RAJCHL, M. Realizace elektromagnetické polohovací platformy pro testování nelineárních řídících algoritmů a identifikačních metod [online]. Brno: Vysoké učení technické v Brně. Fakulta strojního inženýrství. 2020.

Posudek vedoucího

Brablc, Martin

Student Matej Rajchl vytvořil v rámci své diplomové práce experimentální a výukový laboratorní model magnetického manipulátoru. I přes ztížené podmínky vlivem proti-pandemických opatření se mu podařilo realizovat funkční zařízení, i když jeho konstrukci bude ještě možné dopracovat. K řešení přistupoval aktivně, iniciativně a velmi samostatně. Předložená práce je psána dobrou angličtinou a dovolím si tvrdit, že metody v ní použité a také způsob jejich popisu přesahují běžné znalosti studentů magisterského studia v oblastí modelování, řízení a optimalizace. Práci doporučuji k obhajobě s hodnocením A - výborně.

Posudek oponenta

Takács, Gergely

The thesis titled “Design of electromagnetic positioning platform for testing of nonlinear control and identification algorithms” has a total of 63 pages in its electronic version, with a page count of 56 in print. The work is divided into six chapters, contains a bibliography and two appendices. There are 38 bibliographical citations throughout the text, providing a satisfactory cross-section of relevant literature. Note, that the page numbers shown in the review refer to those printed on the page. The reviewer declares no conflict of interest. After a succinct introduction, the author presents the state of the art in Chap. 2. Similar devices are introduced in the beginning of the chapter, where the reviewer feels that just a bit more detail would be welcome on these pieces of hardware. The author continues with an excursion into magnetic models, and continues with a summary of control, estimation, stability and controllability theory. This part on systems control theory is a strong point of the chapter. The text is very well written, readable and the mathematical notation is rigorous and consistent. This part alone may serve as a good introduction or recapitulation to my own students in the future. The next chapter outlines the design of the proposed device. The author provides just about the right amount of context to justify his choices. The text goes through most of the essential components such as the coil, position sensor, power electronics, etc., and presents the solutions developed within the framework of the thesis. I consider the electronic circuit design of high quality. However, the mechanical realization of the device leaves a bit to desire. Given the unusual events caused by the SARS-CoV-2 pandemic, this can certainly be excused. The fourth chapter discusses matters connected with modelling and control algorithm tuning and simulation. First, a nonlinear model of the ball dynamics for a single cool is acquired by a clever combination of experimental measurements and the so-called Sparse Identification of Nonlinear Dynamics (SINDy) algorithm. This is followed by repeating the outlined procedure for all three coils, meanwhile showing graphical controllability and stability analysis. The graphical illustration of controllability analysis is an excellent addition to the discussion, I even dare to say that it is a teachable example of control concepts. The next section serves to test three different control algorithms to compare their performance in simulation. The overall procedure is logical and instructive, the author demonstrates a deep understanding of the concepts presented. The only criticism is that the author did not explain well enough how the identification data was acquired in Fig. 4.2, the reviewer was left to guess that (probably) the ball has been moved manually with different current levels engaged. The fifth chapter is practically the culmination of the work, where the author tests the proposed control algorithms and compares their performance and activity. It is especially exciting to see the imprecise tracking that may be attributed to the spatially-dependent controllability outlined in the previous chapter. The experimental part provides enough context to prove the base functionality of the device, and to compare the performance of various algorithms. Shall the author (or anyone) continue this type of work, a nonlinear model predictive controller (NMPC) would be an exciting candidate – albeit computationally very expensive. The thesis is written in excellent English with a minimal number of typos and awkward formulations found throughout the text. The mathematical notation is rigorous, the typography (prepared in LaTeX) is also of high quality. My only criticism is directed towards certain figures. For example, the author could re-make some line-drawing illustrations ad the fonts used in graphs should match that of the text. Some minor comments: - A list (summary) of mathematical notation would be nice to have at the beginning. - I recommend using abbreviations for standard terms referenced within sentences, such as Fig. instead of figure, Eq. instead of equation, etc.. - p. 7, spelling of TopView(left), etc. is unusual and needs spaces. - p. 8, “solution does not exists” - p. 8, The differentiation terms should be typed with upright font, e.g. use \mathrm{} next time. - p. 9, I would recommend to cite the sources through a single command, e.g. to produce [33,36] instead of [33] [36]. - p. 11, Re-drawing line art such as in Fig. 2.5 would greatly improve the graphical quality of the presented thesis. - p. 12, Even though a list of abbreviations is provided, it is better to spell out abbreviation when used for the first time (e.g. LTI). - p. 13, “in which for simplicity” should not begin in a new paragraph (wrong indentation). - p. 13, line extending the margins “(det(G(x))…” - p. 13, Citations should only be shown in context, e.g. “states [26].” Instead of “states. [26]”. - p. 16, “[…] sensor bias, etc.” instead of “sensor bias etc” - p. 16, “discreetly” discretely - p. 17, the dot should be over z in z_k (just below Eq. (2.40)) - Throughout the thesis: “however” should (almost) be followed by a comma. - p. 18, Re-drawing Fig. 2.8 would not bee too much of a trouble, yet would increase the value of the thesis. - p. 19, Fig. 2.9 (and elsewhere) Font size and type in plots should match that of the general text. - p. 19, Fig. 2.9 Missing axis labels. (e.g. state – input? or state 1 and state 2?) - p. 20, Perhaps an illustration for the Gramian would be nice to see here. - p. 21, The author should have forced LaTeX to place the table better, this enormous white space is distracting. - p. 22, Fig. 3.1 caption: missing space before parentheses - p. 22, missing space before “IR” - p. 24, (possibly elsewhere) Units should be typed upright, with a space after the number, see 0.6ms. - p. 24, Text in the margins. - p. 29, “where…” after Eq. 4.2 should not be indented. - p. 31, “tun-ability” tunability - p. 33, Typing of units, second line below Fig. 4.5. - p. 40. “First we”, “First the” etc. missing a comma. - p. 43. Eq. 4.35, missing parentheses. - p. 44. Fig. 4.16 would be especially hard to read in print (fortunately the reviewer has the electronic version). - p. 46. Missing space after full stop. (“.The”) - p. 48, Tab. 5.1 Typing of units, missing spaces. Conclusion: Overall, the reviewer feels that the presented thesis is of exceptional quality and most definitely recommends its defense in front of a committee. The reviewer recommends the committee to award this master’s thesis by the mark “A”.

eVSKP id 125131