Course Outline (F2019)

ELE829: System Models and Identification

Instructor(s)Gosha Zywno [Coordinator]
Office: ENG463
Phone: (416) 979-5000 x 6105
Email: gzywno@ryerson.ca
Office Hours: Tue 2:30 - 4:30 PM, Thu 3:30 - 5:00 PM
Calendar DescriptionIntroduction to modern methods of linear system identification. Different types of models. Review of classic time- and frequency-based approach to empirical, 'black-box' system modeling. Non-parametric identification: impulse and step weights, spectral analysis. Parametric, discrete transfer function models from I/O data using Least Squares. Data-collection procedures, model structure selection, use of auto- and cross-correlation functions for diagnostics and model validation, overview of different estimation algorithms. Lab work consists of Matlab tutorials and an assignment dealing with identification of an unknown process. Course evaluation includes a group project selected from a list of topics in control system application, and its class presentation.
PrerequisitesELE 639




Compulsory Text(s):
  1. ELE829: Course Notes, M.S. Zywno, Copyrite 1999-2019. The notes are available from the secure course website, (login at https://my.ryerson.ca) as PDF downloadable files.
  2. MATLAB System Identification Toolbox (Matlab R2015) and System Identification Toolbox, User’s Guide, L. Ljung, the MathWorks, Inc., Copyrite 1995-2017, available on the Departmental Network as Matlab help files.
Reference Text(s):
  1. System Identification - Theory for the User, L. Ljung, Prentice Hall, 11th Edition, 2009.
  2. Modeling and Identification of Dynamic Systems, N.K. Sinha, B. Kuszta, VanNostrand, 1983.
Learning Objectives (Indicators)  

At the end of this course, the successful student will be able to:

  1. Demonstrates competency in developing mathematical models for deterministic systems (dynamic processes) and for stochastic systems (noise). Uses relevant computer simulation software - MATLAB System Identification Toolbox. Identifies and carries out steps required in performing a successful model identification procedure. Evaluates the effect of uncertainty in model parameters. (2b)
  2. Applies the tools for system identification to a real-time servomotor system, including obtaining experimental data. Selects appropriate analytical model for the real-time servomotor system, and verifies the model by comparing to experimental results. (3a)
  3. Selects appropriate analytical model for the real-time servomotor system, and verifies the model by comparing to experimental results. Assesses accuracy of the results obtained from the real-time servomotor system, verifying experimental data and explaining sources of possible discrepancies (non-linearity). (3b)
  4. Designs data collection experiments for diagnostics and identification of the model, selects appropriate model structure (BJ model) and noise filter function, and appropriate Least Squares Algorithm. (4b), (4a)
  5. Evaluates the quality of the derived system and noise models by validating against a set criteria, then improves the design until the model is validated. (4c)
  6. Demonstrates proficiency in the use of high-performance engineering modeling and analysis software, including System Identification Toolbox, in this course, and for subsequent engineering practice by completing and demonstrating to the professor the required simulation and analyses to perform system and noise model diagnostics, identification and verification. (5a)
  7. Helps other team members, and accepts help, on technical and team issues. Demonstrates capacity for team leadership while respecting others roles. Evaluates team effectiveness and plans for improvements. (6b)
  8. Produces a professionally prepared technical report using appropriate format, grammar, and citation styles, with figures and tables chosen to illustrate points made, with appropriate size, labels and references in the body of the report. Reports are graded on correctness, completeness, grammar, quality of graphics and layout. (7a)
  9. Responds appropriately to verbal questions from instructors, formulating and expressing ideas, using appropriate technical terminology this is assessed through comprehensive lab interviews by instructors. (7b)
  10. Demonstrates an understanding of project management principles, applying them both to the individual final project and to group tutorials. These include: negotiating the project scope, managing the deadlines, decomposing projects into key tasks and allocating responsibilities and resources according to deadlines. (11b)

NOTE:Numbers in parentheses refer to the graduate attributes required by the Canadian Engineering Accreditation Board (CEAB).

Course Organization

3.0 hours of lecture per week for 13 weeks
1.0 hours of lab/tutorial per week for 12 weeks

Teaching AssistantsTBA
Course Evaluation
Course Activities 15 %
Lab/Tutorial Project (Group) #1 9 %
Lab/Tutorial Project (Group) #2 9 %
Lab/Tutorial Project (Group) #3 9 %
Lab/Tutorial Project (Group) #4 13 %
Final Project Report (Individual) 45 %
TOTAL:100 %

Note: In order for a student to pass a course with "Theory and Laboratory" components, in addition to earning a minimum overall course mark of 50%, the student must pass the Laboratory and Theory portions separately by achieving a minimum of 50% in the combined Laboratory components and 50% in the combined Theory components. Please refer to the "Course Evaluation" section for details on the Theory and Laboratory components.

