In Distributed industrial control systems it is necessary to guarantee certain reliability level. In this sense, Checkpointing and Rollback techniques offer interesting possibilities to achieve fault tolerance without appreciable cost and complexity increment. Several Checkpointing techniques have been proposed. Most of them suppose the presence of stable storage in the system. But distributed industrial control systems usually do not dispose of this kind of storage. So, another storage strategy has to be employed. If Checkpoints were locally stored (Simple Checkpointing), the system tolerates only transient faults. If Checkpoints were locally, at the same node, and, additionally, at another/s node/s of the system stored (Two-level Checkpointing), the system can recover from some permanent faults too. In this article the results of a study of the reliability of these two different Checkpoint storage strategies were presented in order to evaluate if the reliability increase of the Two-level method justifies its greater complexity. In order to accomplish this study, two distributed industrial control systems were presented. Each of them are based on a different node architecture which will have an important effect upon the results of the study.
This article presents a method of modeling based on a data set by means of the GMDH algorithms. The data set contains information about fresh students of the Faculty of Electrical Engineering, Czech Technical University of Prague, Czech Republic (CTU, FEE). The motivation for our investigation was to discover, whether and how a particular student will be successful in his/her university studies. The data set was mined from official study application forms. They reflect student's personality, type and results (grades) from high school and entrance examination. These results serve as input vectors for prediction of his/her study results after the 1st semester. The results produced by the model were then compared with the real ones. The second motivation for our investigation was to find the significance of a particular input vector component, because it enables us to identify possible weak points of a student. As appropriate tools the GMDH neural networks (both linear and nonlinear) have been used.
This article reports a new approach to rule extraction method by using Group Method of Data Handling (GMDH) Algorithm in Data Mining area. The advantages of this method are (1) it accepts both categorical and continuous data at the same time, and (2) rules can be extracted easily from the generated model.We applied GMDH Algorithm to categorical data set of US congress voting records to extract rules. The correction rate of GMDH rules was 97.3% -- higher than Tsukimoto's method of rule extraction from Back-propagation neural network (81.0%). It was also higher than 97.0% of C4.5.
Three self-organizing data mining technologies that employ complementary descriptive languages - parametric regression models (GMDH neural networks), fuzzy rules (self-organizing fuzzy rule induction), and similarity models (analog complexing based clustering and classification) - are applied to generate diagnosis models of different levels of heart disease. The classification results show an accuracy of over 95% in average. Due to the strong knowledge extraction capabilities of the used technologies a nucleus of 4 most relevant variables is identified. The obtained results both classification accuracy and identified nucleus are also important for diagnosis cost reduction considerations.
Prof. Roberto Moreno-Diaz is a very well-known scholar of W.S. McCulloch in the field of neural net theory. We invited Prof. Moreno-Diaz as Chairman of the 7th ERCIM workshop Environmental Modelling (27–29 Sept 2000, Las Palmas de Gran Canaria, Spain), to edit a special issue. He graduated in Physics in 1962 and read his Doctoral thesis at the University of Madrid (now the Complutense), on logical neural nets and electronic models of neurons and neural nets. He has been a Full Professor in electro-magnetism, computer science and artificial intelligence since 1968. From 1962 to 1965 he was Assistant Professor of Industrial Physics at the University of Madrid. From 1965 to 1968 he was a member of staff at the Charles Stark Draper Laboratory at the Technological Institute of Massachussetts, and later worked as a consultant for the same. There he worked on natural and artificial visual processes and their architectures and on neural net theory, under the supervision of W.S. McCulloch. From 1969 to 1979 he was director of the Department of Electricity and Electronics at the University of Zaragoza, where he set up a research group in neural nets, vision and computation. In 1979, he returned to Las Palmas where he found various research groups in neural nets, natural and artificial perception, systems, neurocybernetics and robotic vision, currently existing in the University of Las Palmas de Gran Canaria. Moreno-Dı́az is author and coauthor of over one hundred papers on neurocybernetics, retinal theory and natural and artificial visual models. He has been guest lecturer at many national, European and American Universities. He has organised ten International Conventions on Computer sciences, Computer Aided Systems Theory and Neurocybernetics and is coeditor of 12 volumes on these subjects (publishers: Alianza Editorial, Springer-Verlag, Hemisphere and the MIT Press). He is Academician Correspondent for the Royal Academy of Exact, Physical and Natural Sciences of Madrid since 1981 and Founder Member and Vice-president of the Canary Academy of Sciences since 1986. He was awarded the Canary Research Prize in 1985. He is presently director of the Instituto Universitario de Ciencias y Tecnologı́as Cibernéticas, Universidad de Las Palmas de Gran Canaria.
