LTPDA Training Session 1

This series of help pages consitute the first training session of LTPDA. The various data-packs used throughout the tutorials are available for download on the LTPDA web-site.

  1. Topic 1 - The basics of LTPDA
  2. Topic 2 - Pre-processing of data
  3. Topic 3 - Spectral Analysis
  4. Topic 4 - Transfer function models and digital filtering
  5. Topic 5 - Model fitting

In addition, throughout the course of this training session, we will perform a full analysis of some lab data. The inputs to the analysis are two time-series data streams, the first is the recorded output of an interferometer, the second is a recording of the room temperature in the vicinity of the interferometer. Both are recorded with different sample rates and on different sampling grids. The temperature data is unevenly sampled, and may evem have missing samples.

During each topic of the training session, the data will be manipulated using the tools introduced in that topic (and previous topics). The aim of the data analysis is to determine the influence of temperature on the interferometer output. In particular the steps will be:

  1. Topic 1 Loading and calibrating the raw data.
    1. Read in the raw data files and convert them to AOs
    2. Plot the two data streams
    3. Calibrate the interferometer output to meters (from radians)
    4. Calibrate the temperature data to degrees Kelvin from degrees Celcius
    5. Save the calibrated data series to XML files, ready for the input to the next topic
  2. Topic 2 Pre-processing and data conditioning.
    1. Read in the calibrated AOs from XML files
    2. Trim the data streams to the same time segments
    3. Resample the temperature on to an even sampling grid with no missing samples
    4. Resample to the two data streams to a common 1Hz sample rate
    5. Interpolate the two data streams on to the same time grid
    6. Save the cleaned data to AO XML files
  3. Topic 3 Spectral analysis.
    1. Load the time-series data from Topics 1 and 2
    2. Compare PSDs of the time-series data before and after pre-processing
    3. Check the coherence of temperature and IFO output before and after pre-processing
    4. Measure the transfer function from temperature to IFO output
    5. Save the measured transfer function to disk as an AO XML file
  4. Topic 4 Simulation of the system under investigation.
    1. Make approximate noise-shape models for the temperature and IFO displacement input spectra
    2. Make digital IIR filters matching these noise-shape models
    3. Filter white-noise data streams to produce simulated versions of the temperature and IFO inputs
    4. Make a model of the temperature to IFO coupling
    5. Construct a filter representing this coupling
    6. Filter the simulated temperature data and add it to the simulated IFO input data
    7. Save the simulated temperature and the simulated IFO output data to disk
    8. Repeat the steps from Topic 3, this time using the simulated data
  5. Topic 5 Model fitting and system identification.
    1. Load the measured transfer function from the end of Topic 3
    2. Fit a model transfer function to this measurement
    3. Make a digital filter representation of the fitted model
    4. Filter the temperature data with this filter
    5. Compare the PSD of the filtered temperature data and the IFO output
    6. Subtract the filtered temperature data from the IFO output
    7. Compare the IFO data with the temperature influence subtracted to the original IFO output
    8. (Time permitting) Repeat the exercise for the simulated from Topic 4
    9. (Still need something to do?) Repeat the steps of Topic 4 but this time fit a model to the measured temperature data and use a noise generator to make a simulated temperature data stream

©LTP Team