ESTOCASTIC THEORY AND TIME SERIES ANALYSIS

Author: BOSCH ROIG IGNACIO.
Year: 2005.
University: POLITÉCNICA DE VALENCIA [www.upv.es].
Place of defense: COMUNICACIONES.
Place of preparation: UNIVERSIDAD POLITÉCNICA DE VALENCIA.

Summary: This thesis deals with the development of detection algorithms in distributed systems monosensor. Traditionally, distributed detection has been applied to the merger of decisions taken by sensors spatially dispersed. However, it can be useful in implementing detectors seeking to exploit different kinds of information about the differences between the two scenarios on which to decide. The aim is to replace a single detector complex, if not intractable, for a combination of detectors simpler, implementable from the theory of optimal detection. Faced with the possibility of a comprehensive approach has been preferred to focus on two theses applications of particular interest to the research group, in which seono has developed this thesis: detection of forest fires from infrared echo signals in noise granular background, ultrasonic non-destructive testing. With regard to the detection of forest fires has proposed a scheme detection merging two types of detectors that we call detector detector persistence and growth. The underlying idea is to try to exploit the same information used by the same vigilant human, which it is replacing. The detector is based persistence to turn in the merger decision for small intervals consecutive observation. At each interval set up a filter detector adapted subspace which seeks to exploit the behavior of persistence in the short time period of a fire, compared to other effects of impulsive nature which may also produce sporadic increases in temperature in the cell under surveillance ( cars, sun ,..). For its part detector seeks to exploit growth through a filter adapted to a linear growth rate, the expected growth in temperatures, in the medium and long term, should produce an uncontrolled fire, as compared with controlled fires that could shoot unduly system. The schema design allows you to adjust the probability of false alarm, getting good benefits probability of detection, as we have shown signs simulated and real. In connection with the detection of ultrasonic echoes in the background noise granular we have developed a scheme that combines the philosophy behind the so-called technical "split-spectrum" with the detection distributed. Techniques "split-spectrum" exploit the diversity frequency implicitly comparing the responses to different frequencies of an echo signal, as compared to the responses of noise granular. The first should be much more stable and the latter more erratic. The proposed scheme consists of placing a detector "tuned" to each frequency band and then merge all detections. The tune in each detector is done through a filter adapted to a subspace "paso-banda." We have developed all the analysis required to adjust the desired probability of false alarm, assuring maximizing the probability of detection. The experiments simulated signals has been completed laboratory experiments with pieces of aluminum. We also have applied the techniques proposed in determining the profiles of interfaces between layers corresponding to the dome of the Basilica of the Virgin of Desamparados Valencia, as supported in their restoration.

Author: MONDEJAR JIMENEZ JOSE.
Year: 2005.
University: CASTILLA-LA MANCHA [www.uclm.es].
Place of defense: FACULTAD DE CIENCIAS SOCIALES DE CUENCA.
Place of preparation: FACULTAD DE CIENCIAS SOCIALES DE CUENCA.

Summary: The measurement of economic development along with the definition of the determinants of growth are two of the most interesting issues in recent decades, why have focused interest among scientists in general, and economists in particular. Within this level of study, work focuses on the economic analysis of the economic magnitudes. The economic analysis was conducted with the primary objective of knowing what can happen in the economy, once analyzed their situation and predictable evolution. The prediction appears as a critical aspect, while the most complex of that integrate analysis of the situation. It is also of particular interest to the application of it to the regional level, a factor that hinders its achievement because the information available for the case of European regions is at a level lower than the breakdown of national economies . The practical implementation, focuses on the creation of specific economic indicators for the region of Castilla-La Mancha. This paper comprehensively develop statistical methods that were of particular interest in the current economic world, with special emphasis on their application to the field of quantifying economic magnitudes, which always generate controversy for a high measurement. In addition to the development and implementation of statistical methods presently available for the economic analysis, there are important contributions in the use of models in state space. Although formally equivalent to the known ARIMA models, treatment time series in state space and computational structural advantages that make it particularly suitable for short-term analysis. Therefore, there are increasingly recent developments and applications that make it a clear alternative to the Box-Jenkins approach.

Author: MONDEJAR JIMENEZ JOSE.
Year: 2005.
University: CASTILLA-LA MANCHA [www.uclm.es].
Place of defense: FACULTAD DE CIENCIAS SOCIALES DE CUENCA.
Place of preparation: FACULTAD DE CIENCIAS SOCIALES DE CUENCA.

Summary: The measurement of economic development along with the definition of the determinants of growth are two of the most interesting issues in recent decades, why have focused interest among scientists in general, and economists in particular. Within this level of study, work focuses on the economic analysis of the economic magnitudes. The economic analysis was conducted with the primary objective of knowing what can happen in the economy, once analyzed their situation and predictable evolution. The prediction appears as a critical aspect, while the most complex of that integrate analysis of the situation. It is also of particular interest to the application of it to the regional level, a factor that hinders its achievement because the information available for the case of European regions is at a level lower than the breakdown of national economies . The practical implementation, focuses on the creation of specific economic indicators for the region of Castilla-La Mancha. This paper comprehensively develop statistical methods that were of particular interest in the current economic world, with special emphasis on their application to the field of quantifying economic magnitudes, which always generate controversy for a high measurement. In addition to the development and implementation of statistical methods presently available for the economic analysis, there are important contributions in the use of models in state space. Although formally equivalent to the known ARIMA models, treatment time series in state space and computational structural advantages that make it particularly suitable for short-term analysis. Therefore, there are increasingly recent developments and applications that make it a clear alternative to the Box-Jenkins approach.