CN115032270A - Method and device for quantitatively identifying damage state of building curtain wall based on machine learning algorithm - Google Patents
Method and device for quantitatively identifying damage state of building curtain wall based on machine learning algorithm Download PDFInfo
- Publication number
- CN115032270A CN115032270A CN202210617762.0A CN202210617762A CN115032270A CN 115032270 A CN115032270 A CN 115032270A CN 202210617762 A CN202210617762 A CN 202210617762A CN 115032270 A CN115032270 A CN 115032270A
- Authority
- CN
- China
- Prior art keywords
- curtain wall
- damage
- wall panel
- dynamic response
- response signal
- 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
- 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
- G01N29/045—Analysing solids by imparting shocks to the workpiece and detecting the vibrations or the acoustic waves caused by the shocks
-
- 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
- G01N29/12—Analysing solids by measuring frequency or resonance of acoustic waves
-
- 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
- G01N29/22—Details, e.g. general constructional or apparatus details
-
- 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)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Acoustics & Sound (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a method and a device for quantitatively identifying damage states of a building curtain wall based on a machine learning algorithm, and relates to the technical field of safety detection of the building curtain wall. The method comprises the following steps: acquiring a dynamic response signal of a curtain wall panel to be detected; inputting the dynamic response signal into a constructed support vector machine model optimized based on a wolf algorithm; and obtaining a quantitative identification result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm. The invention can provide an efficient, accurate, digital and intelligent quantitative identification technology for the damage state of the building curtain wall, and can solve the problems that the existing detection method is inaccurate in evaluation, cannot synchronously realize identification of the damage degree and the damage position, and cannot quantitatively identify the damage state.
Description
Technical Field
The invention relates to the technical field of building curtain wall safety detection, in particular to a method and a device for quantitatively identifying a damage state of a building curtain wall based on a machine learning algorithm.
Background
The building curtain wall is introduced from the 80 th century in China, and has become the first major producing and using countries in the world by the beginning of the 21 st century, and the equivalent of the existing building curtain wall is huge. Along with the increase of service life and the influence of environmental erosion, the potential safety hazard problem of the existing building curtain wall is prominent, and the disaster accidents caused by the falling-off of the curtain wall panel are frequent, so that the curtain wall panel becomes an important problem influencing the social life.
In the aspect of the latest curtain wall safety detection theory and technology research, a detection method based on vibration and thermal waves is being widely researched and popularized. Huang, Pan and the like propose detection methods based on indexes such as inherent frequency, relative accumulated error of an origin acceleration frequency response function, vibration transfer rate and the like, but the methods are only methods for judging damage tendency and qualitatively evaluating, screen wall panels with large damage are briefly identified according to relative comparison of the detected indexes, and accurate and quantitative judgment of the damage cannot be realized. In addition, L i n and the like utilize scanning type laser deep heating equipment to heat the hidden frame glass curtain wall, and the structural adhesive damage position is identified according to the temperature difference between the damage position and the non-damage position acquired by the thermal infrared imager, but the damage degree evaluation is not involved, and the method is only suitable for the building curtain wall connected by the structural adhesive.
In addition, according to recent studies of Pan and the like, the evaluation method based on the natural frequency has disadvantages such as insensitivity to small damage, insufficient measurement accuracy, and influence of damage to the measurement result. The detection method based on the relative accumulated error of the origin acceleration frequency response function is contact detection, and cannot realize remote measurement. Based on the analysis, the invention provides a method for quantitatively identifying the damage state of the building curtain wall based on a machine learning algorithm, aiming at the problems that the current curtain wall detection technology is laggard and the theory is incomplete.
Disclosure of Invention
The invention provides a method for detecting a curtain wall, which aims at solving the problems that the existing curtain wall detection technology is laggard and the detection theory is incomplete.
In order to solve the technical problems, the invention provides the following technical scheme:
in one aspect, the invention provides a method for quantitatively identifying damage states of building curtain walls based on a machine learning algorithm, which is realized by electronic equipment and comprises the following steps:
and S1, acquiring a dynamic response signal of the curtain wall panel to be tested.
And S2, inputting the dynamic response signal to a well-constructed support vector machine model optimized based on the gray wolf algorithm.
And S3, obtaining a quantitative recognition result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm.
Optionally, the building process of the support vector machine model based on the grayish wolf algorithm optimization in S2 includes:
s21, acquiring a power response signal of the curtain wall panel; the dynamic response signals comprise dynamic response signals in plates of the curtain wall panel under the working conditions of different damage degrees and dynamic response signals of 4 corners of the curtain wall panel.
S22, calculating damage identification indexes of the curtain wall panels under the working conditions of different damage degrees based on the dynamic response signals; the damage identification index comprises a damage degree identification index relative natural frequency, a damage position identification index relative vibration variance and a damage position identification index relative variation coefficient.
S23, constructing an intelligent classification database based on the damage identification indexes.
S24, constructing a support vector machine model optimized by the wolf algorithm based on the intelligent classification database.
Alternatively, the method of calculating the relative natural frequency of the damage degree identification index in S22 is as shown in the following formula (1):
in the formula, X f For a plurality of different damage degreesRelative natural frequency corresponding to any damage degree in the working conditions;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the damage working condition f;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the working condition that the damage degree is not damaged; and i is the number of the signal acquisition position and has the value range of 1,2,3,4 and 5.