ExaminationsCourse evaluation is ongoing and semester-long, and includes both group work (lab/tutorial reports) and individual effort (final project). All reports include simulations – Matlab codes are submitted and parsed for originality and efficacy. The course professor personally verifies all individual codes submitted with the final report. If the execution of the code does not support claims in the report, the project will automatically receive an automatic deduction in the grade.
Other Evaluation InformationCourse activities are part of the ongoing and semester-long evaluation: there are graded activities in every week of classes (on top of scheduled tutorial/lab reports).
 The graded activities include both individual in-class assessments (pop quizzes), and homework assignments, including computer simulations on the application of theory learned, which are then demonstrated to the professor in class.
 NOTE: The four lab/tutorial reports have different weights.
Other InformationNone

Course Content



Chapters /

Topic, description

Week 1


Goals for the course. Introduction to modern system identification:
 diagnostics identification validation. Types of models. Introduction to Tutorial # 1: Non-parametric models in frequency domain. Frequency
 response for conventional modeling - review of Bode plots. Data
 Collection - PRBS signal. Introduction to Matlab System Identification
 (References: Course Notes Chapters 1 2 3)

Week 2


Non-parametric frequency response models: SPA ETFE the effect of
 noise data filtering. Transfer function models conversions between
 continuous and discrete representations sampling sampling. Simple
 Box-Jenkins (BJ) model structure: OE Model (deterministic process
 white noise).
 (References: Course Notes Chapters 1 2 3)

Week 3


Diagnostic tools in frequency domain - summary. Introduction to
 Tutorial # 2: Non-parametric models in time domain. Review - time
 domain response for conventional modeling (Step and Impulse response
 plots). Review – basic definitions of stochastic processes.
 (References: Course Notes Chapters 2 4)

Week 4


Non-parametric models in time domain: impulse and step weights from
 de-convolution and from correlation analysis. The effect of noise on
 non-parametric models in time domain. Simple parametric non-robust
 discrete transfer function models from impulse weights. Hankel Test of
 system order.
 (References: Course Notes Chapters 2 4)

Week 5


Diagnostic tools in time domain – summary. Review of matrix algebra.
 Introduction to Least Squares method. Robust parametric models. The
 effect of noise on conventional parametric models (non-robust and
 (References: Course Notes Chapters 4 5)

Week 6


Introduction to time-series modeling. Combined dynamic-stochastic
 models - Box-Jenkins structures. Overview of different parameter
 estimation algorithms. Figures of Merit for Model selection: Akaike
 Index Loss Function.
 (References: Course Notes Chapter 6)

Week 7


Refining OE model: ACF PACF and CCF checks. Complete Validation
 for OE Model: Residue whiteness testing - Chi-Square tests Confidence
 Intervals. Introduction to Tutorial # 3: Stochastic noise models.
 (References: Course Notes Chapter 6)

Week 8


Stochastic noise models: AR MA ARMA and “Random Walk”
 processes. Auto- and Partial Auto-Correlation Functions as diagnostic
 tools for stochastic noise models.
 (References: Course Notes Chapters 6 7)

Week 9


Summary of all diagnostic tools for all Box-Jenkins models: non-
 parametric time and frequency domain models Auto- and Partial Auto-
 Correlation functions.
 (References: Course Notes Chapters 7 8)

Week 10


Refining BJ model: ACF PACF and CCF checks. Complete Validation
 for BJ Model: Residue whiteness testing - Chi-Square tests Confidence
 (References: Course Notes Chapters 7 8)

Week 11


Review - designing data collection experiment model structure
 selection complete diagnostics structure revisions and final model
 validation. Examples of a full system identification procedure.
 (References: Course Notes Chapters 6 7 8)

Week 12


Overview of the Final Project (individual): "Black Box" System Identification of two systems ("Easy" and "Difficult"). Questions and
 answers regarding the project.
 (References: Course Notes Chapters 6 7 8)

Week 13


Questions and answers regarding the final project active consultation on
 final project computer simulations.
 (References: Course Notes Chapter 8)

Laboratory/Tutorials/Activity Schedule






Tutorial #1: Diagnostic Tools in Frequency Domain and Simple Model identification - OE Model (2 sessions):
 Part 1: Non-Parametric Models in Frequency Domain as Diagnostic Tools
 Part 2: Simple Model Identification using OE Model
 Part 3: Conventional Parametric Model from Frequency Response Data.



Tutorial #2: Diagnostic Tools in Time Domain and Simple Model identification - OE Model (2 sessions):
 Part 1: Non-Parametric Models in Time Domain as Diagnostic Tools
 Part 2: Simple Model Identification using OE Model
 Part 3: Conventional Parametric Model from Frequency Response Data.



Tutorial #3: Stochastic Noise Models (2 sessions):
 Identify structure of four different noise models.



