identification classification is obtained by training and learning through various machine learning algorithm models, and the model is integrated into a bridge damage detection module and can be freely called. Meanwhile, the evaluation indexes of the bridge damage identification classification models are calculated to guide the selection of the optimal or various bridge damage detection modules for identification detection, so that the damage identification precision is improved. Based on a full-automatic intelligent processing mode, the speed and the precision of detecting the bridge damage are greatly improved, and all-weather real-time detection is realized.
Description
Technical Field
The invention relates to the technical field of bridge safety assessment, in particular to a bridge health online detection module generation method, a bridge health online detection module detection method, a tool box and a device.
Background
By 12 months in 2020, the total number of Chinese bridges exceeds one million, wherein the total number of highway bridges reaches 87.83 ten thousand, 5716 grand bridges and 108344 grand bridges. With the rapid increase of the number of bridges in China and the gradual increase of the bridge ages of parts of bridges, the health and safety problems of bridges are also attracting more and more attention. If the safety problem of the bridge cannot be reflected and processed quickly in time, serious potential safety hazard exists, and even serious life and economic property loss is caused.
For a bridge structure, safety is the first principle, manual detection is mainly used in traditional bridge maintenance management, and the mode has the defects that long-term quantitative tracking of structural performance is difficult, workload is large, information feedback is not timely, and the like. In order to timely and quickly acquire some information of relevant states of a bridge and detect the existence of a fault in the initial stage of hidden danger of the bridge, the conventional common safeguard measure is to install various sensors in the structure during bridge construction, continuously acquire dynamic response data of the bridge during the operation of the bridge, grasp the health state of the bridge and timely find out the safety problem of the bridge. In this case, it is difficult to recognize and judge the damage of the bridge by a human, and even if it is possible to recognize an abnormality of data of some sensors, the accuracy is poor.
Disclosure of Invention
The embodiment of the invention provides a bridge health online detection module generation method, a detection method, a tool box and a device, which are used for eliminating or improving one or more defects in the prior art, solving the problems of difficult damage identification or low accuracy caused by large bridge damage data volume and severe data change and realizing automatic bridge damage identification.
The technical scheme of the invention is as follows:
in one aspect, the invention provides a method for generating a bridge health online detection module, which comprises the following steps:
obtaining a plurality of sample data, wherein each sample data at least comprises a structural dynamic response signal of a sensor arranged at each detection point on a bridge main body and bridge main body state information, the structural dynamic response signal is an acceleration signal of a dynamic test, and the bridge main body state information comprises two types of structural damage and structural non-damage;
counting original extreme value sequences of the structural dynamic response signals collected by all the sensors in each sample data; after noise reduction processing is carried out on the structural dynamic response signals collected by the sensors in each sample data by adopting maximum correlation kurtosis deconvolution, signal characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence of the signal characteristics is counted;
merging the original extreme value sequence and the characteristic extreme value sequence corresponding to each sample data, and adding a label to each sample data according to the bridge main body state information corresponding to each sample data to form a training sample set;
training a plurality of preset classification models by adopting the training sample set to obtain a plurality of bridge damage identification classification models; the preset classification model at least comprises: the system comprises a support vector machine, a decision tree, a full-connection neural network, a long-term and short-term memory neural network and a self-organizing mapping algorithm network;
calculating evaluation indexes of each bridge damage identification classification model, wherein the evaluation indexes comprise accuracy, precision, recall and F1 scores;
and integrating the bridge damage identification and classification models into a bridge damage detection module so as to select one or combine multiple bridge damage identification and classification models according to the evaluation index to perform bridge damage detection.
In some embodiments, the preset classification model further includes a random forest, a boosted tree, and a gradient boosted decision tree obtained based on an ensemble learning method framework in combination with the decision tree.
In some embodiments, the method further comprises:
acquiring a nearest algorithm model, and loading a nearest algorithm for classification calculation;
and calculating the evaluation index of the nearest algorithm model, and integrating the nearest algorithm model serving as a bridge damage identification classification model into the bridge damage detection module so as to select one or combine multiple bridge damage identification classification models according to the evaluation index to perform bridge damage detection.
In some embodiments, before obtaining the plurality of sample data, further comprising:
modeling a variable spring simply supported beam bridge, wherein the variable spring simply supported beam bridge is divided into a first set number of simply supported beam units with a first set length and a spring unit, the variable spring simply supported beam bridge is provided with a second set number of random excitation sources, and a third set number of sensors for detecting acceleration are equidistantly arranged on the variable spring simply supported beam bridge;
and each sensor acquires acceleration data as the structural dynamic response signal according to a set sampling frequency, and Gaussian noise is added to obtain the sample data.
In some embodiments, the method adopts a 5-fold cross validation mode to calculate the evaluation index of each bridge damage identification classification model.
In some embodiments, integrating the bridge damage identification classification models into a bridge damage detection module to select one or more bridge damage identification classification models for bridge damage detection according to the evaluation index includes:
carrying out normalization processing on the accuracy, the precision ratio, the recall ratio and the F1 score in the evaluation index, then carrying out weighted summation to obtain a comprehensive score of each bridge damage identification classification model, and arranging according to the comprehensive score from high to low;
selecting a bridge damage identification classification model with the highest comprehensive score according to the first probability to identify the bridge damage; or selecting a fourth set number of bridge damage identification classification models with higher comprehensive scores according to the second probability to identify the bridge damages, wherein the fourth set number is an odd number, and taking the bridge damage identification results with larger numbers as final results.
On the other hand, the invention also provides an intelligent online bridge health detection method, which comprises the following steps:
acquiring an acceleration data sequence detected by a third set number of sensors on the bridge to be detected at the current moment, wherein the sensors are arranged at equal intervals along the bridge to be detected;
counting to-be-detected original extreme value sequences of the acceleration data acquired by all the sensors in the acceleration data sequence; after noise reduction processing is carried out on the acceleration data sequence by adopting maximum correlation kurtosis deconvolution, the acceleration data sequence characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence to be detected of the acceleration data sequence characteristics is obtained;
and combining the original extreme value sequence to be detected and the characteristic extreme value sequence to be detected, and inputting the combined sequence into a bridge damage detection module in the bridge health online detection module generation method to obtain a damage detection result of the bridge to be detected.