Alternatively, the calculation method of the relative vibration variance of the damage location identification index in S22 is as shown in the following formula (2):
in the formula (I), the compound is shown in the specification,the relative vibration variance of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
Alternatively, the method for calculating the relative variation coefficient of the damage location identification index in S22 is as shown in the following formula (3):
in the formula (I), the compound is shown in the specification,the relative variation coefficient of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
Optionally, the obtaining the dynamic response signal of the curtain wall panel in S21 includes:
and acquiring a dynamic response signal of the curtain wall panel by using a laser Doppler vibration meter.
Optionally, the intelligent classification database is represented by the following formulas (4), (5):
[Y]=[Y 1 ,Y 2 ,Y 3 ,Y 4 ] (5)
in the formula, [ X ]]A feature vector, X, corresponding to any one of a plurality of different damage-level conditions f Representing the relative natural frequency corresponding to the working condition;curtain wall panel respectively representing working conditionsRelative vibration variance at 4 corners;respectively representing the relative variation coefficients corresponding to the 4 corners of the curtain wall panel corresponding to the working conditions.
[Y]Outputting vectors for the intelligent classification database; y is 1 、Y 2 、Y 3 、Y 4 The 4 corner bolt looseness degrees of the curtain wall panel are respectively represented, and the value is 1,2,3, … and N.
Optionally, the obtaining of the result of quantitatively identifying the damage state of the curtain wall panel to be tested according to the dynamic response signal and the support vector machine model optimized based on the grayling algorithm in S3 includes:
and generating a maximum interval hyperplane between the working conditions of different damage degrees according to the dynamic response signal and a support vector machine model optimized based on a wolf algorithm, and judging the working condition of the curtain wall panel to be detected according to the maximum interval hyperplane to obtain a quantitative identification result of the damage state of the curtain wall panel to be detected.
On the other hand, the invention provides a device for quantitatively identifying the damage state of a building curtain wall based on a machine learning algorithm, which is applied to a method for quantitatively identifying the damage state of the building curtain wall based on the machine learning algorithm, and comprises the following steps:
and the acquisition module is used for acquiring the power response signal of the curtain wall panel to be detected.
And the input module is used for inputting the dynamic response signal to the constructed support vector machine model optimized based on the gray wolf algorithm.
And the output module is used for obtaining a quantitative recognition result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm.
Optionally, the input module is further configured to:
s21, acquiring a power response signal of the curtain wall panel; the dynamic response signals comprise dynamic response signals in plates of the curtain wall panel under the working conditions of different damage degrees and dynamic response signals of 4 corners of the curtain wall panel.
S22, calculating damage identification indexes of the curtain wall panels under the working conditions of different damage degrees based on the dynamic response signals; the damage identification index comprises a damage degree identification index relative natural frequency, a damage position identification index relative vibration variance and a damage position identification index relative variation coefficient.
S23, constructing an intelligent classification database based on the damage identification indexes.
S24, constructing a support vector machine model optimized by the wolf algorithm based on the intelligent classification database.
Alternatively, the calculation method of the damage degree identification index relative to the natural frequency is as shown in the following formula (1):
in the formula, X f The relative natural frequency corresponding to any damage degree working condition in a plurality of working conditions with different damage degrees;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the damage working condition f;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the working condition that the damage degree is not damaged; i is the number of the signal acquisition position, and the value range is 1,2,3,4 and 5.
Alternatively, the calculation method of the damage location identification index relative vibration variance is as shown in the following formula (2):
in the formula (I), the compound is shown in the specification,the relative vibration variance of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the acquired vibration speed time-course signal.
Alternatively, the method for calculating the relative variation coefficient of the damage location identification index is as shown in the following formula (3):
in the formula (I), the compound is shown in the specification,the relative variation coefficient of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
Optionally, the input module is further configured to:
and acquiring a dynamic response signal of the curtain wall panel by using a laser Doppler vibration meter.
Optionally, the intelligent classification database is represented by the following formulas (4), (5):
[Y]=[Y 1 ,Y 2 ,Y 3 ,Y 4 ] (5)
wherein [ X ]]Feature vector, X, corresponding to any one of a plurality of different damage levels f Representing the relative natural frequency corresponding to the working condition;respectively representing relative vibration variances at 4 corners of the curtain wall panel corresponding to the working conditions;respectively representing the relative variation coefficients corresponding to the 4 corners of the curtain wall panel corresponding to the working conditions.
[Y]Outputting vectors for the intelligent classification database; y is 1 、Y 2 、Y 3 、Y 4 The 4 corner bolt looseness degrees of the curtain wall panel are respectively represented, and the value is 1,2,3, … and N.
Optionally, the output module is further configured to:
and generating a maximum interval hyperplane between the working conditions of different damage degrees according to the dynamic response signal and a support vector machine model optimized based on a wolf algorithm, and judging the working condition of the curtain wall panel to be detected according to the maximum interval hyperplane to obtain a quantitative identification result of the damage state of the curtain wall panel to be detected.
In one aspect, an electronic device is provided, and the electronic device includes a processor and a memory, where the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the method for quantitatively identifying damage state of building curtain wall based on machine learning algorithm.
In one aspect, a computer-readable storage medium is provided, and at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the method for quantitatively identifying the damage state of the building curtain wall based on the machine learning algorithm.
