Laboratório de Sistemas Eléctric

Laboratório de Sistemas Eléctricos Industriais

Project Name	Duration	Funding Agency
'Intelligent' Electromechanical Systems - Modelling and Control	1995-1998	JNICT (PBIC/C/TPR/2368/95)
The Sciences of Complexity: From Mathematics to Technology to a Sustainable World	1998 - 2001	ZiF (Center for Interdisciplinary Research - University of Bielefeld)
Modelling, Learning and Control of Electrical Drives with Automata Network	1999-2000	IPS
New Topologies for Power Converters with 'Intelligent' Control	1999 - 2001	IPS
SRM Machine - Global Models Development and Performance Analysis as Motor and Generator	2002-2004	FCT (POCTI/ESE/39470/2001)
Determinação de padrões de consumo em subestações de distribuição de energia eléctrica	2002-2003	IPS
RenH2 – Stand-Alone Energy System Supported by Totally Renewable Hydrogen Production	2005 - 2008	FCT (POCTI/ENR/59623/2004)

'Intelligent' Electromechanical Systems - Modelling and Control

The Sciences of Complexity: From Mathematics to Technology to a Sustainable World

Introduction
Many systems in the natural sciences and humanities fall under the heading of "Complex Systems". Many are multi-agent systems, the agents being complex molecules, cells, living organisms, animal groups, human societies, industrial firms or competing technologies.

But what is a complex system?
As always, notions originate from the observation of facts. And the most general observation here is that, when a set of evolving agents interacts, the resulting global system displays collective properties which look qualitatively different from a simple superposition of their elementary behaviors. The common structural features of such systems are non-linearity, interdependence and emergence.

Non-linearity means that the dynamical behavior of the system cannot be viewed as a superposition of the elementary effects of its components, nor reconstructed from elementary "modes". In simple terms, doubling the input does not necessarily double the output.

Interdependence means that the response of each one of the agents, from which the system is built, depends on the evolution of the others, in fact in a self-consistent manner.

Emergence means the creation of collective properties qualitatively different from the individual behavior.

Because these systems share some common structural properties, it is tempting to ask:

Is there a unified theory of complex systems?
Probably not. Unifying ideas like the "the edge of chaos", self-organized criticality, etc, have, at times, been proposed as general paradigms for all complex adaptive systems. But in all cases examples are found of systems that do not follow these general schemes and nevertheless look fairly complex to an unbiased eye. And even if all complex systems arose from the same basic principles, the principles themselves might be of limited utility. (The notion of forest may be useful to understand plant epidemics but is of limited utility to treat a specific tree disease). Complex systems have their specific properties that deserve specialized scientific effort, and for such different corpora different strategies need to be used in order to advance the operative knowledge in each particular field. For complex systems, to learn from their differences is an even more exciting challenge than to find a grand unified theory.

To reach a better understanding of each field and to transfer techniques and insights from one discipline to the others requires the development of a common language and an open forum engaging a broad community to perform a strong cross-fertilization of research. This is why, in this project, mathematicians, physicists, biologists, medical doctors, economists, sociologists, linguists and engineers have attempted to share their experiences, models and the many questions on what role the notion of complexity may play in their fields.

By joint collaborative research and the participation in the seminars and workshops, phenomena and paradigms of fields other than their own became familiar to the scientists that shared the experiences of this project. Simple is what becomes familiar and, by broadening their knowledge on how complexity unfolds in its many forms, they might be able, in the future, to approach the study of complexity through simplicity.

One of the principal objects of theoretical
research in any department of knowledge is to
find the point of view from which the subject
appears in its greatest simplicity.
J. W. Gibbs

Hot spikes
After several preparatory symposia, starting in the April 1998 Madeira Workshop, the Conference "The Sciences of Complexity", held at ZiF, October 6-12, 2000, marked the opening of the Research Year (October 2000 - August 2001). Rather than focus on a particular field it was decided, in conformity with the spirit of the project, to bring together participants from a large spectrum of THE SCIENCES of (with?) complexity. The reader may have a look at this link of the conference, to obtain an insight on how this claim was achieved.

There were special moments when the concentration of participants with specific expertise steered the main stream of the project.

Dynamical models with learning capabilities, for drives and industrial control systems, was at the center of the program during October and November around a group of engineers.

During February a group of experimental physicists carried out, at ZiF, an experiment on shear flow of liquid crystals. This situation, where we can see the coexistence of turbulent and laminar domains, nowadays denoted "space-time intermittency", was at the center of an effort of both experimentalists and theoreticians for a deeper understanding of control and synchronization in extended systems.

Following a first Symposium in October 2000 the dynamics of the immune system was central in March. Here biologists together with computer scientists and physicists focused their interaction on the interplay of cellular and humoral immune systems during the response to viruses, on modeling of vaccination and the isotypic shift.

After a Symposium in July 2000, around May 2001 much attention was paid to the economic development of regions and countries by economists involved on regional and national innovation systems and on mastering complexity in the environment. Information flows may be the central notion to follow, when modeling such systems.

In June, the ability of neural networks (large assemblies of interacting non-linear units) to adapt and perform artificial learning in complex tasks was the topic. Here mathematicians, physicists and engineers collaborated in exploring new supervised and unsupervised learning techniques.