On the other hand, the invention also provides an intelligent online bridge health detection kit, which at least comprises:
the data acquisition module is used for acquiring an acceleration data sequence detected by a third set number of sensors on the bridge to be detected at the current moment;
and the detection module is used for loading the bridge damage detection module in the bridge health online detection module generation method and executing the bridge health intelligent online detection method to obtain a damage detection result of the bridge to be detected.
In some embodiments, the kit further comprises a display module, connected to the detection module, for visually presenting the acceleration data sequence and the damage detection result.
In another aspect, the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the method are implemented.
The beneficial effects of the invention are:
in the bridge health online detection module generation method, the bridge health online detection method, the tool box and the device, the acceleration is acquired by the sensor to be used as a structural dynamic response signal so as to acquire the characteristics of the bridge in dynamic, static and damaged states. Furthermore, extreme value statistics is carried out on the original data of the structural response data, the extreme value statistics is carried out after noise reduction and fast Fourier transform of the structural response data are further carried out, the two statistics results are combined to obtain data characteristics, and the damage characteristics of the bridge can be accurately captured. The bridge damage identification classification model capable of completing bridge damage identification classification is obtained by training and learning through various machine learning algorithm models, and the model is integrated into a bridge damage detection module and can be freely called. Meanwhile, the evaluation indexes of the bridge damage identification classification models are calculated to guide the selection of the optimal or various bridge damage detection modules for identification detection, so that the damage identification precision is improved. Based on the full-automatic intelligent processing mode, the speed and the precision of detecting the bridge damage are greatly improved, and all-weather real-time detection is realized.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
It will be appreciated by those skilled in the art that the objects and advantages that can be achieved with the present invention are not limited to the specific details set forth above, and that these and other objects that can be achieved with the present invention will be more clearly understood from the detailed description that follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:
fig. 1 is a schematic flow chart of a method for generating an online bridge health detection module according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating an intelligent online bridge health detection method according to an embodiment of the present invention;
FIG. 3 is a structural diagram of a variable-spring simply-supported bridge obtained by modeling in a training process according to a bridge health online detection module generation method of the embodiment of the invention;
FIG. 4 is a schematic diagram of model training logic in a bridge health online detection module generation method according to another embodiment of the present invention;
FIG. 5 is a diagram illustrating an original structural dynamic response signal collected by a sensor in sample data generated by a bridge health online detection module according to an embodiment of the present invention;
FIG. 6 is the structure dynamic response signal of FIG. 5 after noise reduction;
FIG. 7 is a graph of the amplitude of FIG. 6 after a fast Fourier transform;
fig. 8 is a distribution diagram of detection results of each bridge damage identification classification model in the bridge health online detection module generation method according to an embodiment of the present invention.
Fig. 9 is a schematic structural diagram of a tool box according to an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings. The exemplary embodiments and descriptions of the present invention are provided to explain the present invention, but not to limit the present invention.
It should be noted that, in order to avoid obscuring the present invention with unnecessary details, only the structures and/or processing steps closely related to the scheme according to the present invention are shown in the drawings, and other details not so relevant to the present invention are omitted.
It should be emphasized that the term "comprises/comprising" when used herein, is taken to specify the presence of stated features, elements, steps or components, but does not preclude the presence or addition of one or more other features, elements, steps or components.
It is also noted herein that the term "coupled," if not specifically stated, may refer herein to not only a direct connection, but also an indirect connection in which an intermediate is present.
In recent years, due to the rapid development of traffic engineering at home and abroad, bridges have become an important part of traffic infrastructure. Among the bridge types, the simply supported bridge has the advantages of simple structure, suitability for various geological conditions and the like, and is a bridge shape which is earliest in application and most widely used in the beam bridge. In order to ensure the safety of the simply supported bridge during operation, bridge maintenance personnel need to know the health state of the simply supported bridge in real time, however, the damage information of the simply supported bridge is difficult to extract due to the lack of an effective data mining means. It is well known that structural damage can cause changes in structural performance (e.g., mass, stiffness, damping) which in turn can affect the structural dynamic response characteristics of the overall system, and the basic principle of damage detection based on structural dynamic response data is to use the dynamic response of these changes to identify or evaluate structural damage. With the rapid development of artificial intelligence technology, structural damage identification combining structural dynamic response information and artificial intelligence has become one of the key research points.
In the prior art, the judgment of the bridge damage mainly depends on the manual identification of professional technicians, and the identification mode has low efficiency and low accuracy and is difficult to adapt to the current requirement of detecting the bridge damage. Therefore, the invention provides a bridge health online detection module generation method, a bridge health online detection module detection method, a tool box and a bridge health online detection module device, which are used for accurately identifying bridge damage and health states.
Specifically, in one aspect, the present invention provides a method for generating a bridge health online detection module, as shown in fig. 1, including steps S101 to S106:
step S101: the method comprises the steps of obtaining a plurality of sample data, wherein each sample data at least comprises a structural dynamic response signal of a sensor arranged at each detection point on a bridge main body and bridge main body state information, the structural dynamic response signal is an acceleration signal of a dynamic test, and the bridge main body state information comprises two types of structural damage and structural non-damage.
Step S102: counting original extreme value sequences of structural dynamic response signals acquired by all sensors in each sample data; and after noise reduction processing is carried out on the structural dynamic response signals acquired by each sensor in each sample data by adopting maximum correlation kurtosis deconvolution, signal characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence of the signal characteristics is counted.
Step S103: and merging the original extreme value sequence and the characteristic extreme value sequence corresponding to each sample datum, and adding a label to each sample datum according to the bridge main body state information corresponding to each sample datum to form a training sample set.
Step S104: training a plurality of preset classification models by adopting a training sample set to obtain a plurality of bridge damage identification classification models; the preset classification model at least comprises: support vector machine, decision tree, full-connection neural network, long-short term memory neural network and self-organizing mapping algorithm network.
