Control Theory (Control Engineering Lab.)

Videos about Control Engineering

I’m Hiroshi Okajima. An associate professor at Kumamoto University, Japan. I majored in control engineering and have been a researcher in control engineering for 20 years.

*Main YouTube channel about Control (7800 subscribers)* is mainly for Japanese person.

English HP of Control System LAB:
sites.google.com/view/okajima-lab/en

In this channel, you can watch videos about control engineering. The main purpose of these videos is to make the outline of control theory widely known and for supplementary materials such as university lectures. There are three main topics of explanation in the videos as follows

(1) Introduction to Control Engineering
(2) Control Theory

In the videos, each item is explained in detail, including many numerical simulations. (Videos longer than 10 minutes often include numerical simulations such as response waveforms. ) .

In some of the videos, Matlab simulations are run and the associated code can be obtained.

View channel on YouTube

Switch Invidious Instance

Videos

Shorts

Playlists

Posts

Control Theory (Control Engineering Lab.)

Upload new video. ACC based on state-space model (state feedback control). #MATLAB

4 months ago | [YT] | 1

View 0 replies

Control Theory (Control Engineering Lab.)

Add new blogs.

Stability Analysis and Pole Placement Control for Discrete-Time Systems
blog.control-theory.com/entry/2025/04/20/122439

Advanced LMI Techniques in Control System Design
blog.control-theory.com/entry/2025/04/20/173054

4 months ago | [YT] | 1

View 0 replies

Control Theory (Control Engineering Lab.)

ieeexplore.ieee.org/document/10858707

On January 30, 2025, our paper on system identification methods for periodic time-varying systems was published.

This paper was published in @IEEEAccess @IEEEorg and is a joint work with Professor Yusuke Fujimoto, Professor Hiroshi Oku, and Haruto Kondo (our former lab member).

For linear time-invariant (LTI) systems, methods such as the prediction error method and subspace methods for obtaining mathematical models have already been well established. However, when dealing with periodic time-varying systems, the problem becomes significantly more challenging. Although various related papers have been published in journals such as Automatica, achieving high accuracy remains difficult. Several ideas utilizing periodic inputs have been proposed.

Our paper employs a method called "cyclic reformulation" which treats periodic time-varying systems as time-invariant ones. This approach was originally proposed before the year 2000 by S. Bittanti. However, since the system order in this representation becomes the product of the state dimension and the period, it has not been widely used (it is few to find relevant literature on this topic).

To summarize our findings: by utilizing the cycling-represented signals and applying existing subspace identification methods, we successfully obtained highly accurate models. The cycling approach had been used in our previous studies on multi-rate systems, which were co-authored with Professors Hosoe and Hagiwara from Kyoto University. Note that I first learned about this method around 2017 from Professor Hosoe, who introduced it to me.

A major contribution of this paper is the characterization of the properties of the Markov parameters in the cyclic representation of periodic time-varying systems. By leveraging these properties, we derive a coordinate transformation matrix for the state-space model of large system order (state dimension × period), enabling us to extract the coefficient matrices of the desired system.

In our numerical examples, we applied random inputs to the original system and identified the system model using input-output pairs, achieving highly accurate mathematical models.

The idea of obtaining a state-space model from cycling-represented signals and then applying a coordinate transformation to derive the cycling representation itself required significant effort to realize.

Rather than proposing a completely new identification method, this research serves as a system integration-type contribution, bridging "cyclic reformulation" and "system identification."

We intend to further advance this research by exploring new theoretical developments and practical applications. We sincerely hope that you will take the opportunity to read our paper through the provided link, as your insights and feedback would be highly valuable for the continued refinement of our work.

Cyclic Reformulation Based System Identification for Periodically Time-varying Systems | IEEE Journals & Magazine | IEEE Xplore