The technical scheme provided by the embodiment of the invention has the beneficial effects that at least:
in the scheme, the defects that the existing detection technology cannot realize synchronous identification of the damage degree and the damage position of the building curtain wall and cannot realize quantitative identification of the damage state are overcome; and the laser vibration measurement technology is combined, so that the remote, lossless, accurate, digital and intelligent quantitative identification of the damage state of the building curtain wall is realized. The composite material can be widely applied to building external hanging structures connected by bolts or similar components, such as point-supported glass curtain walls, stone curtain walls, indoor roof hanging plates, building metal external hanging plate components and the like.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for quantitatively identifying damage states of building curtain walls based on a machine learning algorithm, provided by the embodiment of the invention;
FIG. 2 is a schematic view of panel excitation and panel dynamic response signal acquisition in accordance with the present invention;
FIG. 3 is a graph showing the relative natural frequency variation with the progress of damage under some conditions of the present invention;
FIG. 4 is a diagram of the variance trend of the relative vibration obtained at 4 corners under some conditions;
FIG. 5 is a diagram of the trend of the relative variation coefficients obtained at 4 corners under some conditions of the present invention;
FIG. 6 is a schematic diagram (one) of the support vector machine model principle of the present invention;
FIG. 7 is a schematic diagram of the model principle of the support vector machine of the present invention;
FIG. 8 is a schematic flow chart of the present invention based on the gray wolf algorithm optimization support vector machine model;
FIG. 9 is an effect diagram of the present invention based on the gray wolf algorithm optimization support vector machine model;
FIG. 10 is a block diagram of a device for quantitatively identifying damage states of building curtain walls based on a machine learning algorithm according to an embodiment of the present invention;
fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
As shown in fig. 1, an embodiment of the present invention provides a method for quantitatively identifying a damage state of a building curtain wall based on a machine learning algorithm, where the method may be implemented by an electronic device. As shown in fig. 1, a flow chart of a method for quantitatively identifying damage states of building curtain walls based on a machine learning algorithm, a processing flow of the method may include the following steps:
and S1, acquiring a dynamic response signal of the curtain wall panel to be tested.
In a feasible implementation mode, dynamic response signals of 4 corners and the middle of a curtain wall panel to be tested are collected on site, and a common rubber hammer, an unmanned aerial vehicle capable of launching rubber bullets or environmental excitation can be selected according to specific conditions to excite the panel. The panel to be tested is required to be consistent with the panel specification used when the database is obtained, the vibration excitation position is fixed to the middle of the panel every time, and the damage identification index corresponding to the panel is obtained and calculated.
Optionally, the building process of the support vector machine model based on the grayish wolf algorithm optimization in S2 includes:
and S21, acquiring a dynamic response signal of the curtain wall panel.
The dynamic response signals comprise dynamic response signals in plates of the curtain wall panel under the working conditions of different damage degrees and dynamic response signals of 4 corners of the curtain wall panel.
Optionally, the obtaining of the dynamic response signal of the curtain wall panel in S21 includes:
and acquiring a dynamic response signal of the curtain wall panel by using a laser Doppler vibration meter.
In a feasible implementation mode, the working conditions with different damage degrees are realized by loosening bolts on site, the real working conditions under the actual conditions are simulated through the mutual combination of the loosening number and the loosening degree of the bolts, the data obtained by multi-working-condition simulation are used as the reference, the approximate corresponding working conditions of the panel to be tested are analyzed, and then the safety evaluation is realized. The dynamic response signal is a vibration speed signal of a lower panel which is artificially excited, and the excitation position is constant at the middle part of the panel.
Further, an LDV (Laser Doppler Vibrometer) is prepared as a signal acquisition device, and a curtain wall panel convenient for field manual experiments is selected for damage condition simulation. The curtain wall panel damage simulation can be realized by not hard up bolt, and the not hard up number of turns of bolt accessible or bolt back-out length quantizes. According to different loosening degrees of the bolt, the damage grade of the bolt is divided into: no damage, grade 2 damage, grade 3 damage, … …, grade N damage; the thinner the loosening degree is, the more accurate the damage state evaluation of the curtain wall panel to be finally measured is. Through the mutual combination of the bolt loosening number and the loosening degree, the real working condition under the actual condition is simulated as much as possible, and the damage working condition which is not in line with the actual condition is eliminated according to the actual condition. In addition, the excitation device can be a common rubber hammer or other devices, and the excitation position is constant at the middle part of the panel.
For example, as shown in fig. 2, based on an indoor test of a bolted point-supported glass curtain wall, a support vector machine-based method for quantitatively identifying damage states of a building curtain wall is researched and verified. Ordinary tempered glass and ordinary aluminum alloy bolts with the size of 60cmX52cmX6mm are selected for testing. The diameter of the screw cap is 1.6cm, the diameter of the screw rod is 0.8cm, the thickness of the screw cap is 0.4cm, and the length of the screw rod is 1.8 cm; the schematic of the experiment is shown in FIG. 3. In the test process, the bolt loosening grades are divided into: 1-undamaged, screwing the bolt into contact with the glass panel; 2-2 level damage, and screwing the bolt out by 1 cm; and 3-3 level damage, and completely screwing out the bolt. After the unrealistic working conditions were eliminated according to the actual conditions, the glass panel damage working conditions amounted to 51. Under each working condition, the center of the panel is excited by using a rubber hammer, the sampling frequency of the Doppler laser vibrometer is set to be 1000Hz, the sampling time is set to be 3.4min, and the distance between the Doppler laser vibrometer and the panel to be measured is 5 m. Respectively acquiring data of 5 point positions which are counted by 4 corners of the panel and the center of the panel, wherein the data is a group of data; and each point position acquires data under excitation for 3 times, so that 3 groups of dynamic response signals are acquired at each working condition in the experiment.
