CN114813973A - Bolt loosening detection method combining time reversal technology and BP neural network - Google Patents

Bolt loosening detection method combining time reversal technology and BP neural network Download PDF

Info

Publication number
CN114813973A
CN114813973A CN202210223997.1A CN202210223997A CN114813973A CN 114813973 A CN114813973 A CN 114813973A CN 202210223997 A CN202210223997 A CN 202210223997A CN 114813973 A CN114813973 A CN 114813973A
Authority
CN
China
Prior art keywords
bolt
guided wave
neural network
damage
time reversal
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.)
Pending
Application number
CN202210223997.1A
Other languages
Chinese (zh)
Inventor
段元锋
隋晓东
唐志峰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN202210223997.1A priority Critical patent/CN114813973A/en
Publication of CN114813973A publication Critical patent/CN114813973A/en
Pending legal-status Critical Current

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
    • G01N29/44—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
    • 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
    • G01N29/048—Marking the faulty objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Biochemistry (AREA)
  • Analytical Chemistry (AREA)
  • Evolutionary Computation (AREA)
  • Pathology (AREA)
  • Chemical & Material Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Acoustics & Sound (AREA)
  • Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)

Abstract

The invention discloses a bolt looseness detection method combining a time reversal technology and a BP neural network. A plurality of ultrasonic guided wave sensors are arranged near the tested bolt group, ultrasonic guided waves are excited and received by adopting a time reversal technology, and guided wave signal characteristic values under a specific damage state are extracted. And carrying out local damage positioning and quantitative analysis by adopting a BP neural network, wherein the input of a single training sample is the guided wave signal characteristic values under all the multi-propagation paths corresponding to the excitation and receiving sensors, and the output is the damage position and the local damage degree of the bolt group. In addition, a linear relation between the average characteristic value of the signals under the multiple propagation paths and the overall looseness degree of the bolt group is established, and the overall damage state of the bolt group is predicted. The detection method provided by the invention realizes the damage positioning and damage degree prediction of the bolt group structure composed of a large number of bolts on the premise of adopting fewer sensors, has reliable detection result and strong engineering applicability, and provides guarantee for ensuring the safety of the bolt connection type steel structure.