Step S105: and calculating evaluation indexes of each bridge damage identification classification model, wherein the evaluation indexes comprise accuracy, precision, recall and F1 scores.
Step S106: and integrating the bridge damage identification and classification models into a bridge damage detection module so as to select one or combine multiple bridge damage identification and classification models according to the evaluation index to perform bridge damage detection.
In step S101, sample data may be obtained from real bridge detection data, or may be obtained by computer modeling. In some embodiments, the sample data employed is derived from a structural health monitoring public standard data set. In this embodiment, before step S101, that is, before obtaining a plurality of sample data, the method further includes steps S1011 to S1012:
step S1011: the variable spring simple-supported beam bridge is modeled, the variable spring simple-supported beam bridge is divided into a first set number of first set length simple-supported beam units and a spring unit, the variable spring simple-supported beam bridge is provided with a second set number of random excitation sources, and a third set number of sensors are arranged on the variable spring simple-supported beam bridge at equal intervals and used for detecting acceleration.
Step S1012: and each sensor acquires acceleration data as a structural dynamic response signal according to a set sampling frequency, and Gaussian noise is added to obtain sample data.
Specifically, the simply supported beam bridge is a bridge structure widely applied to medium and small spans, is mainly made of concrete and has a simple structure, so that the simply supported beam bridge is easily deformed and damaged by external force. When a vehicle moves on the bridge in the service process of the simply supported beam bridge, the reaction force response of each support is time-varying and can be changed violently, and the time-varying response is a main cause of support damage. The reasons for such bridge failures are many, including vehicle loading, natural disasters, or improper use of materials. The embodiment is to model the bridge with the structure to obtain data.
Further, in the simply supported beam bridge sensor network, the acquired structural dynamic response signal data include various types, and the data types can be divided into two types: one type is dynamic test data and the other type is static test data. The former type of data includes acceleration, displacement, etc.; the latter kind of data includes static strain, displacement, etc., and the static data is usually not regarded because it cannot acquire modal information (such as vibration mode, damping, natural frequency) of the structure. Therefore, the structural damage identification method based on dynamic testing is mainly adopted in the field of bridge structural damage identification, and the structural dynamic response signal used in the embodiment is an acceleration signal.
Under constant moving loads, the acceleration response of a damaged beam can be assumed to consist of three components: dynamic, static, and corrupt. Taking a single-degree-of-freedom structural system as an example, if the input of a time-varying excitation force or system is f (t), the mass is m, and the time history describing the displacement response of a mass coordinate is x (t), the control equation of forced vibration is formula 1:
wherein the content of the first and second substances,
is a displacement vector at successive time instants t,
representing the first derivative of the displacement vector with respect to time, i.e. the velocity vector, x (t) representing the acceleration vector, f (t) representing the applied external excitation vector.
Natural frequency of the incoming system
And dimensionless viscous damping factor xi c/c of the systemcWherein, in the step (A),
referred to as the critical damping coefficient of the system, equation 1 above can be expressed as equation 2 below:
both ends of the above equation include the acceleration dimension, a (t) ═ f (t)/m, and the characterization force f (t) acts on the free particle m, and the acceleration that can be generated is usually a (t) real function.
The features in the acceleration signal may show damage information of the simple-supported-beam bridge, so that the acceleration signal can be appropriately filtered to highlight the damage component and quantify the severity thereof, and the embodiment will deeply research the acceleration-based simple-supported-beam bridge damage identification model.
And adding bridge main body state information to the sample data formed through actual detection or modeling according to whether the sample data is in a damaged state or not, wherein the sample data is divided into two types of structure damage and structure non-damage.
In step S102, each sample datum is an acceleration signal detected by each sensor in the actual bridge or model at a certain time period, and corresponding bridge body state information. Therefore, the features are extracted in the form of statistical extrema. The extreme value extraction comprises two parts, wherein one part is to count the extreme value aiming at the original structure dynamic response signal, and the other part is to count the extreme value after the noise reduction processing and the fast Fourier transform of the statistical signal characteristic are carried out on the original structure dynamic response signal. The noise reduction processing adopts a mode of deconvolution of maximum correlation kurtosis, so that periodic pulse components can be effectively extracted, and the noise influence of signals is inhibited.
In step S103, a training sample set is formed by adding labels based on the sample data formed in steps S101 and S102.
In step S104, a plurality of preset classification models are trained through the data in the training sample set. Specifically, a support vector machine, a decision tree, a full-connection neural network, a long-short term memory neural network and a self-organizing mapping algorithm network are adopted for classification training, and a corresponding bridge damage identification classification model is formed.
In some embodiments, the predetermined classification model further includes a random forest, a boosted tree, and a gradient boosted decision tree based on an ensemble learning framework in combination with the decision tree. In this embodiment, the integration method is a meta-algorithm that combines several machine learning techniques into a prediction model, so as to achieve the effects of reducing variance (e.g., Bagging frame, Bagging method), deviation (e.g., Boosting frame, weak classifier is assembled into a strong classifier), or improving prediction (Stacking, a model is trained to combine other models). Further, the Boosting framework further includes an adaboost (adaptive Boosting) algorithm and a GBDT (Gradient Boost Decision Tree) algorithm. The adaboost (adaptive boosting) algorithm assigns equal weight to each training example when just starting training, then trains t round with the algorithm to the training set, after each training, assigns larger weight to the training example which fails to train, that is, the learning algorithm is made to pay more attention to the wrong sample after each learning, thereby obtaining a plurality of prediction functions. And gradually reducing the residual error in a residual error fitting mode, and superposing the models generated in each step to obtain a final model. The random forest can be obtained by combining the Bagging framework with the decision tree, and the lifting tree can be obtained by combining the AdaBoost algorithm with the decision tree. Each calculation of the GBDT (gradient Boost Decision tree) algorithm is to reduce the residual error of the previous time, and the GBDT builds a new model in the direction of residual error reduction (negative gradient).
In some embodiments, the method further comprises step S105 and step S106:
step S105: and acquiring a nearest neighbor algorithm model to load a nearest neighbor algorithm for classification calculation.