And S22, calculating damage identification indexes of the curtain wall panel under the working conditions of different damage degrees based on the dynamic response signals.
The damage identification index comprises a damage degree identification index relative natural frequency, a damage position identification index relative vibration variance and a damage position identification index relative variation coefficient.
Alternatively, the calculation method of the damage degree identification index relative to the natural frequency in S22 is as shown in the following equation (1):
in the formula, X f A relative natural frequency corresponding to a certain damage condition;calculating the average natural frequency of the collected signals of the 4 corners in the board corresponding to the damage condition;collecting average natural frequency calculated by signals for 4 corners in the board corresponding to the undamaged working condition; and i is the number of the signal acquisition position and has the value range of 1,2,3,4 and 5.
Indexes representing the damage degree of the curtain wall panel comprise inherent frequency, an accumulated difference value of an origin acceleration frequency response function, vibration transfer rate and the like. In one possible embodiment, the analysis is performed only at the natural frequency. The natural frequency can represent the trend that the natural vibration property of the curtain wall panel changes along with damage. In order to improve the index accuracy, the average value of the natural frequency is obtained by adopting data at 5 point positions of 4 corners in the board, and the relative value under the condition of no damage is obtained in order to avoid the poor final identification effect caused by different dimension and size dimension of various indexes.
Alternatively, the calculation method of the relative vibration variance of the damage location identification index in S22 is as shown in the following formula (2):
alternatively, the method for calculating the relative variation coefficient of the damage location identification index in S22 is as shown in the following formula (3):
in the formula (I), the compound is shown in the specification,the relative vibration variance of the a-th position of a working condition panel is obtained;the relative coefficient of variation of the a-th position of a panel under a certain working condition;is the absolute average of the a-th position,the vibration speed of the facet plate at i is the a-th position. a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
The identification indexes of the damage position of the building curtain wall panel comprise modal curvature, relative vibration variance, relative variation coefficient and the like. In one possible embodiment, the analysis is performed only with the relative vibration variance and the relative coefficient of variation. And describing the distribution of the damage positions by adopting the vibration variance or the proportion of the variation coefficient of a certain corner data acquisition point to the sum of the vibration variances or the variation coefficients of 4 corners.
Wherein, the vibration variance and the variation coefficient are time domain indexes. The vibration variance can represent the signal chaos and the discrete degree, and the coefficient of variation can represent the proportion of the impact signal in the vibration signal, and the larger the proportion is, the larger the coefficient of variation is. Compared with the modal parameter indexes such as natural frequency and the like and the frequency domain indexes, the time domain indexes are more sensitive to structural damage and can generate mutation in small damage. The damage positions can be distinguished by comparing the relative vibration variance or the relative variation coefficient at 4 corner positions of the curtain wall panel.
For example, fig. 3 shows a trend graph of the relative natural frequency of part of the operating conditions with the progress of the damage. In FIG. 3, the bolts that set the top left, top right, bottom right, and bottom left corners of the panel are labeled A, B, C, D, respectively; in this embodiment, the bolt loosening degree is set as: a. the 1 The bolt is just contacted with the panel if the bolt is not damaged; a. the 2 Screwing out the bolt for 1cm under the condition of 2-level damage; a. the 3 In order to damage the level 3, the bolt is completely screwed out; the rest of the positions are similar.
As can be seen from fig. 3, in this embodiment, when the bolt at a or B has 2-level damage, the natural frequency is substantially consistent with the undamaged working condition, that is, the natural frequency cannot identify a small damage; at the same time, from A 1 B 2 C 2 D 2 And A 2 B 1 C 2 D 2 Working conditions A 3 B 2 C 2 D 2 And A 2 B 2 C 2 D 3 Operating conditions and A 1 B 3 C 3 D 1 And A 1 B 1 C 3 D 3 The working condition can be known, when the bolt is not hard up the number and not hard up the degree all unanimous, natural frequency size receives the damage position to influence great, can not accurately distinguish this type of working condition, appears the erroneous judgement when easily leading to the evaluation curtain panel safety condition. To sum up, the variation trend of the natural frequency data in the embodiment is consistent with the results of Huang and Pan,the curtain wall panel safety state evaluation only by means of the natural frequency has great defects.
As shown in FIGS. 4 to 5, comparative example A 1 B 1 C 1 D 1 And A 2 B 1 C 1 D 1 、A 3 B 1 C 1 D 1 And A 3 B 2 C 1 D 1 The working conditions can be known, the curtain wall panel micro-damage can be identified based on the relative vibration variance and the relative variation coefficient, the relative damage degree of 4 bolts in each working condition can be distinguished, and the mixed working conditions are divided. The relative vibration variance is most sensitive to micro-damage but has general stability, and the relative variation coefficient is sensitive to micro-damage but has good stability; the accuracy of dividing damage positions under various working conditions can be improved through the combination of the two indexes.
And S23, constructing an intelligent classification database based on the damage identification indexes.
Optionally, the intelligent classification database of multi-source information fusion is constructed based on natural frequency capable of representing that the natural vibration attribute of the curtain wall panel decreases with the aggravation of damage, vibration variance capable of representing that the chaos and the discrete degree of the panel time-course signal are aggravated with the aggravation of damage, and variation coefficient capable of representing that the ratio of the panel time-course signal to the impact signal component increases with the aggravation of damage. In the intelligent classification database, damage identification indexes are used as characteristic vectors of all working conditions, the damage degree of bolts at all positions is labels corresponding to all working conditions, and the following formulas (4) and (5) are shown:
in the formula, [ X ]]Representing a feature vector, X, corresponding to a condition f Representing the relative natural frequency corresponding to the working condition;respectively representing the relative vibration variances corresponding to the 4 corners of the working condition;respectively representing the corresponding relative variation coefficients of the 4 corners under the working condition.
