Scribd Logo
Upload

Welcome to Scribd!

  • 53 views

    Artificial Neural Networks For Structural Damage Detection Using Modal Data

    Uploaded by

    Kiran Hegde
    This document discusses using artificial neural networks (ANNs) for structural damage detection based on modal data. It describes modal analysis and how damage affects modal parameters like natural frequencies and mode shapes. An ANN is designed and trained on simulated data from a cantilever plate to detect single and multiple damages by their location and magnitude. The ANN accurately predicts damage scenarios, demonstrating its potential for structural health monitoring applications.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd

    Artificial Neural Networks For Structural Damage Detection Using Modal Data

    Uploaded by

    Kiran Hegde
    53 views26 pages
    This document discusses using artificial neural networks (ANNs) for structural damage detection based on modal data. It describes modal analysis and how damage affects modal parameters like natural frequencies and mode shapes. An ANN is designed and trained on simulated data from a cantilever plate to detect single and multiple damages by their location and magnitude. The ANN accurately predicts damage scenarios, demonstrating its potential for structural health monitoring applications.

    Original Description:

    Artificial neural network seminar

    Original Title

    ANN Seminar

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PPT, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses using artificial neural networks (ANNs) for structural damage detection based on modal data. It describes modal analysis and how damage affects modal parameters like natural frequencies and mode shapes. An ANN is designed and trained on simulated data from a cantilever plate to detect single and multiple damages by their location and magnitude. The ANN accurately predicts damage scenarios, demonstrating its potential for structural health monitoring applications.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    53 views26 pages

    Artificial Neural Networks For Structural Damage Detection Using Modal Data

    Uploaded by

    Kiran Hegde
    This document discusses using artificial neural networks (ANNs) for structural damage detection based on modal data. It describes modal analysis and how damage affects modal parameters like natural frequencies and mode shapes. An ANN is designed and trained on simulated data from a cantilever plate to detect single and multiple damages by their location and magnitude. The ANN accurately predicts damage scenarios, demonstrating its potential for structural health monitoring applications.

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PPT, PDF, TXT or read online from Scribd
    Download as ppt, pdf, or txt
    You are on page 1of 26

    ARTIFICIAL NEURAL NETWORKS FOR

    STRUCTURAL DAMAGE DETECTION USING


    MODAL DATA

    Presented By :
    KIRAN H
    VII Sem. C.E.
    Seminar Contents
    Slide no.

     1. Introduction
     2. Modal analysis
     3. Damage detection process
     3. Design of ANN.
     4. Damage detection in cantilever plate
     5. Training data for ANN
     6. Test/Validation data
     7. Damage detection scenarios
     24. Applications of ANN
     25. Conclusion
    Biological Neurons

     Neuron : computational unit in the nervous system. It has:


     (i) Dendrites (inputs) Neuron receives input from other neurons
     (ii) Cell body An electrical pulse that travels from the body, down the
    axon
     (iii) Axon (output) Touch the dendrites or cell body of the next
    neuron.
    Artificial Neurons

     Multiple inputs and a single output.


     The neurons can be trained to fire (or not), for particular input
    patterns.
     Inputs associated with weights.
     These weights correspond to synaptic efficacy in a biological
    neuron.
     Each neuron also has a single threshold value.
     The weighted sum of the inputs is formed, and the threshold
    subtracted, to compose the activation function.
     Note also that weights can be negative, which implies that the
    synapse has an inhibitory rather than excitatory effect on the
    neuron
    Biological and Artificial Neurons

    Human neuron A simple neuron with weight


    Architecture of neural networks

    Feed-forward networks Feedback networks


    Structural damage detection

     Assess the condition of structures.


     Gives improved understanding about structure’s capacity
    and typical performance during its service.
     Can be found out by monitoring change in structural
    responses, natural frequencies and mode shapes and strain
    mode shapes.
     These direct methods can’t locate and quantify damage.
     Modal analysis can give the location and magnitude of
    damage.
    Modal analysis

     The modal vibration test data is used.


