A brief description of the Doctoral School of Electrical Engineering

The doctoral school was established in 2000 to pursue research and development in the field of Electrical Engineering. The school integrates the research activities of ten departments in the Faculty of Electrical Engineering and Informatics, Budapest University of Technology and Economics. The main directions of the scientific activities in the school include embedded systems, electronics, electronic technology and devices, measurement, identification and control, infocommunication systems, electrical machines and drives, power engineering. Since 2002 more than 134 applicants have successfully defended their PhD theses and have been awarded with the degree. The school requires from applicants and supervisors proper scientific research with publications in peer reviewed journals and other professional forums.

2. Research and development projects supported by TÁMOP-4.2.2/B-10/1-2010-0009

The main objective of research supported by the TAMOP project is to investigate problems requiring substantial computing power. The project in the field of electrical engineering include the following R&D subprojects: (i) Embedded systems; (ii) Electromagnetic nondestructive testing; (iii) Analysis of large scale electromagnetic problems; (iv) Modeling, pattern recognition and control; (v) Intelligent robot control;

(vi) Metallurgical and simulation research of metal microstructure formations; (vii) Smart grid analysis with intelligent methods.

2.1 Embedded systems demanding significant computing power

There are several aspect of that field, which are investigated in this project. Model transformations are frequently used means for automating software development in various domains, including embedded systems, to improve quality and reduce production costs. Debugging of model transformations often necessitates identifying parts of the transformation program and the transformed models which have causal dependence on a selected statement. In traditional programming environments, program slicing techniques are widely used to calculate control and data dependencies between the statements of the program. The key point is to introduce program slicing for model transformations where the main challenge is to simultaneously assess data and control dependencies over the transformation program and the underlying models of the transformation.

In bioinformatics related research enormous amount of biological and clinical data should be analyzed. New statistical methods and efficient algorithms are required. The new solutions needed should utilize existing biological knowledge sources, which means that different formats, standards and contents are to be dealt with. Basically MCMC (Markov Chain Monte Carlo) methods have been applied in parallel computing environment. The

methods implemented and tested until now were applied in various biomedical fields, such as the genetic background of asthma, allergy etc.

2.2 Electromagnetic nondestructive testing

Research in this field requires developing various testing methods. Two different methods have been implemented supported by the project for the simulation of electromagnetic field perturbation due to the defect-probe interaction of nondestructive testing. These methods are particularly suitable for the parallelization in the CUDA environment, because of their low memory consumption and atomized nature. One of them is the element-by-element finite element method (EbE-FEM), while the other is the method of moments (MoM). The implemented algorithms are currently under testing.

Two papers have been submitted in this topic, and one of them has already been appeared in the IEEE transactions on Magnetics.

2.3 Analysis of large scale electromagnetic problems

Electromagnetic compatibility is an emerging problem during the design of electronic devices. Standard test procedures are defined in national and international standards to ensure the interoperability of such devices. The tests are performed in special rooms, i.e.

anechoic chambers, that have absorber material on the walls to provide low reflection coefficient. However, full-wave analysis of anechoic chambers is challenging, because (1) the absorber structure covering the walls of the chambers is complex, therefore, requires a dense mesh; (2) the chamber is electrically large at high frequencies, resulting a fine mesh in the analysis domain. The aim of this R&D project is to find a method that can be used for the analysis of such electrically large problems: (i) to build the equivalent model of the spatially periodic absorber structure; and (ii) analyze the full chamber using the equivalent model on the walls.

International literature and standards specify several methods for assessing the level of lightning protection required for certain structures and also provide several methods for assessing the safety provided by certain arrestor arrangements. Two such methods are the rolling sphere and the attractive volume methods. Recent developments in the area permit us to use direct methods for the simulation of lightning events and thus the computation of the lightning paths, which enable us to objectively assess the level of protection provided by an arrangement of arrestors. Such computations enable both the effective design of lightning protection for structures that have unusual characteristics – both if their geometries are unique and if they require a very high level of protection. Also, computer simulation is valuable in validating the effectiveness of arrestor systems that were designed using conventional methods. Such computations require the simulation of many lightning events, each of which – in order to produce meaningful results – must handle large geometries. Both of these aspects make this problem a candidate for acceleration by supercomputing clusters and/or GPGPU computations.

2.4 Modeling, pattern recognition and control

The goal of parametric system identification is to provide estimates for parameters of certain model structures based on given measurement data. This problem can always be presented as an optimization problem with an appropriate choice of cost and constraint functions. Apart from the simplest cases the resulting optimization problems are nonconvex with multiple local minima. Due to the existence of these cicumstances, there are usually no guaranties that the model resulting from a given identification method is a global minimizer. The research aims at the application of semidefiite programming (SDP) relaxation techniques to the optimization problem arising in time domain identification.