[Y]=[Y 1 ,Y 2 ,Y 3 ,Y 4 ] (5)
[Y]Corresponding labels for all working conditions, namely, outputting vectors for the model; y is 1 ,Y 2 ,Y 3 ,Y 4 The bolt looseness degrees at the 4 corners are respectively represented, the values are 1,2,3, … and N, and the corresponding relation with the bolt looseness degrees is 1-undamaged, 2-level damage, 3-level damage and … … and N-level damage.
For example, 153 groups of data are contained in the intelligent classification database in total, and each working condition corresponds to 3 groups of data; each group of data comprises 9 characteristic values which are relative natural frequency, relative vibration variance and relative variation coefficient at 4 corners; each group of data comprises 4 labels with the value of 1 or 2 or 3.
S24, constructing a support vector machine model optimized by the wolf algorithm based on the intelligent classification database.
The classification algorithm in the machine learning algorithm includes a Support Vector machine, a limit gradient elevator, a random forest algorithm, etc., and in a possible implementation, the analysis is performed only by an SVM (Support Vector Machines) model optimized based on GWO (Grey Wolf Optimizer). And analyzing the correlation between the characteristic vector and the output vector of each working condition based on the model to obtain the maximum interval hyperplane between the working conditions. By the method, the problem of identifying the damage of the building curtain wall is converted into the problem of classifying and identifying various working conditions possibly existing in reality. And evaluating the looseness degree of each bolt of the panel by classifying and identifying the corresponding working conditions of the panel under the real condition, thereby realizing the quantitative identification of the damage state of the panel.
In the embodiment, the output of the intelligent recognition model is the damage degree of 4 bolts, which is a multi-label classification problem, so that the multi-label classification model is constructed based on Binary-SVM, specifically, the multi-label classification problem is decomposed into a plurality of single-label two-classification problems for calculation based on a data set decomposition method, and then the recognition results are combined, and finally the multi-label classification problem result [ Y ] is output.
Meanwhile, since most of the panel damage states and damage identification indexes are in a nonlinear correspondence relationship, the model kernel function in the embodiment selects a radial basis kernel function which can take recognition accuracy and model extrapolation effect into consideration, and the expression is shown in formula (6):
in the formula, x is a vector to be classified, and in the embodiment, is an input feature vector corresponding to a certain working condition; x is the number of c Is a radial basis kernel functionA central value of (d); II x-x c ‖ 2 Is x and x c The square of the 2 norm of the difference vector; σ is the scaling factor.
Furthermore, relevant parameters of the intelligent recognition model are set, and the model is trained based on the database.
In a feasible implementation mode, in an GWO-SVM model, an optimization algorithm part is set, the number of wolf clusters is 20, the iteration times are 100 times, 5-fold cross validation is adopted, the value range of a penalty parameter C is [0.001,1000], and the value range of a parameter g is [0.001,1000 ]; the intelligent identification part is used for constructing a multi-label classification model based on Binary-SVM, and a kernel function is a radial basis kernel function; any 102 of 153 sets of data were randomly selected for model training, leaving 51 sets as test sets.
In this embodiment, a schematic diagram of an SVM model principle is shown in fig. 6 and fig. 7, and the original linear indivisible problem is mapped to a high-dimensional space through a radial basis kernel function, and an optimal classification hyperplane is found in the high-dimensional space to divide the original linear indivisible data. The model is based on the principle of minimizing the structural risk, the optimal solution under the current sample is obtained, the problems of small sample, high dimensionality and nonlinear pattern recognition can be effectively solved, and the model is suitable for the field of curtain wall damage detection. The algorithm optimization flow is shown in fig. 8, and the optimization result is shown in fig. 9. In the experiment of this embodiment, the average number of optimization iterations is 5, and the average training duration of the model is 5.68 s. Therefore, the support vector machine model optimized based on the wolf algorithm can obviously improve the identification accuracy and greatly shorten the calculation time.
Table 1 shows comparison of recognition accuracy under various index combinations in the experiment. As can be seen from table 1, the damage recognition index composed of the relative natural frequency, the relative vibration variance, and the relative variation coefficient has the highest accuracy in the intelligent recognition model, which can reach 92.86%, while the recognition model composed of only the relative natural frequency or the local index has an accuracy less than 60%. In summary, the best recognition effect can be achieved through the combination of 3 indexes.
TABLE 1
And S3, obtaining a quantitative recognition result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm.
Optionally, the obtaining of the result of quantitatively identifying the damage state of the curtain wall panel to be tested according to the dynamic response signal and the support vector machine model optimized based on the grayling algorithm in S3 includes:
and generating a maximum interval hyperplane between the working conditions of a plurality of working conditions with different damage degrees according to the dynamic response signal and a support vector machine model optimized based on a wolf algorithm, and judging the working condition of the curtain wall panel to be detected according to the maximum interval hyperplane to obtain a quantitative recognition result of the damage state of the curtain wall panel to be detected.
In a feasible implementation mode, the damage identification index corresponding to the panel is obtained and calculated, the maximum interval hyperplane between the working conditions generated by the trained intelligent identification model is used for intelligently judging the concrete working conditions which can be approximately corresponding to the panel, the damage degree of each bolt is identified, and further the quantitative identification of the damage state of the curtain wall is realized.
