## Diploma Thesis Distributed Control of a Deformable System – Analysis of a Neocybernetic Framework

### BibTeX

@MISC{-ing_diplomathesis,

author = {Professor Dr. -ing and Dr. H. C. Michael Zeitz},

title = {Diploma Thesis Distributed Control of a Deformable System – Analysis of a Neocybernetic Framework},

year = {}

}

### OpenURL

### Abstract

The thesis at hand was written at the Control Engineering Laboratory of the Helsinki

### Citations

4286 |
Neural Networks: A Comprehensive Foundation
- Haykin
- 1999
(Show Context)
Citation Context ...nal grid is faster than the change of the input data u(l). 2.4.2 Modeling of a Hebbian Neuron Based on the linear Hebbian neuron model, complex structures could already be developed as can be read in =-=[6]-=- or [18], for example. On the other hand, models have become increasingly complex, what makes it difficult to find efficient methods for the analysis. Basis for the following mathematical consideratio... |

4196 | Artificial Intelligence: A Modern Approach - Russell, Norvig - 1994 |

283 |
Cybernetics: Or Control and Communication
- Wiener
- 1949
(Show Context)
Citation Context ...le of Symbols 125 References 139Chapter 1 Introduction It was Norbert Wiener who brought up the term cybernetics when his book Cybernetics: Or Control and Communication in the Animal and the Machine =-=[25]-=- was published in 1948. The general idea of cybernetic systems is that observed complex behavior in systems is caused by the dynamics of present actors. These dynamics are a result of interactions and... |

233 | Wireless Sensor and Actor Networks: Research Challenges - Akyildiz, Kasimoglu |

217 | Concepts and Applications of Finite Element Analysis - Cook, Malkus, et al. - 1989 |

121 |
Principal Component Neural Networks: Theory and Applications
- Diamantaras, Kung
- 1996
(Show Context)
Citation Context ...d is faster than the change of the input data u(l). 2.4.2 Modeling of a Hebbian Neuron Based on the linear Hebbian neuron model, complex structures could already be developed as can be read in [6] or =-=[18]-=-, for example. On the other hand, models have become increasingly complex, what makes it difficult to find efficient methods for the analysis. Basis for the following mathematical considerations are n... |

47 |
The organisation of behaviour: a neuropsychological theory
- Hebb
- 1949
(Show Context)
Citation Context ...ind efficient methods for the analysis. Basis for the following mathematical considerations are neurophysiological observations that can be ascribed to the physician Donald O. Hebb half a century ago =-=[7]-=-. The Hebbian law can be formulated as follows: • Hebbian law. Synaptic connection between the neuron and an incoming signal becomes stronger if the signal and the current neuron activity correlate wi... |

11 |
Multivariate Regression − Techniques and Tools
- Hyötyniemi
(Show Context)
Citation Context ...hese data vectors the modeling machinery tries to construct appropriate connections among the available data segments. As the data normally turns out to be redundant, one can use multivariate methods =-=[9]-=- and corresponding mathematical tools for a better analysis of neocybernetic models [15]. This bottom-up analysis of data can be carried out by Hebbian neurons, for example, and is presented later on ... |

4 | Neocybernetics in Biological Systems
- Hyötyniemi
- 2006
(Show Context)
Citation Context ...ramework for the combination of control theory, information theory and communication theory and as an interdisciplinary research area it can be applied to different domains like biology or technology =-=[15]-=-[14]. Until now, traditional control purposes are mainly implemented in a centralized manner. Hence, the total potential of cybernetic thinking in control theory is not exploited yet, if one thinks of... |

3 |
Nervensystem. In Allgemeine und spezielle Pathologie
- Schlote, Riede, et al.
- 1993
(Show Context)
Citation Context ...layers, soles of feet, ...), muscles of human beings or animals are getting better evolved, if they are used more intense and so on. In biology this phenomenon is known as hyperplasia and hypertrophy =-=[23]-=-. Even if the ongoing process in the case of the steel plate adaptation is active, the emergent result can be implemented passively, by adding extra layers at these positions, where the actuators evol... |

2 |
Emergent Coordination in Distributed Sensor Networks
- Hyötyniemi
(Show Context)
Citation Context ...eded. A system theoretic approach reveals that Hebbian/anti-Hebbian learning results in emergent system behavior, what can, for example, be used for smart data analysis in distributed sensor networks =-=[11]-=-. Even stronger views can be derived from the new neocybernetic approach to "elastic systems" [15]. These systems are determined by strong interconnections with their regarded environment caused by ma... |

2 |
Hebbian Neuron Grids: System Theoretic Approach
- Hyötyniemi
- 2004
(Show Context)
Citation Context ...his bottom-up analysis of data can be carried out by Hebbian neurons, for example, and is presented later on in Section 2.4 in order to get a better understanding of neocybernetic modeling approaches =-=[12]-=-.Chapter 2. About Neocybernetics 7 To achieve a consistent mathematical framework that is easy enough to provide a simple analysis of complex systems and still captures the essence of the considered ... |

1 |
Femlab 3.1 Modeling Guide. Femlab AB
- AB
- 2004
(Show Context)
Citation Context ...hat the shortcomings of the conventional simulative control algorithm are addressed. 5.1.1 Interconnection of Matlab and Femlab In order to simulate mechanical deformable systems, the tool Femlab 3.1 =-=[1]-=-, [2] is used for the numerical simulation of deformed systems, where different forces (external and actuator) are applied. Femlab is a multiphysics modeling tool, where physical processes are describ... |

