Cleveland State University

Department of Electrical and Computer Engineering

EEC 645/745, ESC 794

Intelligent Control Systems

Homework Assignments, Fall 2010

Homework problems are from Neuro-Fuzzy and Soft Computing, by Jang, Sun, and Mizutani (1997)

Homework 1 – Due Thursday September 2

·          Write a Matlab simulation of a simple dynamic system of your choice. Hand in a description of the system, your Matlab source code, and some input and output plots.

Homework 2 – Due Thursday September 9

·          Problems 2.5, 2.6, 2.11, 2.12, 2.20, 2.23(e) – For problem 2.23(e), email your Matlab code to the instructor

Homework 3 – Due Tuesday September 14

·          Problems 3.4, 3.6 (check errata for corrections), 3.9

Homework 4 – Due Tuesday September 21

·          Problems 4.1, 4.3, 4.6, 4.9

Homework 5 – Due Thursday September 30

·         Design a fuzzy controller for the system of your choice. Do some experiments. Some issues that you might want to investigate are: (1) How does the performance change with different fuzzy inference systems, or with different defuzzificiation methods? (2) How does the performance change with different rule bases? (3) How does the performance change with different membership functions, or with different membership function parameters? (4) How does the performance change with a different number of fuzzy classes for the inputs and outputs? (5) How does the performance change with a different fuzzy partition of the input space? (6) After you obtain a good fuzzy controller, how does the performance of the system change if your model parameters change? (7) Other issues that can be explored are limited only by your imagination.
You can use CruiseControl.zip, which is available on the course web site, as a template for your control system. Note that if your system has noisy inputs or outputs, then you will need to run a set of Monte Carlo simulations to obtain the average performance of your control system.

·         Present your findings to the class in a 10-15 minute informal presentation.

Homework 6 – Due Tuesday October 19

·         Investigate the performance of a neural network on the two-dimensional XOR problem. Study how various network training options affect performance. Present your findings in class

Midterm – Tuesday October 26

·         Open book, open notes. No electronic devices allowed.

Homework 7 – Due Thursday November 4

·         Review the course notes about neural networks. Design a neural controller for the system of your choice. Do some experiments and study some tradeoffs. Present your findings to the class in an informal presentation. Email your overhead slides to me.

Homework 8 – Due Tuesday November 16

·         Submit a project proposal to www.turnitin.com (don’t give me a hard copy). See the course web site for the class ID and password. The proposal should contain an abstract, introduction (including literature review), problem description, preliminary results, proposed work (including timeline), conclusion, and references. See http://academic.csuohio.edu/simond/courses/ReportTemplate.pdf for a paper template. I am not looking for a specific format, and you don’t have to follow the template, but it should give you an idea of what I am looking for.

Homework 9 – Due Tuesday November 30

·         Prove two forms of the matrix inversion lemma (see NeuralNets4.ppt slides 16 and 20).

·         Derive Equation 27 on slide 21 of NeuralNets4.ppt.

Homework 10 – Due Tuesday December 7

·         Final project presentations.

Homework 11 – Due Friday December 10

·         Submit your final project report to www.turnitin.com (don’t give me a hard copy).

Homework 12 – Due Friday December 10 (doctoral students only)

·         Submit your technical paper to www.turnitin.com (don’t give me a hard copy). This can be the same as your final project report, but the quality requirements of a conference or journal paper or much higher than those of a term project. See http://pages.cs.wisc.edu/~markhill/the_task_of_the_referee.pdf for an idea of the criteria which I will use to evaluate your paper. Send me an email with a link to the conference or journal site to which you are planning to submit your paper.