Description

Bolt loosening detection method combining time reversal technology and BP neural network
Technical Field
The invention belongs to the technical field of constructional engineering, and relates to a time reversal technology, a BP neural network and a structural health monitoring technology.
Background
The bolt connection is a typical steel member connection mode, and is widely applied to multiple industries including civil engineering, aerospace, mechanical engineering, electric power engineering and the like, and the tightness degree of the bolt is related to the operation safety and the structure safety of equipment. The tightness degree of the bolt is influenced by various effects such as environmental load, traffic load and the like, the situation that a single bolt or a plurality of bolts are loosened inevitably exists in the using process, the bolt loosens to a certain degree and then can cause sudden damage to the structure, huge economic loss and severe social influence are caused, the health monitoring of the bolt connection structure is realized, and the safety monitoring device has important significance for improving the safety performance of equipment.
The current detection methods for bolt looseness include strain gauge detection methods, manual detection of a torquemeter, ultrasonic guided wave detection methods, impedance methods, machine vision technologies and the like. The manual detection cost is too high, the detection precision of a strain gauge detection method is not high, the number of sensors needing to be installed is large, the detection area of an impedance method is not large, and an ultrasonic guided wave technology and a machine vision technology become research hotspots in recent years.
In the aspect of bolt looseness detection based on ultrasonic guided waves, currently adopted detection methods comprise an energy method, a time reversal method, a frequency method and the like. Due to the fact that the surface of the bolt plate is not smooth, the actual contact area between the bolt plates is determined by the pretightening force of the bolt, ultrasonic guided wave signals are excited from one end of the bolt plate and transmitted to the other end of the bolt plate through the contact area, and the energy transmitted by the signals is determined by the actual contact area of the bolt plate. The detection technology based on the energy method utilizes the characteristic of signals to establish the relationship between the energy of the transmission signals and the bolt torque, however, the actual contact area of the bolt plate does not change after the bolt torque reaches a certain degree, so that the method cannot detect the bolt looseness in early stage. Similarly, the change of the signal frequency is not obvious enough in the detection mode based on the frequency method, and the detection mode is easily influenced by environmental noise and the like. The time reversal technology realizes accurate prediction of bolt loosening degree by focusing the energy of the guided wave signals at the signal receiving sensor.
Patent document CN113433214A proposes a damage positioning method for a plate structure, but the bolt structure has a more complicated structure form, and the propagation path of the guided wave is difficult to predict. Furthermore, the number of sensors used in the method is too large to be popularized in actual engineering application.
Disclosure of Invention
The invention aims to solve the technical problems and provides a bolt looseness detection method combining a time reversal technology and a BP neural network, which detects the overall and local damage degree of a complex bolt group through a small number of sensors, positions the damaged area, ensures the safety of a bolt connection structure during operation and prevents disasters.
In order to solve the technical problems, the invention adopts the following technical scheme:
a bolt loosening detection method combining a time reversal technology and a BP neural network comprises the following steps:
s1: arranging a plurality of ultrasonic guided wave excitation and receiving sensors with the same quantity near a tested bolt group, and exciting and receiving ultrasonic guided waves in a single-transmitting and single-receiving mode;
s2: under a lossless state, combining all excitation sensors and all receiving sensors one by one, wherein each combination represents different guided wave propagation paths, acquiring an ultrasonic guided wave reconstruction signal between a specific excitation sensor and a specific receiving sensor by adopting a time reversal technology, and extracting the maximum value of the reconstruction signal on a time domain;
s3: under the damage state, acquiring ultrasonic guided wave reconstruction signals under all excitation and receiving sensor combinations by adopting a time reversal technology, and calculating the ratio of the maximum value of the reconstruction signals under the damage state to the maximum value of the reconstruction signals under the corresponding propagation path in the nondestructive state to be used as the characteristic value of the guided wave signals on the propagation path;
s4: dividing the bolt group into a plurality of independent areas, wherein each area comprises a plurality of independent bolts;
s5: building a BP neural network, wherein the input of a single sample is characteristic values of guided wave signals on all propagation paths in a damaged state, the number of output nodes is the number of divided regions of a bolt group, the positions of the output nodes represent the regions of bolt loosening, and the size of the output node value represents the damage degree of the regions of bolt loosening;
s6: training and testing a BP neural network through a large amount of data, so as to accurately identify the bolt loosening position and the local bolt loosening degree;
s7: and averaging the characteristic values of the guided wave signals on all propagation paths under the condition of single damage, and establishing a linear relation with the overall looseness degree of the bolt group through a large number of damage samples to realize the prediction of the overall damage degree of the bolt group.
Further, the bolt group may be divided into a plurality of independent regions according to structural characteristics.
Furthermore, the ultrasonic guided wave excitation and receiving sensor is characterized by comprising an ultrasonic guided wave sensor which is not limited to a piezoelectric type and a magnetostrictive type and can be flexibly selected according to an actual measured component;
furthermore, the ultrasonic guided wave is characterized by comprising Lamb waves, SH waves and other forms, and the excitation frequency of the ultrasonic guided wave is reasonably selected according to the frequency dispersion characteristic of a structure to be detected;
furthermore, the plurality of ultrasonic guided wave sensors are characterized in that the number of the sensors for exciting and receiving signals is more than or equal to two, but is far less than the total number of the bolts, and the number of the exciting and receiving sensors is flexibly selected according to the measured size of the measured component and the total number of the bolts;
further, the time reversal technique is characterized by comprising the following steps:
s2-1: the excitation sensor excites a Gaussian pulse signal delta (t), and the guided wave signal is transmitted from one side of the bolt plate to the signal receiving sensor on the other side of the bolt plate through the connecting region of the bolt plate. Assuming that the structural impulse response is h (t), the received signal is y (t) ═ h (t);
s2-2: 0-t of received signal y (t) in time domain 0 Time-reversal of segments, inverted signal y (t) 0 -t)＝h(t 0 -t) is retransmitted as an input signal to the excitation sensor;
s2-3: acquisition of reconstructed signals from receiving sensors
Figure BDA0003538558540000031
Extraction of t 0 Reconstructed signal value at time instant
Figure BDA0003538558540000032
Further, the BP neural network input is characterized in that, for the BP neural network input, the guided wave signal characteristic value on a single guided wave propagation path is between 0 and 1 (lossless).
Further, the BP neural network output is characterized in that, for the output of the BP neural network, the output value of a single neuron node is between 0 (lossless) and 1, and represents the local region damage degree.
Further, the training data set of the BP neural network may be obtained through a finite element model or through actually acquired signals, and the hyper-parameters of the neural network are determined through repeated tests.
The method provided by the invention needs a small number of sensors, can be suitable for complex bolt connection structures, can detect the damage degree specifically to an independent area, and has high detection accuracy. The invention creatively combines the BP neural network and the time reversal technology, takes the damage characteristic values on different guided wave propagation paths as the judgment basis, accurately identifies the damage position and realizes high-efficiency detection.
Drawings
The above advantages of the present invention will become more apparent and more readily appreciated from the detailed description set forth below when taken in conjunction with the drawings, which are intended to be illustrative, not limiting, of the invention and in which:
FIG. 1 is a schematic view of an exemplary bolted connection in an embodiment of the present invention;
FIG. 2 is a graph of a Gaussian pulse signal excited by excitation sensor A1;
fig. 3 is a signal map received by the receiving sensor R1;
fig. 4 is a signal map received by R1 with time reversal;
FIG. 5 is a reconstructed signal map obtained by the receiving sensor R1;
FIG. 6 is a diagram of a BP neural network architecture;
FIG. 7 is a diagram of BP neural network prediction results according to an embodiment of the present invention;
fig. 8 is a diagram showing the prediction result of the overall damage degree of the bolt group according to the embodiment of the present invention.
Detailed Description
The following describes a bolt loosening detection method combining a time reversal technique and a BP neural network in detail with reference to the accompanying drawings.
The examples described herein are specific embodiments of the present invention, are intended to be illustrative and exemplary in nature, and are not to be construed as limiting the scope of the invention. In addition to the embodiments described herein, those skilled in the art will be able to employ other technical solutions which are obvious based on the disclosure of the claims and the specification of the present application, and these technical solutions include technical solutions which make any obvious replacement or modification for the embodiments described herein.
The invention provides a bolt looseness detection method combining a time reversal technology and a BP neural network, which comprises the following steps in combination with a specific structural form shown in figure 1:
s1: the tested bolt group shown in FIG. 1 comprises 8 independent bolts, 2 ultrasonic guided wave excitation and receiving sensors are respectively arranged at two ends of a bolt connecting area, ultrasonic guided waves are excited and received in a single-transmitting and single-receiving mode, and the propagation path of the ultrasonic guided waves comprises A1-R1, A1-R2, A2-R1 and A2-R2;
s2: in a nondestructive state, a Gaussian pulse signal shown in fig. 2 is input into an excitation sensor A1, an ultrasonic guided wave signal is transmitted from a left bolt plate to a right bolt plate through a bolt connecting region, and a receiving signal shown in fig. 3 is obtained by a receiving sensor R1;
s3: select the received signal of FIG. 3 from 0-1.5 × 10 -3 s is used as effective signal part and is 0-1.5 × 10 -3 Reversing the time-course signal in the interval s to obtain a reversed signal shown in FIG. 4, and inputting the reversed signal into the excitation sensor A1 again;
s4: the receiving sensor R1 obtains the reconstructed signal shown in FIG. 5, and extracts 1.5 × 10 in the lossless state -3 Amplitude Y of s-time signal intact,A1-R1 ；
S5: repeating S2-S4 to obtain Y intact,A1-R2 ，Y intact,A2-R1 ，Y intact,A2-R2 ；
S6: loosening the bolt, repeating the operations S2-S5, and obtaining Y under the current damage state damage,A1-R1 ，Y damage,A1-R2 ，Y damage,A2-R1 ，Y damage,A2-R2 ；
S7: the guided wave signal characteristic values DI on 4 propagation paths in the current damage state are calculated, and are illustrated by propagation paths A1-R1, DI A1-R1 The calculation formula is as follows:
Figure BDA0003538558540000051
DI is 1 in the lossless state;
s8: dividing a bolt group into 4 rows, wherein each row comprises two independent bolts;
s9: different DI values on 4 guided wave propagation paths can be obtained under a single damage state, and the DI values serve as a training sample of a group of neural networks, and the labels of the sample are 4 numbers from 0 to 1 (lossless). The position of the sample label number represents the number of damaged lines of the bolt, and the size of the sample number represents the local damage degree of the line of the bolt;
s10: collecting a training sample set of the neural network through a large number of damage conditions;
s11: inputting the training sample set into the neural network shown in fig. 6 for training, adjusting the hyper-parameters of the neural network to improve the accuracy of the training result, and then putting the tested data into the trained neural network for prediction to obtain the prediction result shown in fig. 7. FIG. 7 shows that the damage occurred in line 1 and line 2, and the corresponding predicted damage degree value is similar to the actual value;
s12: and calculating the average value of DI on 4 propagation paths in a single damage state, and establishing a linear relation between the DI average value and the overall damage state of the bolt group through a large number of samples. Fig. 8 is a graph showing a relationship between the average DI value and the overall damage degree of the bolt group in different states, and a linear relationship of fig. 8 is obtained by data fitting and used for predicting the overall damage degree of the bolt group.
The present invention is not limited to the above embodiments, and any other products in various forms can be obtained by the teaching of the present invention, but any changes in the shape or structure thereof, which are the same as or similar to the technical solutions of the present invention, fall within the protection scope of the present invention.