The traditional assignment approaches for elevator group systems, which are simple and rough, are only fit for low-rise buildings. Hence, they are not suitable in the high-rise buildings with complex traffic requests. In this article, a so-called Score Index Assignment (SIA) approach is proposed to design the real-time scheduling for elevator group systems. Simulation results are given to show that the SIA approach can improve the total service performance of the elevator group system subjected to uncertainties. The SIA approach can fit in any different traffic request conditions by tuning the control (assignment) parameters in the so-called total service evaluation index function of elevator.
Robot manipulators are built to meet certain predetermined performance requirements. The question of whether the robot will have the desired functionality (e.g. dexterity, accuracy, reliability, speed, etc.) needs to be answered before the robot is actually built.We have developed a software package that can greatly ease the design of a generic 6-DOF manipulator with a spherical wrist. Our package will accept as input the configuration of a generic robot in D-H parameter form and the robot dynamics parameters and produce a variety of closed form solutions that are essential to the robot designer. The package can also be used as a simulation tool that can tell the designer whether the manipulator meets the desired functionality. It will also optimize several control and structure parameters for the generic manipulator based on simulated task descriptions.
In this paper the process of gearing information supply to information demand in river basin management is discussed. The focus is on the design of information systems for river basin management, interdisciplinary modeling and a discussion on the large-scale patterns which emerge if micro scale models and macro scale models are coupled. The theoretical concepts are illustrated with a number of case studies.
In this article, the usability of genetic algorithms for signal approximation is discussed. Due to recent developments in the field of signal approximation by wavelets, this work concentrates on signal approximation by wavelet-like functions. Signals are approximated by a finite linear combination of elementary functions and a genetic algorithm is employed to find the coefficients to such an approximation. The algorithm maintains a population of different approximations, encoded in the form of 'chromosomes'. From this population 'parents' are selected according to their 'fitness', and the 'children' that constitute the next generation are produced from these parents using mutation and crossover operators.Fitness functions employed to evaluate different approximations are the L1, L2, L4, and L∞ norms. Experiments are carried out on several test signals, using Gabor and spline wavelets, both to evaluate the quality of different fitness functions, encoding schemes, and operators, and to assess the usefulness of genetic algorithms in the realm of signal approximation.Although other existing methods are faster while providing comparable approximation quality, the algorithm offers a great deal of flexibility in terms of different elementary functions, fitness criteria, etc.
This paper is concerned with recursive estimation, testing and forecasting of the volatility of daily returns in Standard and Poor's 500 Composite Index in the presence of outliers, or significant spikes in the volatility of daily returns, and model misspecification. The empirical analysis increases the sample size up to 12000 observations recursively to examine the effects of outliers and misspecification on: (i) the parameter estimates of the ARCH(1) and GARCH (1,1) process; (ii) their associated asymptotic and robust t-ratios; (iii) the second and fourth moment conditions for stationarity, consistency and asymptotic normality; and (iv) the forecast performance for periods with significant spikes in volatility and for periods of relative calm.
Models of visual processing in living systems can be successfully developed for the front end sensorial part, which in vertebrates correspond to the retina. The appropriate modelling tools are the tools of classical systems theory. In this line, we present first analytical models for the frog's retina and for higher vertebrate retinae, leading to the generalized structure of receptive fields processing for modelling linear and non-linear behaviour. Next, transformations on input data receptive fields are presented, including the novel concept of micro-structures of retinal receptive fields in form of Newton Filters, the concepts of transformation on input and output spaces and the concepts of transformation on receptive fields by means of partitions.
Due to exterior dynamical constraints, affecting the manipulator system, some deflections from the theoretical movement trajectory arise for the holder. They depend on the manipulator arms configuration. Analysis of the system vibrations occurring near the position of the static balance enables us to determine the size of the area, in which the manipulator holder can really be found. This information is critical for applying the proper control system and for eliminating the most undesirable conditions for the manipulator in action. When a manipulator is constructed, its dynamical characteristics of the manipulator are needed already at the design stage. Performing adequate simulations is necessary to change the geometrical or material parameters of the links to achieve the demanded level of accuracy if the manipulator positioning in certain working conditions.