1 |
Femlab 3.1 User’s Guide. Femlab AB
- AB
- 2004
(Show Context)
Citation Context ...he shortcomings of the conventional simulative control algorithm are addressed. 5.1.1 Interconnection of Matlab and Femlab In order to simulate mechanical deformable systems, the tool Femlab 3.1 [1], =-=[2]-=- is used for the numerical simulation of deformed systems, where different forces (external and actuator) are applied. Femlab is a multiphysics modeling tool, where physical processes are described wi... |

1 |
Konvergenz einer Vektor-Folge
- Boßle, Streit
- 2006
(Show Context)
Citation Context ...ngle transformed sensor displacements s o i,∗(l) are only dependent on their own predecessor as matrix Λ(l) is diagonal. The overall process converges now, if all single sequences so i,∗ (l) converge =-=[3]-=-. For a single sequence i it follows from (5.13) with (5.14) s o i,∗ (l) = ˜si,∗(l) + λi(l)s o−1 i,∗ (l) (5.15) and with respect to the initial value, the sequence can be written as a geometric progre... |

1 |
Le Chatelier’s Principle
- Clark
- 2002
(Show Context)
Citation Context ... to external tensions deformations take place, but if the external pressure is released the system returns to its original state. In chemistry for example that manner is called Le Chatelier Principle =-=[4]-=- where the dynamic equilibrium of chemical reactions moves in order to counteract external changes. Shortly it can be said that every neocybernetic system is identical with elastic systems, characteri... |

1 |
Presentation: Towards "Cybernetic Control
- Hyötyniemi
(Show Context)
Citation Context ...tivariate distributed controllers are adapting data-based and learning fromChapter 6. Simulation: Results and Interpretations 95 different variations in the observable environment completely locally =-=[8]-=-. Anyhow, the results of this local adaptation processes show emergent behavior in the system. In the case of the steel plate adaptation processes the coupling parameter q determines directly the expe... |

1 |
Cybernetics: Towards a Unified Theory
- Hyötyniemi
- 2004
(Show Context)
Citation Context ...n. Furthermore, the idea of stable systems can be relaxed. Systems can also include oscillating parts, as long as it keeps its integrity and statistical terms can be used to describe the system [15], =-=[10]-=-. 2.3.2 High dimensionality As the real structure of the environment can not be assumed to be known, one has to take measurements of the environmental responses into account in order to build an appro... |

1 | Information and Entropy in Cybernetic Systems - Hyötyniemi - 2004 |

1 |
it is here, it is hot
- Cybernetics
- 2006
(Show Context)
Citation Context ...work for the combination of control theory, information theory and communication theory and as an interdisciplinary research area it can be applied to different domains like biology or technology [15]=-=[14]-=-. Until now, traditional control purposes are mainly implemented in a centralized manner. Hence, the total potential of cybernetic thinking in control theory is not exploited yet, if one thinks of dis... |

1 |
Höhere Mathematik 1
- Meyberg, Vachenauer
- 1999
(Show Context)
Citation Context ...ry geometric progression converges. In case of the geometric progression (5.16), this holds when |λi(l)| < 1 ∀ i = 1, ..., m (5.17)Chapter 5. Derivation of a New Simulation Algorithm 60 is fulfilled =-=[17]-=-. In other terms, one can state out that the first order iteration is stable as long, as the magnitudes of the eigenvalues of the matrix Z(l) are smaller than one |eig{Z(l)}|i < 1 ∀ i = 1, ..., m (5.1... |

1 |
Method of Steepest Descent
- Weisstein
(Show Context)
Citation Context ... expressed as The closed form solution for x is dJ dx (x) = ϕT W ϕx − ϕ T W u (2.22) ¯x = ( ϕ T W ϕ ) −1 ϕ T W u (2.23) If one wants to search the minimum iteratively with the steepest descent method =-=[24]-=- the formulation can be presented as x(κ + 1) = x(κ) − β dJ dx (x(κ)) = x(κ) − βϕ T W ϕx(κ) + βϕ T W u (2.24) If one compares this equation to (2.4) one can see a certain similarity, if the different ... |

1 |
Die freie Enzyklopädie. Minimum Total Potential Energy Principle. 2006. http://en.wikipedia.org/wiki/Minimum_total_potential_energy_ principle
- Wikipedia
(Show Context)
Citation Context ...efinite. These conditions are fulfilled if (3.8) holds. With these assumptions, (3.5) becomes the difference between the stored potential energies in the system. Principle of Minimum Potential Energy =-=[27]-=- defines that a pressured structure finds an equilibrium that minimizes this criterion as it tries to exhaust the external forces with minimum change of internal deformations. Any balanced neocybernet... |

1 |
The free Encyclopedia. Diagonalmatrix
- Wikipedia
- 2006
(Show Context)
Citation Context ...lgorithm 59 on principal axes with the transformation s(l) = Ds∗(l) (5.12) Matrix Z(l) is diagonalizable, if all algebraic and geometric multiplicities of the eigenvalues of the matrix Z(l) are equal =-=[28]-=-. The transformation matrix D from (5.12) is the matrix of eigenvectors of Z(l) and the diagonalized matrix Λ(l) contains the eigenvalues of Z(l) respectively. (5.11) delivers with (5.12) the transfor... |

1 |
The free Encyclopedia. Trace (linear algebra). 2006. http://en. wikipedia.org/wiki/Matrix_trace
- Wikipedia
(Show Context)
Citation Context ...he matrix trace is the sum of the diagonal elements and simultaneously it is the sum of the matrix eigenvalues. Additionally, matrices within trace can be rotated if they are appropriately compatible =-=[29]-=-. This result means that the completely adapted system maximizes the sum of the n most significant eigenvalue squares as seen from within the system. If the criterion from (3.5) is used, the optimum r... |