Goals of this program are to educate most qualified specialists for research and scientific development as well.

Graduates have outstanding knowledge in special area, determined by his/her PhD. project.
Special orientation is given by thesis;graduates are qualified for high positions in all branch.
Leader of research team,managers in industry.

The graduate of the subject field obtains broad knowledge of subject of cybernetics control and/or measuremet. The knowledge is build mainly on theoretical background of the subject. Moreover the graduate will obtain deep special knowledge aimed in direction of his/her thesis. The graduate will be able to perform scientific and/or applied research based on up to date theoretical knowledge. The graduate will be able to organize and lead a team of researchers in the studied subject.

prof. Ing. Pavel Václavek, Ph.D.

2. round (applications submitted from 04.07.2016 to 20.07.2016)

1. Application of orthogonal functions for signal prediction

   Application of orthogonal function for prediction. Focus on such orthogonal functions which are called double orthogonal, especially prolate spheroidal wave functions. Analyze the fundamental problems of this prediction such as the influence of sampling frequency, the effects of noise, etc. Propose a suitable algorithm on the base of the analysis and compare results with known methods.

   Tutor: Jura Pavel, prof. Ing., CSc.

2. Improving frequency resolution using orthogonal function.

   Focus on such orthogonal functions which are called double orthogonal, especially prolate spheroidal wave functions. Analyze the fundamental problems of approximation of signals such as the influence of sampling frequency, the effects of noise, etc. Propose a suitable method on the base of the analysis and compare your results with the current state of the art.

   Tutor: Jura Pavel, prof. Ing., CSc.

3. Localization of sound and vibration sources with non-contact methods

   The topic is aimed to the research of the methods and algorithms for non-contact localization and characterization of sound and vibration sources. Issues related especially with analysis of the sources with near-field acoustic holography method using microphone array will be studied with respect to its applicability for localization in confined space with reflections and other noise sources and also to increase the prediction accuracy with data fusion from other spatial measurement systems. In addition to the theoretical work, practical implementation of these methods and optimization of calculation algorithms for use in the field of non-contact vibrodiagnostics and localization of noise sources in mechanical systems will be carried out.

   Tutor: Beneš Petr, doc. Ing., Ph.D.

4. Self-tuning controllers in electrical drives

   Proposed topic concentrates on a field of automatic control which deals with self-tuning controllers for electromechanical systems. The aim is to propose identification experiments leading to system model with successive automatic controller / controllers tuning. The results are expected in a form of identification and control algorithms being tested in simulations in the environment of Matlab Simulink or Modelica and finally implemented and validated on a real hardware platforms and real electrical drives.

   Tutor: Blaha Petr, doc. Ing., Ph.D.

1. round (applications submitted from 01.04.2016 to 15.05.2016)

1. Data fusion for indoor localization applications

   Research and developmnent of data fusion methods for purposes of indoor location and tracking, taking advantage from particle filtering based metyhods, data fusion algoritms and enhanced sensor data representation in order to increase location accuracy and accuracy quantification. The research will be based on preexisting knowhow and will intend to improve estimation accuracy by enhancing amount of information that is available for the data fusion based estimation of location, velocity and acceleration. Part of the research will also be devoted to paralelization of relevant algorithms in order to tak eadvantage of multi-core computing platforms.

   Tutor: Fiedler Petr, doc. Ing., Ph.D.

2. Methods for determining the influence of mechanical impacts on the parameters of the sensors

   The topic is focused on research methods for testing systems using the SRS (Shock Response Spectrum). The goal is to design methodology and its verification for testing the resistance of MEMS sensors to mechanical shocks with SRS of defined shape. It will be necessary to solve the issue of automated measurement and generation of shocks with the appropriate time history using the available facilities, e.g. an electrodynamic or pneumatic shaker.

   Tutor: Beneš Petr, doc. Ing., Ph.D.

3. Methods for measurement of position and motion using fiber optics

   The topic is focused on research and development of advanced methods for measurement of position and mechanical motion based on optical principles using fibers. As part of the work, doctoral student will familiarize with interferometric and resonant principles employed in fiber-optic sensors for position and motion, especially with Sagnac effect for measurement of angular velocity. Subsequently, research will be focused on methods for processing electrical signals from optical detectors and their use for driving fiber-optic sensor system in a closed loop for practical implementation in all fiber sensors and assuming the use of conventional piezoelectric and electro-optical modulation components. The subject of the work will also include methodology for calibration of angular velocity sensors with focus on precise fiber-optic sensors for inertial navigation. This topic will be solved in relation to the national Competence center project focused on research and development of advanced sensors and sensor data processing methods.

   Tutor: Beneš Petr, doc. Ing., Ph.D.

4. Methods for vibrodiagnostics of non-rotating machines

   The topic is aimed to the research of the methods and algorithms for vibration diagnostic of non-rotating machines, especially for diagnostic method of bearings and gearboxes working under non-stationary conditions. Research will be focused on limiting factors in identification of actual methods for machines with a variable speed. Possibilities of modification or combination of these methods will be studied for achieving mechanical particles state estimation with better accuracy. In addition to the theoretical work, practical implementation and verification of these methods will be accomplished on mock-up.

   Tutor: Beneš Petr, doc. Ing., Ph.D.

5. Research of methods to improve localization accuracy inside buildings

   Research of indoor localization algorithms and methods based on wireless localization and communication system. Research is focused on assuring of quantitative localization parameters in wireless localization systems to assure quantification of quality and reliability of distance (location) measurement. The research will take advantage of data fusion methods in order to improve location accuracy based on combination of wireless localisation and information from inertial sensors. Research will therefore cover methods of data fusion, filtration and model based estimation to achieve acceptable indoor localization accuracy in challenging indoor conditions. One of the expected research outcomes is a methodology for placement of radio-beacons in order to minimize amount of beacons while maximizing location accuracy.

   Tutor: Bradáč Zdeněk, doc. Ing., Ph.D.

6. Supervised and Semi-supervised Machine Learning Methods for Tasks of Optical Character Recognition

   A student will aim at research of machine learning methods for such tasks of optical character recognition, where explicit algorithm is not possible to create. This topic is focused on classification of annotated inputs to several classes, i.e. on supervised learning and semi-supervised learning methods with only part of the input data annotated. An extra attention must be paid to a so-called structure data learning with output value of a structure or sequence. Such learning approach corresponds to complex tasks as a LPR characters recognition or general text recognition. A verification of designed model accuracy by e.g. cross-validation method must forms a final stage of the research.

   Tutor: Horák Karel, Ing., Ph.D.