In the embodiment of the invention, the defects that the prior detection technology cannot realize synchronous identification of the damage degree and the damage position of the building curtain wall and cannot realize quantitative identification of the damage state are overcome; and the laser vibration measurement technology is combined, so that the remote, nondestructive, accurate, digital and intelligent quantitative identification of the damage state of the building curtain wall is realized. The composite material can be widely applied to building external hanging structures connected by bolts and similar components, such as point-supported glass curtain walls, stone curtain walls, indoor roof hanging plates, building metal external hanging plate parts and the like.
As shown in fig. 10, an embodiment of the present invention provides a device 1000 for quantitatively identifying an injury state of a building curtain wall based on a machine learning algorithm, where the device 1000 is applied to implement a method for quantitatively identifying an injury state of a building curtain wall based on a machine learning algorithm, and the device 1000 includes:
the obtaining module 1010 is used for obtaining a dynamic response signal of the curtain wall panel to be tested.
An input module 1020 for inputting the dynamic response signal to the constructed support vector machine model optimized based on the gray wolf algorithm.
And the output module 1030 is used for obtaining a quantitative identification result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the grayish wolf algorithm.
Optionally, the input module 1020 is further configured to:
s21, acquiring a power response signal of the curtain wall panel; the dynamic response signals comprise dynamic response signals in a plate of the curtain wall panel under the working conditions of different damage degrees and dynamic response signals of 4 corners of the curtain wall panel.
S22, calculating damage identification indexes of the curtain wall panels under the working conditions of different damage degrees based on the dynamic response signals; the damage identification index comprises a damage degree identification index relative natural frequency, a damage position identification index relative vibration variance and a damage position identification index relative variation coefficient.
S23, constructing an intelligent classification database based on the damage identification indexes.
S24, constructing a support vector machine model optimized by the wolf algorithm based on the intelligent classification database.
Alternatively, the method for calculating the relative natural frequency of the damage degree identification index is as shown in the following formula (1):
in the formula, X f The relative natural frequency corresponding to any damage degree working condition in a plurality of working conditions with different damage degrees;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the working condition f;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the working condition that the damage degree is not damaged; and i is the number of the signal acquisition position and has the value range of 1,2,3,4 and 5.
Alternatively, the calculation method of the damage location identification index relative vibration variance is as shown in the following formula (2):
in the formula (I), the compound is shown in the specification,the relative vibration variance of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
Alternatively, the calculation method of the relative variation coefficient of the damage location identification index is as shown in the following formula (3):
in the formula (I), the compound is shown in the specification,the relative variation coefficient of the a-th position of the curtain wall panel under any one damage degree in a plurality of working conditions with different damage degrees; n is the number of collected signal samples;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
Optionally, the input module 1020 is further configured to:
and acquiring a dynamic response signal of the curtain wall panel by using a laser Doppler vibration meter.
Optionally, the intelligent classification database is as shown in the following formulas (4), (5):
[Y]=[Y 1 ,Y 2 ,Y 3 ,Y 4 ] (5)
wherein [ X ]]Feature vector, X, corresponding to any one of a plurality of different damage levels f Representing the relative natural frequency corresponding to the working condition;respectively representing relative vibration variances at 4 corners of the curtain wall panel corresponding to the working conditions;respectively representing the relative variation coefficients corresponding to the 4 corners of the curtain wall panel corresponding to the working conditions.
[Y]Outputting vectors for the intelligent classification database; y is 1 、Y 2 、Y 3 、Y 4 The 4 corner bolt looseness degrees of the curtain wall panel are respectively represented, and the value is 1,2,3, … and N.
Optionally, the output module 1030 is further configured to:
and generating a maximum interval hyperplane between the working conditions of different damage degrees according to the dynamic response signal and a support vector machine model optimized based on a wolf algorithm, and judging the working condition of the curtain wall panel to be detected according to the maximum interval hyperplane to obtain a quantitative identification result of the damage state of the curtain wall panel to be detected.
In the embodiment of the invention, the defects that the prior detection technology cannot realize synchronous identification of the damage degree and the damage position of the building curtain wall and cannot realize quantitative identification of the damage state are overcome; and the laser vibration measurement technology is combined, so that the remote, nondestructive, accurate, digital and intelligent quantitative identification of the damage state of the building curtain wall is realized. The composite material can be widely applied to building external hanging structures connected by bolts and similar components, such as point-supported glass curtain walls, stone curtain walls, indoor roof hanging plates, building metal external hanging plate parts and the like.
Fig. 11 is a schematic structural diagram of an electronic device 1100 according to an embodiment of the present invention, where the electronic device 1100 may generate a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 1101 and one or more memories 1102, where the memory 1102 stores at least one instruction, and the at least one instruction is loaded and executed by the processors 1101 to implement the following method for quantitatively identifying a damage state of a building curtain wall based on a machine learning algorithm:
and S1, acquiring a dynamic response signal of the curtain wall panel to be detected.
And S2, inputting the dynamic response signal to a well-constructed support vector machine model optimized based on the gray wolf algorithm.
And S3, obtaining a quantitative recognition result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm.
In an exemplary embodiment, a computer-readable storage medium, such as a memory, including instructions executable by a processor in a terminal, is further provided, where the instructions are configured to perform the method for quantitatively identifying the damage state of the building curtain wall based on the machine learning algorithm. For example, the computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit and scope of the present invention.
