PhD Students


Despoina Karamichailidou

Thesis Title:

Development of computational intelligence methods with emphasis on adaptive techniques for big data

Thesis Synopsis:

The objective of this dissertation is the development of computational intelligence methods with increased efficiency, in order to be able to deal with large-scale data; emphasis will be given on neural network training algorithms. Big Data analytics is increasingly receiving attention and importance, as it holds unique promises, but on the other hand, it is characterized by numerous computational and statistical challenges. The improvement and further development of conventional and non-conventional methods, together with the corresponding adaptive algorithms will be crucial for the proper management of large-scale data. The produced models will be tested on large-scale data, coming from various scientific areas, in order to evaluate their ability to handle Big Data challenges, aiming at both statistical accuracy and computational efficiency for real-world applications.

 

 

Myron Papadimitrakis

Thesis Title:

Development of computational intelligence methods for system optimization and control

Thesis Synopsis:

The purpose of this PhD Thesis is to develop algorithms based on computational intelligence with application in optimization and automatic control of complex and non-linear systems. The development of high-accuracy models using neural network and machine learning methods, as well as multi-objective optimization techniques using evolutionary computation and swarm intelligence methodologies is of high priority. As far as automatic control is concerned, special importance is placed on model predictive control, where new, optimized methodologies will be applied to develop distributed controllers based on neural network prediction models and evolutionary computation methods.

 

 

Aristotelis Kapnopoulos

Thesis Title:

Automatic control of unmanned vehicles using computational intelligence

Thesis Synopsis:

The objective of this PhD Thesis is related to the development of automatic control methods in order to create intelligent unmanned aerial vehicles (UAVs). Emphasis will be given to the study of quad-copters, as their control comprises a challenging task due to its dynamic behavior, which exhibits nonlinear, under-actuated and strongly coupled terms. Computational intelligence methods as fuzzy logic, neural networks, evolutionary computation and swarm intelligence will be used in order to provide elements of intelligence to the aerial vehicles. Special importance will be placed on the development of model predictive control (MPC) methods, and particularly in solving the optimization problem formulated during the implementation of an MPC controller using swarm intelligence methodologies and in developing new dynamic predictive models based on neural networks.

 

 

Ioannis Kordatos

Thesis Title:

Development of model predictive control methods for nonlinear systems using machine learning techniques

Thesis Synopsis:

The objective of this PhD dissertation is the development of MPC (Model Predictive Control) and EMPC (Economic Model Predictive Control) algorithms for non-linear systems using computational intelligence and machine learning models for system modelling. Emphasis will be given to RBFNs (Radial Basis Function Networks) for predicting the behavior of these dynamical systems. Comparative analysis of the ML models in terms of accuracy and computational time for system modelling will be conducted by testing them to a number of open-source datasets from various scientific fields, in order to capture the performance of the models that prevail. From the perspective of automatic control, the development of MPC schemes will be realized by integrating the ML models, as predictive models utilized by the controller. Additional study will be implemented for tuning the controllers regarding the suitable operating points, while also considering stability and robustness issues. Finally, MPC controllers in conjunction with the developed ML models, will find real-world application in a waste treatment plant.

 

 

Theodoros Protoulis

Thesis Title:

Nonlinear system control using computational intelligence techniques

Thesis Synopsis:

The objective of the present PhD thesis is the development of control methods, suitable for controlling complex multiple input - multiple output nonlinear dynamical systems. Particular emphasis will be given to the development of economic predictive control (EMPC) techniques, indented to be applied to complex physicochemical processes such as wastewater treatment plants. The control of these processes deals with many challenges due to their high complexity and as a result, development of controllers able to satisfy certain specifications not only regarding the operating points of these processes, but also regarding the consumption of energy resources, is indeed a challenging task. For the purpose of achieving these goals, in the context of this PhD thesis, methods of system identification based on computational intelligence techniques such as fuzzy logic, neural networks, evolutionary computation and swarm intelligence will be developed. Further to the aforementioned objectives, particular emphasis will be devoted to the study of stability and robustness of the designed controllers. More specifically, in order to prove the global stabilization of the dynamical systems with the integration of the controllers, methods of studying stability and robustness based on Lyapunov functions and theorems will be applied. Finally, part of the present PhD thesis will be committed to ways of formulating and solving the mathematical optimization problems to be found in the procedures of system identification and predictive control.