Step S106: and calculating an evaluation index of the nearest algorithm model, and integrating the nearest algorithm model serving as a bridge damage identification classification model into the bridge damage detection module so as to select one or combine multiple bridge damage identification classification models according to the evaluation index to perform bridge damage detection.
In step S105 and step S106, unlike the other models in step S104, the nearest algorithm model does not require additional training, and determines the category of the current sample according to the category of the k samples closest to the data to be evaluated, where the distance calculation methods are various, the common distance calculation method is the euclidean distance, and the nearest algorithm is simple, but the generalization error rate thereof is not more than twice the error rate of the bayesian optimal classifier.
In step S105, the method calculates an evaluation index of each bridge damage identification classification model by means of 5-fold cross validation. And randomly dividing the sample data into 5 equal parts, wherein 4 parts are used as training sets for training each preset classification model, and the remaining 1 part is used for testing, calculating and evaluating indexes.
In some embodiments, step S106 is to integrate the bridge damage identification and classification models into a bridge damage detection module, so as to select one or more bridge damage identification and classification models according to the evaluation index for bridge damage detection, including step S1061 and step S1062:
step S1061: and carrying out normalization processing on the accuracy, precision ratio, recall ratio and F1 score in the evaluation indexes, weighting and summing to obtain a comprehensive score of each bridge damage identification classification model, and arranging according to the comprehensive score from high to low.
Step S1062: selecting a bridge damage identification classification model with the highest comprehensive score according to the first probability to identify the bridge damage; or selecting a fourth set number of bridge damage identification classification models with higher comprehensive scores according to the second probability to identify the bridge damages, wherein the fourth set number is an odd number, and taking the bridge damage identification results with larger numbers as final results.
In other embodiments, a model may be trained to combine other models by improving the prediction, the output of each bridge damage recognition classification model is used as the input of a new model, and the final output is obtained by training the new model. And combining all the bridge damage identification classification models and the new model to form a final bridge damage detection module.
On the other hand, the invention also provides an intelligent online bridge health detection method, as shown in fig. 2, comprising steps S201 to S203:
step S201: and acquiring an acceleration data sequence detected by a third set number of sensors on the bridge to be detected at the current moment, wherein the sensors are arranged at equal intervals along the bridge to be detected.
Step S202: counting original extreme value sequences to be detected of acceleration data acquired by all sensors in the acceleration data sequence; and after noise reduction processing is carried out on the acceleration data sequence by adopting maximum correlation kurtosis deconvolution, the acceleration data sequence characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence to be detected of the acceleration data sequence characteristics is obtained.
Step S203: and combining the original extreme value sequence to be detected and the characteristic extreme value sequence to be detected, and inputting the combined sequence into the bridge damage detection module in the bridge health online detection module generation method in the steps S101 to S106 to obtain a damage detection result of the bridge to be detected.
On the other hand, the invention also provides an intelligent online detection tool box for bridge health, which at least comprises: the device comprises a data acquisition module and a detection module.
And the data acquisition module is used for acquiring the acceleration data sequence detected by a third set number of sensors on the bridge to be detected at the current moment.
And the detection module is used for loading the bridge damage detection module in the bridge health online detection module generation method in the steps S101 to S106, and executing the bridge health intelligent online detection method in the steps S201 to S202 to obtain a damage detection result of the bridge to be detected.
In some embodiments, the kit further comprises a display module, wherein the display module is connected with the detection module and is used for visually presenting the acceleration data sequence and the damage detection result.
In another aspect, the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the method are implemented.
The invention is illustrated below with reference to specific examples:
in the embodiment, the collected dynamic response data of the simply supported girder bridge structure are classified by using an artificial intelligence algorithm, so that the simply supported girder bridge structure is diagnosed or damage is predicted. The most common algorithms in the traditional machine learning, such as a support vector machine, a decision tree, discriminant analysis and a k-nearest neighbor algorithm, and a higher-level deep learning algorithm, such as a fully-connected neural network and a long-short term memory neural network, are adopted, and the neural network structure can automatically extract high-level features without manually extracting the features.
The core of the support vector machine is to project the features of the data to a high dimension, then find a hyperplane, and segment different classes of data points, and the greater the separation degree, the better. To find the optimal parameters of SVM, GridSearchCV method is adopted, and its unraveling is Grid Search (GS) and Cross Validation (CV). The parameters searched by the grid search method are adjusted according to the step length in a certain interval, the parameters with the highest precision on the verification set are obtained by utilizing the training learner after the parameters are adjusted, and the principle of the method is similar to that of finding the maximum value in an array. In the field of structural damage recognition, a support vector machine is often used for damage recognition to judge the health state of a structure, and different SVMs according to a kernel function can be classified into a Linear SVM and a Cubic SVM.
The decision tree model comprises a classic ID3 algorithm, an improved C4.5 algorithm and a powerful CART algorithm, the DT model represents a mapping between object attributes and object values, each node in the tree represents a judgment condition of the object attributes, branches of the tree represent objects meeting the node conditions, and leaf nodes of the tree represent prediction results to which the objects belong. And continuously classifying the nodes to obtain a final diagnosis result.
The nearest neighbor algorithm KNN is one of the simplest classification methods in a machine learning method, the KNN determines the category of the current sample according to the category of the nearest k samples, the calculation methods of the distances are various, the common distance calculation method is the Euclidean distance, and although the KNN method is simple, the generalization error rate of the KNN method is not more than twice of the error rate of a Bayesian optimal classifier. Among them, the Fine KNN algorithm is commonly used.
The fully-connected neural network comprises an input layer, a plurality of hidden layers and an output layer. When training the fully-connected neural network, firstly initializing neuron parameters of each layer, then transmitting sample data to an output layer after layer-by-layer calculation, and finally, training the neural network is supervised learning, namely, an input X has a real value Y corresponding to the input X, and Loss between the output Y' of the neural network and the real value Y is used for adjusting parameters of network back propagation.