Following a Workshop in February 2000, and always in collaboration with the Tycho Brahe Project, linguists assembled in July 2001. Here two important and long-standing linguistic questions were addressed: how does language change proceed in time, and what triggers syntactic change. Continuing the well established tradition of the Tycho Brahe Project, linguists worked closely with mathematicians, physicists and computer scientists. Such collaboration of the disciplines was one of the hallmarks of the ZiF complexity project.

The newly emerging school of economics called "evolutionary economics" challenges most mainstream approaches in the field by insisting that most, if not all economic phenomena have to be viewed as non-linear out-of-equilibrium dynamics. The working group on economics and learning systems addressed these questions.

In June-July and also at several times during the year probabilistic theories of neural representation, neural networks with higher-order interactions and quantum control were focal points of research at ZiF.

In July and August computational non-linear dynamics and extended systems close to threshold were actively pursued.

To study the dynamics of the evolution at molecular resolution, one must master complexity with incomplete information. The subject was a continually progressing hot spike.

The notion of Self-Organized Criticality has become a new paradigm when studying different phenonema in natural and social sciences. Active research in this area took place throughout the whole year.

Finally, a special week on "Mathematics and Finance" highlighted the lively interactions between these two fields.

Modelling, Learning and Control of Electrical Drives with Automata Network

No controlo de accionamentos electromecânicos de velocidade variável têm especial importância os sistemas com capacidades de adaptação e aprendizagem. Tais sistemas, nomeadamente os baseados na máquina assíncrona de rotor em gaiola, de grande utilização industrial, pela dificuldade de medição directa das variáveis rotóricas representam um difícil problema de modelização quando se pretendem obter altos índices de precisão e desempenho.

A modelização clássica, recorrendo a equações do campo electromagnético, termodinâmica e mecânica, revela-se insuficiente, imprecisa e deste modo inadequada à obtenção de um desempenho eficaz. As relações fortemente não lineares entre as várias variáveis presentes, a dificuldade em obter algumas dessas variáveis, a variação de parâmetros em função das condições de funcionamento, a difícil modelização de certos fenómenos térmicos, mecânicos e electromagnéticos, originam modelos que, apesar de constituírem uma boa representação do accionamento electromecânico apresentam insuficiências que se podem revelar importantes em aplicações de elevada precisão.

Alguns estudos foram já efectuados utilizando-se as capacidades de aprendizagem de algoritmos linguísticos (lógica fuzzy) e/ou conexionistas (redes neuronais) na construção do modelo do sistema.

Subsistem, no entanto, dificuldades na representação de situações em que as sequências temporais são determinantes para a dinâmica.

Neste sentido propõe-se a exploração de redes de autómatos para a modelização e controlo de accionamentos electromecânicos de velocidade variável. As suas capacidades de aprendizagem e representação de relações funcionais serão analisadas e verificadas experimentalmente.

Prevê-se a aplicação desta metodologia ao controlo por orientação de campo, controlo adaptativo e controlo óptimo dos referidos sistemas de accionamento electromecânico.

Os resultados pretendidos revestem-se de importância assinalável para os processos industriais, pelo que o estudo teórico desenvolvido será acompanhado de uma necessária validação experimental, resultante da implementação de sistemas laboratoriais.

New Topologies for Power Converters with 'Intelligent' Control

SRM Machine - Global Models Development and Performance Analysis as Motor and Generator

Determinação de padrões de consumo em subestações de distribuição de energia eléctrica

RenH2 – Stand-Alone Energy System Supported by Totally Renewable Hydrogen Production

Rural and remote sites electrification is nowadays an important market for renewable energy based electrical production systems.

Autonomous electrical production systems, based on renewable energies are the most competitive economical option, when compared with solutions based only on diesel generators. However in order to supply 24 hours a day, one must consider a diesel generators to fulfil the energetic needs unsatisfied by the other system’s components: photovoltaic cells and wind turbines. From the environmental point of view this is not the ideal solution, since one of the system’s components uses a non renewable energy source – diesel.

There has been a significant investment increase, both international and national, within the hydrogen technology as the future energy vector. Its use as energy storage potency the role of renewable energies within the energetic sector. The hydrogen provides a effective way of energy storage solving the renewable energy fluctuation problem, which is a problem within autonomous systems.

The main objective of this project is the development of a fully autonomous system, where every component is based upon renewable energies. In this way we propose the development of an autonomous energy production system based on the hydrogen technology. Correlated objectives of the project are the optimization and integration of production modules, storage and energy conversion within fuel cells, using photovoltaic cells and wind turbines to achieve the production of hydrogen.

The obtained system is then a suitable choice regarding the actual stand-alone systems, based upon diesel generators and lead acid batteries energy storage. It meets the sustainability and environmental respect criteria regarding the energetic solutions of the future – zero emitting either on production or consumption.

Conceived to meet the energetic needs of a rural facility, typically 1kWp, this project considers the energy storage within hydrogen fuel cells and the connection to several energy sources, which may be placed where the receptive resource is by far more abundant (Annexed diagram). A future connection to the power grid could also be considered as an extension of this project.