From the solution of the defined sequence of SDPs a sequence of system models can be extracted that converges to the globally optimal system model. In the project the SDP technique for polynomial optimization problems (POP) will be investigated to solve identification problems.

Although computerized verification of scanned, handwritten signatures has been extensively studied in the past three decades, the results in the field are still unsatisfactory. Because the lack of a widely accepted formalism, advancements achieved by different researchers are hard to combine and even to compare. To address these problems a simplified formal model for off-line signature verification is proposed, where each of the verification steps can be mathematically modeled. Feature extraction aims at setting up a proper feature space and sample signatures are to be analyzed to predict the achievable precision of a signature verification system based on the cardinality and the quality of input samples.

Developing the control of modern power converters is a very expensive and time-consuming task. Time to Market can take unacceptable long. FPGA-based real-time simulation of a power stage with analog measured signals can reduce significantly the cost and time of testing a product. This new approach is known as HIL (Hardware-In-the-Loop) testing. A general power converter consists of two main parts: a power level (main circuit) and a digital controller unit, which is usually realized by using some kind of DSP.

Testing the controller HW and SW is quite problematic: live tests with a completely assembled converter can be dangerous and expensive. A low-power model of the main circuit can be built under laboratory conditions, but it will have parameters (e. g. time constants and relative losses) differing from the ones of the original system. The solution is the HIL simulation of the main circuit. With this method the simulator can be completely transparent for the controller unit, unlike other computer based simulation methods. The goal of the research in this field is to develop such a real-time simulator using FPGA. The modeled circuit is a three-phase inverter, which is widely used in power converters of renewable energy sources.

2.5 Intelligent robot control

Research goals in this field include developing efficient algorithms supporting various applications of mobile robots. The algorithms refer to the following areas:

• Navigation algorithms in known and unknown environment

• Algorithms to create maps

• Simultaneous navigation and map construction.

Mobile robots – in general – are moving under the assumption of either known or (partially) unknown environment conditions. Within the frame of the ongoing research efficiency of robot navigation is to be improved by using the DWA (Dynamic Window Approach) methodology accompanied by the control policy called Receding Horizon Control. The main goal is to elaborate methods not requiring the repeated evaluation of the navigation function.

Machine vision systems for mobile robotics demand high computational capacity. We not only deal with intensity images but depth maps also which increases computational costs.

We utilize both images simultaneously to improve landmark detection and redetection rates to create precise 3D maps for robot localization and navigation algorithms in unknown environments. Although these algorithms must be optimized for energy efficient field applications, during research and development phase of the methods, a less optimized version is implemented to generate more data thus aiding evaluation and refinement processes. For the optimization step parallelization of machine vision problems is being evaluated.

2.6 Metallurgical and simulation research of metal microstructure formations Due to the RoHS (Restriction of Hazardous Substances) directive of the European Union, novel manufacturing trends in the electronics industry point towards the lead-free soldering technology. This field applies a lot of different surface coatings and differing process parameters compared to the common leaded soldering technology. This results in a lot of new (or previously solved) reliability issues evoking more and more research topics. In this project we focus to the following fields: intermetallic layer formation during common convection reflow process and vapour phase soldering; and the phenomenon of electrochemical migration. The aim of this R&D project is: (i) to develop simulation models of the vapor phase soldering and the electrochemical migration, to implement and verify them with experimental results; (ii) and to investigate the role of the intermetallic layer formation in different reliability issues such as whisker formation or electrochemical migration.

2.7 Smart grid analysis with intelligent methods

The traditionally hierarchical structure of electric networks are being gradually transformed, the main drivers of this change include:

− Large-scale integration of renewable energy sources

− The generation of a significant part of these small plants are volatile and often stochastic

− Info-communication and measurement technology are becoming cheaper, allowing for an increasing level of instrumentation and consumer interactivity

− Increasing security of supply requirements

− The continued development of electricity markets and regional market coupling efforts.

The evolving intelligent network (Smart Grid) raises a number of questions regarding its expected future operation. The goal of the project includes research and testing of new system management principles and algorithms, and making recommendations to prevent breakdowns and system recovery at various levels of system control hierarchy. Further scope of the research covers investigations of the forthcoming Europe-wide interconnection of organized electricity markets and its effects on efficient utilization of network components, system security and overall economic operation.