The long-short term Neural Network (LSTM) is one of the time cycle Neural networks (RNN), and the cycle Neural Network (RNN) is a Neural Network for processing sequence data, and can process data with sequence change compared with a general Neural Network, but the general cycle Neural Network has a long-term dependence problem, and in order to specially solve the problem, an LSTM structure can be introduced.
Further, the present embodiment also introduces an Ensemble Learning method (Ensemble Learning), also called multi-classifier system, which completes the Learning task by combining a plurality of weak learners, and if used for the classification task, is called Ensemble Classifiers (Ensemble Classifiers). The current ensemble learning algorithms can be classified into the following categories: boosting (algorithm to promote weak learner to strong learner) class; bagging (a representation of parallel ensemble learning methods) and random forest (RF for short, an extended variant of Bagging) classes. Common ensemble classifiers include Boosted Trees, Bagged Trees, Subspace Discriminiant and Subspace kNN.
In this embodiment, the adopted data set is derived from a public standard data set for structural health monitoring provided by European works hop, the data set is derived from a 1.4m simple beam, the cross section of the simple beam is a uniform rectangle with 50mm × 5mm, as shown in fig. 3, a spring 612.5mm away from a support is arranged below the simple beam to form a variable spring simple beam bridge, and a spring constant k and temperature have a nonlinear relationship, as shown in the following formula 3:
k＝k0+aT3； (3)
wherein k is0100kN/m, a-0.8 (band compatible units), so that T is uniformly randomly distributed between-20 ℃ and +40 ℃ taking into account seasonal variations in temperature T.
The beam has three equal sections, Young's modulus E in section iiWith corresponding independent dimensionless environment variables ZiThe following equation 4 is shown in a linear relationship:
Ei＝E0+σiZi(i＝1,2,3)； (4)
wherein E is0＝207GPa，ZiIs a normalized gaussian variable: ziN (0, 1), and the standard deviation σ of the different segmentsiComprises the following steps: sigma1＝5GPa，σ2＝3GPa，σ3＝7GPa。
In the analysis and data acquisition process, a variable spring simply supported beam bridge and three independent loads (namely, excitation sources) are arranged, the structure of the variable spring simply supported beam bridge is divided into 144 simply supported beam units and one spring unit for modeling, the independent random excitation sources with different amplitudes excite at three position points, and the vibration response is analyzed by using a vibration mode superposition method of a static correction program. The lateral acceleration is measured at 47 equidistant points along the beam, with a standard deviation σ of 0.01m/s for each sensor2The average noise level of the Gaussian noise is about 1 percent of the signal, the sampling frequency is 571 Hz, the number of sequences of data measured each time is 2859, and the external environmentIs slowly varying and is basically considered to be performed in a constant environment. The first 50 measurements were from undamaged structures and the last 50 measurements were from damaged structures. There are five different levels of damaged beam height: 4.5, 4, 3.5, 3 and 2.5mm, the extent of each injury is composed of 10 different measurements.
In summary, there are three factors that affect the external force variation: (1) a spring having a non-linear relationship between temperature and spring rate; (2) independently varying young's modulus of the three regions; (3) the random load at three points is distributed.
The data collected form a data set containing 100 files, each file containing a variable y (47x2859 matrix) representing 47 accelerometers and 2859 time series. The first 50 files are uncorrupted (labeled as unidamaged) samples, and the last 50 files are corrupted (labeled as damaged) samples.
The method comprises the steps of firstly extracting an extreme value of an original signal, carrying out filtering and noise reduction by adopting an MCKD algorithm, then carrying out fast Fourier transform on the noise-reduced signal, extracting extreme value characteristics, combining the characteristics and the extreme value of the original signal into a new data set, and adding damaged or undamaged signals as labels to form a training sample set. Finally, inputting the training sample set into the machine learning algorithm adopted in the embodiment to perform classification training on the samples, and establishing a simply supported bridge damage identification model according to the classification training, wherein the machine learning algorithm comprises the following steps: support vector machine, decision tree, full-connection neural network, long-short term memory neural network and self-organizing mapping algorithm network. Specifically, fineTree, Linear Discriminiant, Logistic Regression, Linear SVM, Cubic SVM, Fine kNN, Subspace kNN and Subspace Discriminiant are adopted. The main process is shown in fig. 4.
For example, as shown in FIG. 5, the acceleration data of the first sensor in the first sample is significantly fluctuating.
And (4) counting the data extreme value of each sensor in each sample, and obtaining 94 characteristics comprising 47 acceleration maximum values and 47 acceleration minimum values of each sensor in each sample. And one hundred samples can be obtained, containing 50 undamaged, 50 damaged samples. A temporary data set (hereinafter referred to as an old data set) is obtained, and partial data is as follows
Shown in table 1.
Table 1 temporary data set partial data
In order to effectively detect the pulse wave of the signal, the embodiment first of all reduces noise of original data by using an MCKD algorithm, selects an optimal parameter as an experimental parameter by using a fixed-step search method, and sets the parameter: the filter length is 400 and the impulse signal period is 300, with best results, and the results are shown in fig. 6. The spectrum of the signal is then analyzed or features of the signal are extracted by means of signal processing. The invention adopts Fast Fourier Transform (FFT) to extract the characteristics of the signal, and can directly obtain each frequency spectrum component of the waveform, thereby being a powerful tool for analyzing the harmonic waveform. The FFT of a sensor acceleration signal for a sample is characterized as shown in fig. 7.
It is clear that the curve has a plurality of extrema and that the extrema fall mainly within the frequency range 0, 80, so that the range is divided into 7 non-uniform cells, one extremum point per statistic is taken as the abscissa, and if there is no extrema in a cell, the last extremum is used instead. Thus, there are 7 additional features per sample, which are added after the old data set to form a new data set containing 101 features, with the new data set portion data shown in table 2 below.
TABLE 2 New data set part data
Performing learning classification by using a MATLAB platform to run FineTree, Linear cognitive, Logistic Regression, Linear SVM, Cubic SVM, Fine kNN, Subspace kNN and Subspace cognitive:
first, under 5-fold cross validation, the old data set was classified using a machine learning algorithm, and the results are shown in table 3 below.
