Tutorial 2 Instructions Spring 2019

PwnHero ELE829 Spring 2019 1 Tutorial # 2: DIAGNOSTIC TOOLS & NON-PARAMETRIC MODELS IN TIME DOMAIN Objective The objectives of this Tutorial are: 1) to introduce you to non-parametric models in time domain that can be used as diagnostic tools for Modern System Identification of parametric models; 2) to reinforce your understanding of a simple parametric model identification using Ljung's OE Model that you were introduced to in Tutorial # 1; 3) to review a classical approach to model identification using time domain tools and to contrast this approach with the modern one. Logistics: Tutorial # 2 consists of three Parts. In spring semesters, it has to be completed in one two-hour session. The Tutorial is a MATLAB simulation and can be completed outside of the Control Systems lab. You will complete this Tutorial in pairs. Each pair will be working on a different set of data - you will be assigned a data set number in the first session. You should download the tutorial2_S2019.m file and work with only the assigned process. Each pair of partners will be briefly interviewed by the Course TA when handing in the Tutorial Report. Please refer to the posted Marking Sheet for the grading rubric. In Part 1 of this tutorial, you will try out the new techniques of "diagnosing" the discrete system structure (i.e. [𝑛𝑏 𝑛𝑓 𝑛𝑘] – see Theory section for definitions) in time domain - you will try to apply these techniques to a “noise-free” system, as well as to a “noisy” system. You will observe and comment on the specific diagnostic tools (subroutines impweigh, cra, correl and hanktest): how they compare in ease of extracting the required information, how much effect noise has on them. You will also assess how the diagnostics tools depend on the quality of input data (number of samples, frequency content and amplitude). Finally, you will determine the discrete system structure, based on the diagnostics. In Part 2 of this tutorial, you will obtain the system OE model from simulated data, as a first introduction to the modern identification approach. In Part 3 of this tutorial, you will apply a conventional approach to finding the unknown system transfer function using step response obtained in Part 1, based on what you have learned in ELE639. HINT: knowing the outcome (Part 2) should help you finding the correct solution. PwnHero.com