System identification

In the final exam, students will be provided with a dataset. Each student must prepare a report and submit it on Moodle by the end of the exam period.

The report should include the following sections:

1. Input Data Analysis

• Identify the type of input signal (e.g., PRBS, White Noise, Sine-Sweep, etc.).

• Determine and explain the number of periods if the signal is periodic.

• Detect and remove any DC offset and trends from the data.

2. Frequency Response Identification

• Identify the system’s frequency response.

• Justify the choice of analysis method (e.g., Fourier analysis or spectral analysis), including the selection of windows and data grouping strategy.

3. System Structure and Order Estimation

• Estimate the system’s order and structure.

• Support your choice using multiple criteria such as frequency response, loss function, zero-pole cancellation, rank of Q, and delay estimation.

4. Model Identification and Validation

• Split the data into identification and validation sets.

• Identify and compare several parametric models (e.g., ARX, IV, ARMAX, OE, BJ, SS).

• Select the best model based on performance in:

• Time-domain validation

• Frequency-domain validation

• Statistical validation (including parameter variances)

5. Code and File Submission

• All MATLAB code used for analysis must be included in the final report.

• Submit only one PDF file, containing both the report and the MATLAB code.

The file Gyro400.mat contains the input and output data of a gyroscope system with a sampling time of 20 ms. This file can be used as a reference to help students prepare for the final exam.

It is strongly recommended that students prepare the structure of their report and MATLAB scripts in advance, so that during the exam only the necessary modifications—such as file names and model selections—need to be made.

Files and subfolders