TABLE 3 Classification Effect of traditional machine learning Algorithm on old data sets
Under 5-fold cross validation, the new data set was classified using a machine learning algorithm, with the results shown in table 4 below:
TABLE 4 classification effect of machine learning method on New data sets
Wherein, the algorithm of the linear support vector machine sets parameters: the effect is best when the Kernel function is Linear, the Kernel scale is Automatic, the Box constraint level is 1, and the Multiclass method is One-vs-One, and the accuracy reaches 92%.
Further, evaluation indexes of the bridge damage recognition classification model obtained after FineTree, Linear classifier, Logistic Regression, Linear SVM, Cubic SVM, Fine kNN, Subspace kNN and Subspace classifier training are calculated, wherein the evaluation indexes comprise accuracy, precision, recall and F1 score.
In the binary problem, it is assumed that the samples contain a total of two classes: positive and Negative, as shown in FIG. 8, when the classifier prediction is over, a Confusion Matrix (fusion Matrix) is obtained.
Thus, in the binary model, the mathematical expressions of Accuracy, Precision, Recall, and F1 Score (F1 Score) are:
and integrating the bridge damage identification and classification models into a bridge damage detection module so as to select one or combine multiple bridge damage identification and classification models according to the evaluation index to perform bridge damage detection.
And carrying out normalization processing on the accuracy, precision ratio, recall ratio and F1 score in the evaluation indexes, weighting and summing to obtain a comprehensive score of each bridge damage identification classification model, and arranging according to the comprehensive score from high to low. Selecting a bridge damage identification classification model with the highest comprehensive score according to the first probability to identify the bridge damage; or selecting a fourth set number of bridge damage identification classification models with higher comprehensive scores according to the second probability to identify the bridge damages, wherein the fourth set number is an odd number, and taking the bridge damage identification results with larger numbers as final results. Meanwhile, one of the bridge damage identification classification models can be freely called to identify the bridge damage.
The embodiment further provides a tool kit, which is based on the operation of a computer, a single chip microcomputer or other electronic devices capable of being used for storing and operating programs, and is used for building the machine learning tool kit of the bridge health online intelligent monitoring system, as shown in fig. 9, and the tool kit comprises a data importing module, a data preprocessing module, a data visualization module and a training machine learning module.
The data importing module is used for importing sample data.
The preprocessing data module is used for counting original extreme value sequences of the structural dynamic response signals acquired by all the sensors in each sample data; and after noise reduction processing is carried out on the structural dynamic response signals acquired by each sensor in each sample data by adopting maximum correlation kurtosis deconvolution, signal characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence of the signal characteristics is counted. And merging the original extreme value sequence and the characteristic extreme value sequence corresponding to each piece of sample data, and adding a label to each piece of sample data according to the bridge main body state information corresponding to each piece of sample data to form a training sample set.
And the data visualization module is used for visually presenting the data processing result in the forms of multi-dimensional analysis, data drawing and animation display.
The training machine learning module is used for training a plurality of preset classification models by adopting the training sample set to obtain a plurality of bridge damage identification classification models; the preset classification model at least comprises: the method comprises the steps that a support vector machine, a decision tree, a full-connection neural network, a long-term and short-term memory neural network and a self-organizing mapping algorithm network are used for calculating evaluation indexes of each bridge damage identification classification model, wherein the evaluation indexes comprise accuracy, precision, recall and F1 scores; and integrating the bridge damage identification and classification models into a bridge damage detection module so as to select one or combine multiple bridge damage identification and classification models according to the evaluation index to perform bridge damage detection.
In summary, in the bridge health online detection module generation method, the bridge health online detection method, the tool box and the device, the acceleration is acquired by the sensor and is used as the structural dynamic response signal, so as to acquire the characteristics of the bridge in dynamic, static and damaged states. Furthermore, extreme value statistics is carried out on the original data of the structural response data, the extreme value statistics is carried out after noise reduction and fast Fourier transform of the structural response data are further carried out, the two statistics results are combined to obtain data characteristics, and the damage characteristics of the bridge can be accurately captured. The bridge damage identification classification model capable of completing bridge damage identification classification is obtained by training and learning through various machine learning algorithm models, and the model is integrated into a bridge damage detection module and can be freely called. Meanwhile, the evaluation indexes of the bridge damage identification classification models are calculated to guide the selection of the optimal or various bridge damage detection modules for identification detection, so that the damage identification precision is improved. Based on the full-automatic intelligent processing mode, the speed and the precision of detecting the bridge damage are greatly improved, and all-weather real-time detection is realized.
Those of ordinary skill in the art will appreciate that the various illustrative components, systems, and methods described in connection with the embodiments disclosed herein may be implemented as hardware, software, or combinations of both. Whether this is done in hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention. When implemented in hardware, it may be, for example, an electronic circuit, an Application Specific Integrated Circuit (ASIC), suitable firmware, plug-in, function card, or the like. When implemented in software, the elements of the invention are the programs or code segments used to perform the required tasks. The program or code segments may be stored in a machine-readable medium or transmitted by a data signal carried in a carrier wave over a transmission medium or a communication link. A "machine-readable medium" may include any medium that can store or transfer information. Examples of a machine-readable medium include electronic circuits, semiconductor memory devices, ROM, flash memory, Erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, Radio Frequency (RF) links, and so forth. The code segments may be downloaded via computer networks such as the internet, intranet, etc.
It should also be noted that the exemplary embodiments mentioned in this patent describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above-described steps, that is, the steps may be performed in the order mentioned in the embodiments, may be performed in an order different from the order in the embodiments, or may be performed simultaneously.