     The natural frequencies and mode shapes can be considered
    to represent the state of the structure.
     Modal parameters depend only on the mechanical
    characteristics of the structure and not on the excitation
    applied.
     Theoretically the structure can be represented by
    measurements taken at a single location.
     Reduction in time and cost of performing damage
    monitoring and predictive maintenance.
    Damage Detection Process

     Every structure has its own natural vibration (or resonance)


    frequencies.
     Each vibration mode has a different energy distribution.
     Any localized damage will affect each mode differently
    depending on the location and severity of the damage.
     Modal parameters are sensitive to boundary conditions, i.e.,
    physical constraints of the structure.
     Damage will soften the structure and thus modify its
    dynamic characteristics such as frequency and mode shape.
    Damage Detection Process
    Artificial Neural Network

     The feed-forward, multilayered, supervised neural network


    with the error back propagation algorithm is used.
     Two stages.1st is data feed forward.
     The output of each node is defined as 
    netj = ij Oi + ϴj  
    Oj = f( netj )
     Threshold function is
    {1 x>1
    f(x) = {x -1≤x≤1
    {-1 x<-1
    Artificial Neural Network

     The 2nd stage is error back propagation and weights


    adjustment.
     The system error is defined as E
    E= (1/2P)* p=1∑P n=1∑N (dpn – opn)
    P - number of instances in the training set,
    dpn ,opn - desired and calculated output of the nth output node for
    the pth instance, respectively
    Damage Detection in Cantilever plate

     Cantilever steel plate properties


     Cross-section of 50.75 x 6.0 mm2 and mass density of 7670
    kg/m3.
     The first 6 natural frequencies for the intact and damaged
    configurations for this cantilever plate are taken from
    literature.
     The damage configuration was introduced by saw cuts at
    the clamped end (Cut-1) and the mid-span (Cut-2) of the
    plate.
     The depth (d) of the cut is 20 mm.
    Damage Detection in Cantilever plate
     Damage made by cuts at locations 1 and 5 for the cantilever
    plate

     Division into 9 components for simulation of damage in


    FEA
    Measured and computed (FEA) frequencies.

    Frequencies, Hz
    Case Mode 1 Mode 2 Mode 3 Mode 4 Mode 5 Mode 6

    Measured 5.67 35.58 100.33 196.38 319.50 485.71


    (undamaged)

    Measured 5.11 31.60 94.44 179.97 308.89 452.71


    (damaged)

    FEA 5.68 35.61 99.70 195.36 322.90 482.27


    (undamaged)
    The Neural Network Architecture

     A feed forward back-propagation neural network has been


    used.
     In input stage, 6 processing elements in the input layer
    representing a vector of the first 6 natural frequencies of the
    beam.
     In output stage,10 PEs for the output layer, where the first
    node consists of damage magnitude and the remaining 9
    nodes represent the damage indicators at each of 9 locations.
     The damage indicators '0' and '1' represent the undamaged
    and fully damaged conditions, respectively.
    The Neural Network Architecture

     In the hidden layer design ,18 PEs in single hidden layer


    network was chosen as it results in less RMS error.
    Training Data

     To obtain training data finite element (FE) model has been


    generated for the cantilever plate with 36 finite elements.
     The damage was defined as the fractional loss of the second
    moment of area over one location.
     The data was generated for a wide range of damage
    magnitudes of 1%, 5%, 10%, 15%, 20%, 25% and 30%.
     Hence, the database consists of training sets corresponding
    to 7 damage states in each of 9 locations of the cantilever
    beam. This results in 63 training sets.
    Training Data
     For multiple damage scenario, namely, simultaneous
    damage in two locations, training data has been generated
    for 1%, 10% and 20% damage magnitudes in each of two
    damage locations simultaneously which resulted in 144
    training sets.
    Test /Validation Data
     For validating the network, a test data set has been
    generated for 3 damage magnitudes of 3%, 13% and 23%.
     This test database consist of data pertaining to 3 damage
    levels for each of 9 damage locations which resulted in 27
    test sets for single damage scenario.
     For validating multiple damage state, test data set
    corresponding to the above three damage magnitudes in
    three selected pairs of damage locations, namely, at (1,3),
    (1,5) and (1,7), have been generated which resulted in 9 test
    sets.
    Neural Network Training

     The learning rate parameter (η) for the network is chosen as


    0.01 and the error tolerance is chosen as 0.01 after trials.
     Fractional change in resonant frequencies used to get
    accurate results.
     zi = (fui – fdi ) / fui
    zi is the fractional change in the i-th mode,
    f u and f d are the frequencies of undamaged and damaged
    structure, respectively.
    Damage detection using ANN
    Multiple damage detection
    Applications for Neural Networks

     Detection of medical phenomena


     Stock market prediction
     Credit assignment
     Monitoring the condition of machinery.
     Engine management
     In marketing
     Facial recognition
     Robotics
    Conclusion

     The detection and identification of structural damage is a


    vital part of the monitoring and servicing of structural
    systems during their lifetime.
     It was observed that prediction capability of ANN is better
    by using fractional difference of frequencies instead of
    using the frequencies.
     The results show that the ANN method is capable of
    predicting the location and magnitude of damage with good
    accuracy for single and multiple damage scenarios.

    You might also like