Claims (7)

1. A bolt loosening detection method combining a time reversal technology and a BP neural network is characterized by comprising the following steps:
s1: arranging a plurality of ultrasonic guided wave excitation and receiving sensors with the same quantity near a tested bolt group, and exciting and receiving ultrasonic guided waves in a single-transmitting and single-receiving mode;
s2: under a lossless state, combining all excitation sensors and all receiving sensors one by one, wherein each combination represents different guided wave propagation paths, acquiring an ultrasonic guided wave reconstruction signal between a specific excitation sensor and a specific receiving sensor by adopting a time reversal technology, and extracting the maximum value of the reconstruction signal on a time domain;
s3: under the damage state, acquiring ultrasonic guided wave reconstruction signals under all excitation and receiving sensor combinations by adopting a time reversal technology, and calculating the ratio of the maximum value of the reconstruction signals under the damage state to the maximum value of the reconstruction signals under the corresponding propagation path in the nondestructive state to be used as the characteristic value of the guided wave signals on the propagation path;
s4: dividing the bolt group into a plurality of independent areas, wherein each area comprises a plurality of independent bolts;
s5: building a BP neural network, wherein the input of a single sample is characteristic values of guided wave signals on all propagation paths in a damaged state, the number of output nodes is the number of divided regions of a bolt group, the positions of the output nodes represent the regions of bolt loosening, and the size of the output node value represents the damage degree of the regions of bolt loosening;
s6: training and testing a BP neural network through a large amount of data, so as to accurately identify the bolt loosening position and the local bolt loosening degree;
s7: and averaging the characteristic values of the guided wave signals on all propagation paths under the condition of single damage, and establishing a linear relation with the overall looseness degree of the bolt group through a large number of damage samples to realize the prediction of the overall damage degree of the bolt group.
2. The method of claim 1 wherein the ultrasonic guided wave excitation and reception sensors comprise piezoelectric and magnetostrictive ultrasonic guided wave sensors.
3. The method for detecting the loosening of the bolt by combining the time reversal technology and the BP neural network as claimed in claim 1, wherein the ultrasonic guided wave types include Lamb waves and SH waves, and the excitation frequency is reasonably selected according to the frequency dispersion characteristics of the structure to be detected.
4. The method of claim 1, wherein the number of the exciting or receiving sensors is greater than or equal to two but much smaller than the total number of the bolts of the bolt group, and the number of the exciting and receiving sensors is determined according to the size of the tested member and the number of the bolts.
5. The method for detecting the loosening of the bolt by combining the time reversal technology and the BP neural network according to claim 1, wherein the time reversal technology comprises the following steps:
s2-1: the excitation sensor excites Gaussian pulse signals, and guided wave signals are transmitted to the signal receiving sensor on the other side from one side of the bolt plate through the connection area of the bolt plate;
s2-2: the received signal is in the time domain 0-t 0 Time reversal is carried out on the segments, and the reversed signals are used as input signals to be sent to the excitation sensor again;
s2-3: obtaining a reconstructed signal from the receiving sensor, extracting t 0 The time of day signal value.
6. The method for detecting the loosening of the bolt by combining the time reversal technology and the BP neural network as claimed in claim 1, wherein the guided wave signal characteristic value calculated under a single guided wave propagation path is between 0 and 1 (lossless) for the input of the BP neural network.
7. The method of claim 1, wherein for the output of the BP neural network, the positions of the output neuron nodes represent local bolt regions, and the output value of a single neuron node is between 0 (lossless) and 1, which represents the damage degree of the local region.
CN202210223997.1A 2022-03-09 2022-03-09 Bolt loosening detection method combining time reversal technology and BP neural network Pending CN114813973A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210223997.1A CN114813973A (en) 2022-03-09 2022-03-09 Bolt loosening detection method combining time reversal technology and BP neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210223997.1A CN114813973A (en) 2022-03-09 2022-03-09 Bolt loosening detection method combining time reversal technology and BP neural network