Complex real world systems are currently developing to become a decisive instrument for IT-supported problem solving of a great number of problems posed by science, economy and society. By this, we for the first time face the chance of being able to find new solutions, visions, and strategies for sustainable development on the basis of complex, real world simulations. M3-Simulations is a simulation approach that actively involves man in a simulated world based on scientifically-founded simulation models driven by current data of measurement. This new kind of simulation concept combines reality-based modeling of scientific simulation systems with the intuitive graphic representation, complex communication and interaction structures of virtual worlds in order to support the process of investigation into the possibilities of sustainable development.
In this paper, we address the problems of guaranteed cost control for a class of large-scale discrete-time systems with parameter uncertainties. The parameter uncertainties are real time-varying norm-bounded. Using the Lyapunov method, a robust decentralized state-feedback controller is constructed to render the closed-loop system robustly stable while guaranteeing a prescribed level of performance. The developed result is expressed in terms of linear matrix inequalities. A numerical example is included to illustrate the design procedure.
A summary of results of extensive theoretical and numerical investigations on the dynamics and control of an inverted pendulum with rigid unilateral constraints is presented. The system is subjected to optimal excitations which permit to reduce the theoretically chaotic region in the parameter plane, and to suitably modify the nonlinear dynamics. Several bifurcations, both classical and nonclassical, determine the transition between different dynamical regimes. Attention is devoted to local and global bifurcations, and to the analysis of phenomena which are directly related to the nonsmooth nature of the system. The performances of the optimal excitations are numerically evaluated by comparison with the reference case of harmonic force, and it is shown how it is possible to improve some technical requirements of the dynamics through proper implementations of the optimal excitations.
In real-time applications a discrete supervisory controller should stabilise the fuzzy control system. For this reason a discrete supervisory controller was developed and applied together with the fuzzy controller. The discrete supervisory controller performs like a continuous one. The main difference is in the sign function of the appended controller. We have proved that the fuzzy control system equipped with the discrete supervisory controller is globally stable only if we can predict the sign of supervisory action. On the other hand, discrete supervisory controllers require the knowledge of the bounds of the non-linear functions in the system, same as continuous one. Application to a balance control system was successful, both in simulations and in real-time control.
We describe in this paper a new method for adaptive model-based control of non-linear dynamic plants using Neural Networks, Fuzzy Logic and Fractal Theory. The new neuro-fuzzy-fractal method combines Soft Computing (SC) techniques with the concept of the fractal dimension for the domain of Non-Linear Dynamic Plant Control. The new method for adaptive model-based control has been implemented as a computer program to show that our neuro-fuzzy-fractal approach is a good alternative for controlling non-linear dynamic plants. We illustrate in this paper our new methodology with the case of controlling biochemical reactors in the food industry. For this case, we use mathematical models for the simulation of bacteria growth for several types of food. The goal of constructing these models is to capture the dynamics of bacteria population in food, so as to have a way of controlling this dynamics for industrial purposes.
Recent advances in computer, robotics and Internet technologies have enhanced the importance of digital signal processing tools, parallel processing and multi-processing research. The recent developments in microelectronics and fiber optics technologies have called for faster digital signal processors for data compression, coding, secure communication, and multimedia applications. Man-machine interaction, the adaptive control of fast moving objects, spaceships and increasingly intelligent and complex robotics have motivated research in faster computers and massively parallel processors. Adaptive on-line system identification of fast time varying systems is a problem that has received greater attention in recent years. Generalized spectral analysis transforms and factorizations thereof lead to highly efficient tools for parameter identification in the presence of noise and highly efficient coding techniques. This chapter presents novel spectral analysis definitions and general base parallel processors for the implementation of DSP algorithms, and modern generalized spectral analysis transforms destined for such high-speed applications. The Weighted Laplace and z-transform generalized spectra are shown to unmask the poles and zeros of finite duration continuous time and discrete time systems. The Generalized Chrestenson-Walsh transform is shown to be a generalization of the Fourier Transform and as such provides a wider range of coding and multimedia compression possibilities. Massive parallelism is implemented using general base hypercube transformations.
The self-organizing Kohonen map is a reliable and efficient way to achieve vector quantization. Typical application of such algorithm is image compression. Moreover, Kohonen networks realize a mapping between an input and an output space that preserves topology. This feature can be used to build new compression schemes which allow to obtain better compression rate than with classical method as JPEG without reducing the image quality. Compared to JPEG, our lossy compression scheme shows better performances (in terms of PSNR) for compression rates higher than 30 [C. Amerijckx, M. Verleysen, P. Thissen and J.-D. Legat (1998). Image compression by self-organized Kohonen map. IEEE Transactions on Neural Networks, 9(3), 503-507.]. For lossless compression, this rate is about 2.7 for standard images.