Features that are described and/or illustrated with respect to one embodiment may be used in the same way or in a similar way in one or more other embodiments and/or in combination with or instead of the features of the other embodiments in the present invention.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes may be made to the embodiment of the present invention by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A bridge health online detection module generation method is characterized by comprising the following steps:
obtaining a plurality of sample data, wherein each sample data at least comprises a structural dynamic response signal of a sensor arranged at each detection point on a bridge main body and bridge main body state information, the structural dynamic response signal is an acceleration signal of a dynamic test, and the bridge main body state information comprises two types of structural damage and structural non-damage;
counting original extreme value sequences of the structural dynamic response signals acquired by all the sensors in each sample data; after noise reduction processing is carried out on the structural dynamic response signals collected by each sensor in each sample data by adopting maximum correlation kurtosis deconvolution, signal characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence of the signal characteristics is counted;
merging the original extreme value sequence and the characteristic extreme value sequence corresponding to each sample data, and adding a label to each sample data according to the bridge main body state information corresponding to each sample data to form a training sample set;
training a plurality of preset classification models by adopting the training sample set to obtain a plurality of bridge damage identification classification models; the preset classification model at least comprises: the system comprises a support vector machine, a decision tree, a full-connection neural network, a long-term and short-term memory neural network and a self-organizing mapping algorithm network;
calculating evaluation indexes of each bridge damage identification classification model, wherein the evaluation indexes comprise accuracy, precision, recall and F1 scores;
and integrating the bridge damage identification and classification models into a bridge damage detection module so as to select one or combine multiple bridge damage identification and classification models according to the evaluation index to perform bridge damage detection.
2. The method for generating the bridge health online detection module according to claim 1, wherein the preset classification model further comprises a random forest, a lifting tree and a gradient lifting decision tree which are obtained based on an ensemble learning method framework and a decision tree.
3. The method for generating the bridge health online detection module according to claim 1, further comprising:
acquiring a nearest algorithm model, and loading a nearest algorithm for classification calculation;
and calculating the evaluation index of the nearest algorithm model, and integrating the nearest algorithm model serving as a bridge damage identification classification model into the bridge damage detection module so as to select one or more bridge damage identification classification models according to the evaluation index for bridge damage detection.
4. The method for generating the bridge health online detection module according to claim 1, further comprising, before acquiring a plurality of sample data:
modeling a variable spring simply supported beam bridge, wherein the variable spring simply supported beam bridge is divided into a first set number of simply supported beam units with a first set length and a spring unit, the variable spring simply supported beam bridge is provided with a second set number of random excitation sources, and a third set number of sensors for detecting acceleration are equidistantly arranged on the variable spring simply supported beam bridge;
and each sensor acquires acceleration data as the structural dynamic response signal according to a set sampling frequency, and Gaussian noise is added to obtain the sample data.
5. The method for generating the bridge health online detection module according to claim 1, wherein the method adopts a 5-fold cross validation mode to calculate the evaluation index of each bridge damage identification classification model.
6. The method for generating the bridge health online detection module according to claim 5, wherein the bridge damage identification classification models are integrated into a bridge damage detection module, so as to select one or more bridge damage identification classification models for bridge damage detection according to the evaluation index, and the method comprises the following steps:
carrying out normalization processing on the accuracy, the precision ratio, the recall ratio and the F1 score in the evaluation index, then carrying out weighted summation to obtain a comprehensive score of each bridge damage identification classification model, and arranging according to the comprehensive score from high to low;
selecting a bridge damage identification classification model with the highest comprehensive score according to the first probability to identify the bridge damage; or selecting a fourth set number of bridge damage identification classification models with higher comprehensive scores according to the second probability to identify the bridge damages, wherein the fourth set number is an odd number, and taking the bridge damage identification results with larger numbers as final results.
7. An intelligent online bridge health detection method is characterized by comprising the following steps:
acquiring an acceleration data sequence detected by a third set number of sensors on the bridge to be detected at the current moment, wherein the sensors are arranged at equal intervals along the bridge to be detected;
counting to-be-detected original extreme value sequences of the acceleration data acquired by all the sensors in the acceleration data sequence; after noise reduction processing is carried out on the acceleration data sequence by adopting maximum correlation kurtosis deconvolution, the acceleration data sequence characteristics are extracted through fast Fourier transform, and a characteristic extreme value sequence to be detected of the acceleration data sequence characteristics is obtained;
combining the original extreme value sequence to be detected and the characteristic extreme value sequence to be detected, and inputting the combined sequence into the bridge damage detection module in the bridge health online detection module generation method according to any one of claims 1 to 6, so as to obtain a damage detection result of the bridge to be detected.
8. The utility model provides a healthy intelligent on-line measuring toolbox of bridge which characterized in that includes at least:
the data acquisition module is used for acquiring an acceleration data sequence detected by a third set number of sensors on the bridge to be detected at the current moment;
the detection module is used for loading the bridge damage detection module in the bridge health online detection module generation method according to any one of claims 1 to 6, and executing the bridge health intelligent online detection method according to claim 7 to obtain a damage detection result of the bridge to be detected.