Publications (1)

Publication Number Publication Date
CN114813973A true CN114813973A (en) 2022-07-29

Family

ID=82528938

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210223997.1A Pending CN114813973A (en) 2022-03-09 2022-03-09 Bolt loosening detection method combining time reversal technology and BP neural network

Country Status (1)

Country Link
CN (1) CN114813973A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114324584A (en) * 2021-12-16 2022-04-12 浙江工业大学 Steel structure detection method based on intelligent algorithm and ultrasonic phased array technology
CN116735705A (en) * 2023-04-10 2023-09-12 三峡大学 Damage detection method and device based on ultrasonic guided wave linear and nonlinear characteristics

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114324584A (en) * 2021-12-16 2022-04-12 浙江工业大学 Steel structure detection method based on intelligent algorithm and ultrasonic phased array technology
CN114324584B (en) * 2021-12-16 2023-09-05 浙江工业大学 Steel structure detection method based on intelligent algorithm and ultrasonic phased array technology
CN116735705A (en) * 2023-04-10 2023-09-12 三峡大学 Damage detection method and device based on ultrasonic guided wave linear and nonlinear characteristics
CN116735705B (en) * 2023-04-10 2024-01-23 三峡大学 Damage detection method and device based on ultrasonic guided wave linear and nonlinear characteristics

Similar Documents

Publication Publication Date Title
CN114813973A (en) Bolt loosening detection method combining time reversal technology and BP neural network
CN110389170B (en) Train component crack damage detection method and system based on Lamb wave imaging
Park et al. Impedance-based health monitoring of civil structural components
Park et al. Impedance‐Based Structural Health Monitoring
CN101566601B (en) System for recognizing tensile damage state of 16Mn steel force-bearing part by adopting neural network and coupling iteration
CN109649432B (en) System and method for monitoring integrity of steel rail of cloud platform based on guided wave technology
CN103760243A (en) Microcrack nondestructive testing device and method
CN110044533B (en) Bolt pretightening force monitoring method based on combination of EMD (empirical mode decomposition) energy entropy and support vector machine
CN114323375B (en) GIS basin-type insulator flange stress detection method and system
CN106909733A (en) Health state monitoring method for high-strength bolt of key component of steel truss bridge
CN104406901B (en) Concrete internal steel bar corrosion monitoring sensor based on Lamb wave principle
CN103940908A (en) Ultrasonic detecting device and method based on DBSCAN (Density-based Spatial Clustering Of Applications With Noise) and cross-correlation algorithms
CN112014470A (en) Quantitative assessment method and system for bolt connection state
CN103776902A (en) Nonlinear ultrasound evaluating method for impact fatigue damage of metal bonding interface
CN107192492A (en) A kind of tools for bolts ' pretension torque supersonic guide-wave monitoring method based on improvement time reversal
Chakraborty et al. Embedded ultrasonic transmission sensors and signal processing techniques for structural change detection in the Gliwice bridge
CN116840135A (en) Steel gate accelerated degradation test bed with simultaneous effects of corrosion and fatigue and post-degradation running state evaluation method
CN101000325A (en) Method for investigating failure connection of engineering structure by signal focus
CN203745428U (en) Microcrack nondestructive test device
CN110412132A (en) Based on the multi-modal ultrasonic phase array detection system of multi-mode and nonlinear imaging method
CN114509506B (en) Online crack evaluation method based on guided wave time spectrum difference and convolutional neural network set
CN113298805B (en) Structure surface defect detection method based on active Lamb wave acoustic emission
CN114235257A (en) Bolt axial force online accurate monitoring method and device based on data fusion
CN112710740B (en) Time reversal multi-bolt damage ultrasonic detection method and system based on PN code
CN115078540B (en) Wall nondestructive testing system and wall nondestructive testing method for wall climbing detection robot

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