Claims (10)
1. A method for quantitatively identifying the damage state of a building curtain wall based on a machine learning algorithm is characterized by comprising the following steps:
s1, acquiring a power response signal of the curtain wall panel to be tested;
s2, inputting the dynamic response signal to a constructed support vector machine model optimized based on the wolf algorithm;
and S3, obtaining a quantitative recognition result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm.
2. The method according to claim 1, wherein the construction process of the support vector machine model based on the wolf' S algorithm optimization in S2 includes:
s21, acquiring a power response signal of the curtain wall panel; the dynamic response signals comprise dynamic response signals in plates of the curtain wall panel under a plurality of working conditions with different damage degrees and dynamic response signals of 4 corners of the curtain wall panel;
s22, calculating damage identification indexes of the curtain wall panel under the working conditions of different damage degrees based on the dynamic response signals; the damage identification index comprises a damage degree identification index relative natural frequency, a damage position identification index relative vibration variance and a damage position identification index relative variation coefficient;
s23, constructing an intelligent classification database based on the damage identification indexes;
and S24, constructing a support vector machine model optimized by the gray wolf algorithm based on the intelligent classification database.
3. The method according to claim 2, wherein the method for calculating the damage degree identification index relative to the natural frequency in S22 is represented by the following formula (1):
in the formula, X f The relative natural frequency corresponding to any damage degree working condition in a plurality of working conditions with different damage degrees;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the damage working condition f;calculating average natural frequency of collected signals of a plate center of the curtain wall panel and 4 corners of the curtain wall panel corresponding to the working condition that the damage degree is not damaged; and i is the number of the signal acquisition position and has the value range of 1,2,3,4 and 5.
4. The method according to claim 2, wherein the calculation method of the damage location identification index relative vibration variance in S22 is as shown in the following formula (2):
in the formula (I), the compound is shown in the specification,the relative vibration variance of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the time point i is the a position;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
5. The method according to claim 2, wherein the method for calculating the relative coefficient of variation of the damage location identification index in S22 is represented by the following formula (3):
in the formula (I), the compound is shown in the specification,the relative variation coefficient of the a-th position of the curtain wall panel under any damage degree in a plurality of working conditions with different damage degrees;the vibration speed of the curtain wall panel at the moment i at the a-th position is set;is the absolute average of the a-th position,a is the serial number of 4 corner positions of the curtain wall panel, and the values are 1,2,3 and 4; and N is the number of sample points of the collected vibration speed time-course signal.
6. The method of claim 2, wherein the obtaining the dynamic response signal of the curtain wall panel in S21 comprises:
and acquiring a dynamic response signal of the curtain wall panel by using a laser Doppler vibration meter.
7. The method of claim 2, wherein the intelligent classification database is represented by the following formulas (4), (5):
[Y]=[Y 1 ,Y 2 ,Y 3 ,Y 4 ] (5)
wherein [ X ]]A feature vector, X, corresponding to any one of a plurality of different damage-level conditions f Representing the relative natural frequency corresponding to the working condition;respectively representing relative vibration variances at 4 corners of the curtain wall panel corresponding to the working conditions;respectively representing relative variation coefficients corresponding to 4 corners of the curtain wall panel corresponding to the working conditions;
[Y]outputting vectors for the intelligent classification database; y is 1 、Y 2 、Y 3 、Y 4 The 4 corner bolt looseness degrees of sign curtain wall panel respectively, the value is 1,2,3, …, N.
8. The method as claimed in claim 1, wherein the obtaining of the result of quantitatively identifying the damage state of the curtain wall panel to be tested according to the dynamic response signal and the optimized support vector machine model based on the wolf algorithm in S3 comprises:
and generating a maximum interval hyperplane between the working conditions of different damage degrees according to the dynamic response signal and a support vector machine model optimized based on a wolf algorithm, and judging the working condition of the curtain wall panel to be detected according to the maximum interval hyperplane to obtain a quantitative identification result of the damage state of the curtain wall panel to be detected.
9. The utility model provides a building curtain damage state quantitative recognition device based on machine learning algorithm which characterized in that, the device includes:
the acquisition module is used for acquiring a power response signal of the curtain wall panel to be detected;
the input module is used for inputting the dynamic response signal into a constructed support vector machine model optimized based on a grayish wolf algorithm;
and the output module is used for obtaining a quantitative identification result of the damage state of the curtain wall panel to be detected according to the dynamic response signal and the support vector machine model optimized based on the wolf algorithm.