9. The intelligent online bridge health detection kit of claim 8, further comprising a display module, wherein the display module is connected to the detection module and is configured to visually present the acceleration data sequence and the damage detection result.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method according to any of claims 1 to 7 are implemented when the processor executes the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111592343.8A CN114563150B (en) | 2021-12-23 | 2021-12-23 | Bridge health online detection module generation method, detection method, tool box and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111592343.8A CN114563150B (en) | 2021-12-23 | 2021-12-23 | Bridge health online detection module generation method, detection method, tool box and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114563150A true CN114563150A (en) | 2022-05-31 |
| CN114563150B CN114563150B (en) | 2023-04-07 |
Family
ID=81711574
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111592343.8A Active CN114563150B (en) | 2021-12-23 | 2021-12-23 | Bridge health online detection module generation method, detection method, tool box and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114563150B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116642947A (en) * | 2023-05-26 | 2023-08-25 | 广东科伟工程检测有限公司 | Bridge vibration defect detection method, system and terminal based on optical fiber sensor |
| CN116804534A (en) * | 2023-06-29 | 2023-09-26 | 中交第二航务工程局有限公司 | Auxiliary determination method for continuous beam bridge construction process |
| CN116861292A (en) * | 2023-07-10 | 2023-10-10 | 哈尔滨工业大学 | Real-time identification method for road monitoring data loaded dynamic response |
| CN117688480A (en) * | 2024-02-04 | 2024-03-12 | 四川华腾公路试验检测有限责任公司 | Bridge damage identification method based on damage frequency panorama and random forest |
| CN117723782A (en) * | 2024-02-07 | 2024-03-19 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117933730A (en) * | 2024-03-21 | 2024-04-26 | 四川华腾公路试验检测有限责任公司 | Comprehensive bridge safety assessment method based on random forest |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108170925A (en) * | 2017-12-21 | 2018-06-15 | 中国铁道科学研究院铁道建筑研究所 | A kind of bridge damnification method for early warning based on arma modeling |
| CN110060368A (en) * | 2019-04-22 | 2019-07-26 | 苏州大学 | Mechanical method for detecting abnormality based on potential feature coding |
| CN112528564A (en) * | 2020-12-08 | 2021-03-19 | 中国矿业大学 | Bridge structure damage identification method based on sparse Bayesian learning |
| CN113553749A (en) * | 2021-09-22 | 2021-10-26 | 深圳市城市交通规划设计研究中心股份有限公司 | Bridge health monitoring method, system, computer and storage medium |
-
2021
- 2021-12-23 CN CN202111592343.8A patent/CN114563150B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108170925A (en) * | 2017-12-21 | 2018-06-15 | 中国铁道科学研究院铁道建筑研究所 | A kind of bridge damnification method for early warning based on arma modeling |
| CN110060368A (en) * | 2019-04-22 | 2019-07-26 | 苏州大学 | Mechanical method for detecting abnormality based on potential feature coding |
| CN112528564A (en) * | 2020-12-08 | 2021-03-19 | 中国矿业大学 | Bridge structure damage identification method based on sparse Bayesian learning |
| CN113553749A (en) * | 2021-09-22 | 2021-10-26 | 深圳市城市交通规划设计研究中心股份有限公司 | Bridge health monitoring method, system, computer and storage medium |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116642947A (en) * | 2023-05-26 | 2023-08-25 | 广东科伟工程检测有限公司 | Bridge vibration defect detection method, system and terminal based on optical fiber sensor |
| CN116642947B (en) * | 2023-05-26 | 2024-06-07 | 广东科伟工程检测有限公司 | Bridge vibration defect detection method, system and terminal based on optical fiber sensor |
| CN116804534A (en) * | 2023-06-29 | 2023-09-26 | 中交第二航务工程局有限公司 | Auxiliary determination method for continuous beam bridge construction process |
| CN116804534B (en) * | 2023-06-29 | 2024-03-12 | 中交第二航务工程局有限公司 | Auxiliary determination method for continuous beam bridge construction process |
| CN116861292A (en) * | 2023-07-10 | 2023-10-10 | 哈尔滨工业大学 | Real-time identification method for road monitoring data loaded dynamic response |
| CN117688480A (en) * | 2024-02-04 | 2024-03-12 | 四川华腾公路试验检测有限责任公司 | Bridge damage identification method based on damage frequency panorama and random forest |
| CN117688480B (en) * | 2024-02-04 | 2024-05-14 | 四川华腾公路试验检测有限责任公司 | Bridge damage identification method based on damage frequency panorama and random forest |
| CN117723782A (en) * | 2024-02-07 | 2024-03-19 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117723782B (en) * | 2024-02-07 | 2024-05-03 | 山东大学 | Sensor fault identification positioning method and system for bridge structure health monitoring |
| CN117933730A (en) * | 2024-03-21 | 2024-04-26 | 四川华腾公路试验检测有限责任公司 | Comprehensive bridge safety assessment method based on random forest |
| CN117933730B (en) * | 2024-03-21 | 2024-06-07 | 四川华腾公路试验检测有限责任公司 | Comprehensive bridge safety assessment method based on random forest |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114563150B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114563150B (en) | Bridge health online detection module generation method, detection method, tool box and device | |
| Smarsly et al. | Machine learning techniques for structural health monitoring | |
| Zheng et al. | A rolling bearing fault diagnosis method based on multi-scale fuzzy entropy and variable predictive model-based class discrimination | |
| Kankar et al. | Rolling element bearing fault diagnosis using wavelet transform | |
| CN102265227B (en) | Method and apparatus for creating state estimation models in machine condition monitoring | |
| CN112660745B (en) | Intelligent diagnosis method and system for carrier roller fault and readable storage medium | |
| Parisi et al. | Automated location of steel truss bridge damage using machine learning and raw strain sensor data | |
| Joshuva et al. | Crack detection and localization on wind turbine blade using machine learning algorithms: a data mining approach | |
| CN114169374B (en) | Cable-stayed bridge stay cable damage identification method and electronic equipment | |
| CN116380445B (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN114357594A (en) | Bridge abnormity monitoring method, system, equipment and storage medium based on SCA-GRU | |
| CN113901977A (en) | Deep learning-based power consumer electricity stealing identification method and system | |
| Arul et al. | Data anomaly detection for structural health monitoring of bridges using shapelet transform | |
| Son et al. | Deep learning-based anomaly detection to classify inaccurate data and damaged condition of a cable-stayed bridge | |
| CN111538311A (en) | Flexible multi-state self-adaptive early warning method and device for mechanical equipment based on data mining | |
| Ye et al. | A deep learning-based method for automatic abnormal data detection: Case study for bridge structural health monitoring | |
| CN112682273A (en) | Wind turbine generator fault detection method based on cost-sensitive lightweight gradient elevator | |
| CN117824756B (en) | Building structure monitoring method, equipment and storage medium | |
| CN110956331A (en) | Method, system and device for predicting operation state of digital factory | |
| CN109840386B (en) | Damage identification method based on factor analysis | |
| CN114638039B (en) | Structural health monitoring characteristic data interpretation method based on low-rank matrix recovery | |
| CN116204825A (en) | Production line equipment fault detection method based on data driving | |
| Purarjomandlangrudi et al. | Application of anomaly technique in wind turbine bearing fault detection | |
| Vagnoli et al. | A machine learning classifier for condition monitoring and damage detection of bridge infrastructure | |
| CN114429032A (en) | Bridge health online intelligent monitoring system |
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 | ||
| GR01 | Patent grant |