Scribd Logo
Upload

Welcome to Scribd!

  • 201 views

    System Identification Assignment

    Uploaded by

    teuku arriessa sukhairi
    This document describes steps taken to create a mathematical model of a cantilever beam using system identification techniques in MATLAB. Experimental vibration data was collected from the beam and used to identify state space models of varying orders. The model that best fit the data had an order of 19 and captured 94.16% of the measured output variance. Frequency analysis of this model found the first four resonant frequencies of the beam. The identified model was then assigned to a Simulink model to simulate the beam's time response under sinusoidal input at the first resonant frequency.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    System Identification Assignment

    Uploaded by

    teuku arriessa sukhairi
    201 views9 pages
    This document describes steps taken to create a mathematical model of a cantilever beam using system identification techniques in MATLAB. Experimental vibration data was collected from the beam and used to identify state space models of varying orders. The model that best fit the data had an order of 19 and captured 94.16% of the measured output variance. Frequency analysis of this model found the first four resonant frequencies of the beam. The identified model was then assigned to a Simulink model to simulate the beam's time response under sinusoidal input at the first resonant frequency.

    Original Description:

    MATLAB, Advanced Control Method, Vibration

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document describes steps taken to create a mathematical model of a cantilever beam using system identification techniques in MATLAB. Experimental vibration data was collected from the beam and used to identify state space models of varying orders. The model that best fit the data had an order of 19 and captured 94.16% of the measured output variance. Frequency analysis of this model found the first four resonant frequencies of the beam. The identified model was then assigned to a Simulink model to simulate the beam's time response under sinusoidal input at the first resonant frequency.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    201 views9 pages

    System Identification Assignment

    Uploaded by

    teuku arriessa sukhairi
    This document describes steps taken to create a mathematical model of a cantilever beam using system identification techniques in MATLAB. Experimental vibration data was collected from the beam and used to identify state space models of varying orders. The model that best fit the data had an order of 19 and captured 94.16% of the measured output variance. Frequency analysis of this model found the first four resonant frequencies of the beam. The identified model was then assigned to a Simulink model to simulate the beam's time response under sinusoidal input at the first resonant frequency.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 9

    HOME WORK I:

    SYSTEM IDENTIFICATION ASSIGNMENT

    T.Arriessa Sukhairi
    (108013235855)

    Advanced Control Methods for Adaptive Mechanical Systems

    Problem: A cantilever beam is given but without knowledge of its properties (i.e. dimensions,
    mass, flexural rigidity (EI), etc.). One is interested to do frequency analysis for this cantilever
    beam using MATLAB\ System Identification Toolbox. In order to do so, a number of steps have
    to be performed.

    Step 1 - Data measurement


    The beam is attached with 4 piezoelectric patches to act as actuators. Experimental equipments
    are arranged as in Figure 1. Data acquisition equipment (DAQ) feeds 4 band-limited white
    noise excitation signals to the 4 actuators on the beam through a power amplifier. Vibration
    near tip of the beam is recorded by a velocity sensor (vibrometer).

    Figure 1. Experiment setup for data collection

    Plant: Single-sided clamped aluminum beam with 4 piezoelectric patches and a laser
    vibrometer (MISO system).

    Advanced Control Methods for Adaptive Mechanical Systems

    Plant-input: voltage signals to the four piezoelectric patches u[k].

    Plant-output: deflection velocity near to the end point y[k].


    Measurement was already done in 10 seconds with sampling period Ts = 0.001s. Four input
    signals and one output signal are stored in Data_samplingfreq_1000_hz.mat file
    which can be downloaded from blackboard in the directory Kursunterlagen\Sytem
    Identification_matlab files.zip.

    Step 2 - Model Identification: Indentify a linear state-space model for the plant using
    the subspace state-space system identification method N4SID in MATLAB\ System
    Identification Toolbox (using GUI or command lines). Perform this step with various
    model orders, e.g. 8, 10, 12, 14, etc. Describe the identification process.
    By using the system identification tool in MATLAB we process our data from
    Data_samplingfreq_1000_hz.mat.

    Figure 2. System Identification Toolbox

    Then we input our data in left side of the toolbox by choosing time domain data
    and then insert the value U from the Data_samplingfreq_1000_hz.mat file
    as input variable and Y from the Data_samplingfreq_1000_hz.mat file as
    output variable. Set the sampling interval as 0.001 s

    Advanced Control Methods for Adaptive Mechanical Systems

    Figure 3. Import Data

    Then we will get the data as shown in the figure 4. Select the range in which we
    are going to do the analysis. Here I choose the range between 3 7.

    Figure 4. Input and Output Signals

    Advanced Control Methods for Adaptive Mechanical Systems

    Figure 5. Range 3 7

    Figure 6. Combination the Input and Output and Range 3 7

    Step 3 - Model Verification: Compare the simulated output using the models obtained from
    task 2 with the measured output (use the command compare if you do system identification by
    command lines). Comment on the percentage of fit when model order is increased and choose
    simplest model that fits the measured data best.

    For this task, we need to compare the percentage result that we get by increasing or
    decreasing the model order, to find and choose the simplest model that fits the measured
    data best.

    We can change the value in Specify value to get the state space models (mathematical
    model) that we are going to compare with the model from the experiment.
    5

    Advanced Control Methods for Adaptive Mechanical Systems

    Figure 7. Choosing Model Order

    Here I choose the value 15, 18, 20, 19, 22.

    Then we compare the value as can be seen in the figure 8

    Figure 8. Model Output

    Advanced Control Methods for Adaptive Mechanical Systems

    The percentage that I get :


    15 (SS2)

    18 (SS3)

    20 (SS4)

    19 (SS5)

    22 (SS6)

    91.07 %

    92.51 %

    91.35 %

    94.16 %

    90.26 %

    Table 1. The percentage from comparing the simulated output using the models obtained from task 2
    with the measured output

    The best percentage that I get is when I specify the value as 19.

    Figure 9. The Percentage at Specify Value 19

    From the table 1 can be seen that the percentage value will increase until it reach the
    peak (in this case is 19) then it will decrease.

    Step 4 - Frequency analysis: From the best model selected in step 3, plot the frequency
    response of the plant and record the first 4 resonant frequencies of the beam.

    Plot the frequency response

    Figure 10. Frequency Response

    Advanced Control Methods for Adaptive Mechanical Systems

    From the plot of the frequency response we get the beam resonant frequencies
    1

    84.7881 rad/s

    455.8061 rad/s

    1242.594 rad/s

    2465,7624 rad/s

    Table 2. Resonant Frequencies

    Figure 11. 1st Resonant Frequency

    Step 5 Use the identified model: From the identified state-space model of the beam, assign
    those system properties to the given Simulink model state_space_model_beam.mdl and plot
    the model time response when the input for all 4 channels excitation is of the form
    u(t) = 10*sin(2f1t)
    where f1 is the first resonant frequency.

    After we get the best fit model for our system, then send the data into workspace. By using the
    data that we have, then we store the value a, b, c, and k that we have from the data to create
    variable A, B, C, and K in which will be used in Simulink. Then change the value f1 from the
    formula with the value of 1 in this case 84.7881 rad/s in source block parameter (see figure
    12). Then do the simulation and we can see the velocity as been shown in figure 13.

    Advanced Control Methods for Adaptive Mechanical Systems

    Figure 12. Simulink Model

    Figure 12. Source Block Parameter Input

    Figure 13.Output Velocity

    You might also like