10. The apparatus of claim 9, wherein the input module is further configured to:
s21, acquiring a power response signal of the curtain wall panel; the dynamic response signals comprise dynamic response signals in plates of the curtain wall panel under the working conditions of different damage degrees and dynamic response signals of 4 corners of the curtain wall panel;
s22, calculating damage identification indexes of the curtain wall panel under the working conditions of different damage degrees based on the dynamic response signals; the damage identification index comprises a damage degree identification index relative natural frequency, a damage position identification index relative vibration variance and a damage position identification index relative variation coefficient;
s23, constructing an intelligent classification database based on the damage identification indexes;
and S24, constructing a support vector machine model optimized by the gray wolf algorithm based on the intelligent classification database.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210617762.0A CN115032270B (en) | 2022-06-01 | 2022-06-01 | Method and device for quantitatively identifying damage state of building curtain wall based on machine learning algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210617762.0A CN115032270B (en) | 2022-06-01 | 2022-06-01 | Method and device for quantitatively identifying damage state of building curtain wall based on machine learning algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115032270A true CN115032270A (en) | 2022-09-09 |
CN115032270B CN115032270B (en) | 2023-03-28 |
Family
ID=83123364
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210617762.0A Active CN115032270B (en) | 2022-06-01 | 2022-06-01 | Method and device for quantitatively identifying damage state of building curtain wall based on machine learning algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115032270B (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109520688A (en) * | 2018-11-15 | 2019-03-26 | 北京科技大学 | Building curtain wall safe condition remote detecting method based on vibration measurement with laser technology |
CN110132596A (en) * | 2019-04-24 | 2019-08-16 | 昆明理工大学 | A method of the rolling bearing fault diagnosis based on wavelet packet and GWO-SVM |
CN111024433A (en) * | 2019-12-30 | 2020-04-17 | 辽宁大学 | Industrial equipment health state detection method for optimizing support vector machine by improving wolf algorithm |
JP2020128951A (en) * | 2019-02-12 | 2020-08-27 | 国立研究開発法人防災科学技術研究所 | Building damage state estimation system and building damage state estimation method |
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 |
CN113618491A (en) * | 2021-08-23 | 2021-11-09 | 浙江工业大学 | Method for establishing broach wear state recognition model |
CN114358423A (en) * | 2022-01-05 | 2022-04-15 | 腾讯科技(深圳)有限公司 | Method for processing building data for curtain wall |
-
2022
- 2022-06-01 CN CN202210617762.0A patent/CN115032270B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109520688A (en) * | 2018-11-15 | 2019-03-26 | 北京科技大学 | Building curtain wall safe condition remote detecting method based on vibration measurement with laser technology |
JP2020128951A (en) * | 2019-02-12 | 2020-08-27 | 国立研究開発法人防災科学技術研究所 | Building damage state estimation system and building damage state estimation method |
CN110132596A (en) * | 2019-04-24 | 2019-08-16 | 昆明理工大学 | A method of the rolling bearing fault diagnosis based on wavelet packet and GWO-SVM |
CN111024433A (en) * | 2019-12-30 | 2020-04-17 | 辽宁大学 | Industrial equipment health state detection method for optimizing support vector machine by improving wolf algorithm |
CN113618491A (en) * | 2021-08-23 | 2021-11-09 | 浙江工业大学 | Method for establishing broach wear state recognition model |
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 |
CN114358423A (en) * | 2022-01-05 | 2022-04-15 | 腾讯科技(深圳)有限公司 | Method for processing building data for curtain wall |
Non-Patent Citations (2)
Title |
---|
胡璇 等: "《改进灰狼算法优化支持向量机在风力机齿轮箱故障诊断中的应用》" * |
郝文峰 等: "《基于经验模态分解-支持向量机的玻璃幕墙开胶损伤的玻璃幕墙开胶损伤预测研究》" * |
Also Published As
Publication number | Publication date |
---|---|
CN115032270B (en) | 2023-03-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Berti et al. | Gravitational-wave spectroscopy of massive black holes with the space interferometer LISA | |
CN102445640B (en) | GIS device intelligent recognition method based on vector machine and artificial fish swarm optimization | |
CN103617684B (en) | Interference-type optical fiber circumference vibrating intruding recognizer | |
Li et al. | Damage detection of bridges using response of vehicle considering road surface roughness | |
CN105740404A (en) | Label association method and device | |
Jayasundara et al. | Vibration-based dual-criteria approach for damage detection in arch bridges | |
Solari et al. | Dynamic response of structures to thunderstorm outflows: Response spectrum technique vs time-domain analysis | |
CN110888186A (en) | Method for forecasting hail and short-time heavy rainfall based on GBDT + LR model | |
Alvanitopoulos et al. | Seismic intensity feature construction based on the Hilbert–Huang transform | |
CN110321518A (en) | A method of determining Hydrological Time Series trend type | |
CN110889440A (en) | Rockburst grade prediction method and system based on principal component analysis and BP neural network | |
CN113836999A (en) | Tunnel construction risk intelligent identification method and system based on ground penetrating radar | |
Entezami et al. | Condition assessment of civil structures for structural health monitoring using supervised learning classification methods | |
CN107632010B (en) | Method for quantifying steel sample by combining laser-induced breakdown spectroscopy | |
CN113935384B (en) | Bridge health monitoring method and system for self-adaptive decomposition and recognition of signals | |
CN104198998B (en) | Clustering treatment based CFAR (Constant False Alarm Rate) detection method under non-uniform background | |
Pourgholi et al. | Influence of Hankel matrix dimension on system identification of structures using stochastic subspace algorithms | |
CN115032270B (en) | Method and device for quantitatively identifying damage state of building curtain wall based on machine learning algorithm | |
CN107437112B (en) | A kind of mixing RVM model prediction methods based on the multiple dimensioned kernel function of improvement | |
CN109029711B (en) | Dynamic bridge structure multi-order frequency identification method | |
Lofrano et al. | A pseudo-modal structural damage index based on orthogonal empirical mode decomposition | |
CN109211814B (en) | It is a kind of to be set a song to music the soil profile kind identification methods of face partition characteristics based on three-dimensional light | |
CN105651537B (en) | A kind of truss structure damage real-time monitoring system of high susceptibility to damage | |
Hu et al. | Blind modal estimation using smoothed pseudo Wigner–Ville distribution and density peaks clustering | |
Andreadis et al. | Intelligent seismic acceleration signal processing for damage classification in buildings |
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 |