Tutorial #4: Simple System Identification of a Real-Life System (Servomotor) (3 sessions):
 Part 1: Obtaining Experimental Frequency and Time Domain Responses
         from the Servo-motor
 Part 2: Model Identification and Comparisons with Nominal Values Model.



Consultations on Final Project
 Final System Identification Project - “Black Box” Models for two systems: OE-type and BJ- or PEM-type (3 sessions). Students use the tutorial time to work on their final projects (diagnostics identification validation).

Policies & Important Information:

  1. Students are required to obtain and maintain a Ryerson e-mail account for timely communications between the instructor and the students;
  2. Any changes in the course outline, test dates, marking or evaluation will be discussed in class prior to being implemented;
  3. Assignments, projects, reports and other deadline-bound course assessment components handed in past the due date will receive a mark of ZERO, unless otherwise stated. Marking information will be made available at the time when such course assessment components are announced.
  4. Refer to our Departmental FAQ page for information on common questions and issues at the following link: https://www.ee.ryerson.ca/guides/Student.Academic.FAQ.html.

Missed Classes and/or Evaluations

When possible, students are required to inform their instructors of any situation which arises during the semester which may have an adverse effect upon their academic performance, and must request any consideration and accommodation according to the relevant policies as far in advance as possible. Failure to do so may jeopardize any academic appeals.

  1. Health certificates - If a student misses the deadline for submitting an assignment, or the date of an exam or other evaluation component for health reasons, they should notify their instructor as soon as possible, and submit a Ryerson Student Health Certificate AND an Academic Consideration Request form within 3 working days of the missed date. Both documents are available at https://www.ryerson.ca/senate/forms/medical.pdf.. If you are a full-time or part-time degree student, then you submit your forms to your own program department or school;
  2. Religious, Aboriginal and Spiritual observance - If a student needs accommodation because of religious, Aboriginal or spiritual observance, they must submit a Request for Accommodation of Student Religious, Aboriginal and Spiritual Observance AND an Academic Consideration Request form within the first 2 weeks of the class or, for a final examination, within 2 weeks of the posting of the examination schedule. If the requested absence occurs within the first 2 weeks of classes, or the dates are not known well in advance as they are linked to other conditions, these forms should be submitted with as much lead time as possible in advance of the absence. Both documents are available at www.ryerson.ca/senate/forms/relobservforminstr.pdf. If you are a full-time or part-time degree student, then you submit the forms to your own program department or school;
  3. Academic Accommodation Support - Before the first graded work is due, students registered with the Academic Accommodation Support office (AAS - www.ryerson.ca/studentlearningsupport/academic-accommodation-support) should provide their instructors with an Academic Accommodation letter that describes their academic accommodation plan.

Academic Integrity

Ryerson's Policy 60 (the Academic Integrity policy) applies to all students at the University. Forms of academic misconduct include plagiarism, cheating, supplying false information to the University, and other acts. The most common form of academic misconduct is plagiarism - a serious academic offence, with potentially severe penalties and other consequences. It is expected, therefore, that all examinations and work submitted for evaluation and course credit will be the product of each student's individual effort (or an authorized group of students). Submitting the same work for credit to more than one course, without instructor approval, can also be considered a form of plagiarism.

Suspicions of academic misconduct may be referred to the Academic Integrity Office (AIO). Students who are found to have committed academic misconduct will have a Disciplinary Notation (DN) placed on their academic record (not on their transcript) and will normally be assigned one or more of the following penalties:

  1. A grade reduction for the work, ranging up to an including a zero on the work (minimum penalty for graduate work is a zero on the work);
  2. A grade reduction in the course greater than a zero on the work. (Note that this penalty can only be applied to course components worth 10% or less, and any additional penalty cannot exceed 10% of the final course grade. Students must be given prior notice that such a penalty will be assigned (e.g. in the course outline or on the assignment handout);
  3. An F in the course;
  4. More serious penalties up to and including expulsion from the University.

The unauthorized use of intellectual property of others, including your professor, for distribution, sale, or profit is expressly prohibited, in accordance with Policy 60 (Sections 2.8 and 2.10). Intellectual property includes, but is not limited to:

  1. Slides
  2. Lecture notes
  3. Presentation materials used in and outside of class
  4. Lab manuals
  5. Course packs
  6. Exams

For more detailed information on these issues, please refer to the Academic Integrity policy(https://www.ryerson.ca/senate/policies/pol60.pdf) and to the Academic Integrity Office website (https://www.ryerson.ca/academicintegrity/).

Important Resources Available at Ryerson

  1. The Library (https://library.ryerson.ca/) provides research workshops and individual assistance. Inquire at the Reference Desk on the second floor of the library, or go to library.ryerson.ca/guides/workshops
  2. Student Learning Support(https://www.ryerson.ca/studentlearningsupport) offers group-based and individual help with writing, math, study skills and transition support, and other issues.