Este trabalho de pesquisa objetivou a obtenção de uma nova política de operação que melhor caracterizasse o comportamento ótimo dos sistemas hidrelétricos de potência, mesmo diante das mais variadas condições hidrológicas. Este trabalho teve duas linhas de investigação. Uma tratou do problema de previsão de vazões afluentes mensais, na busca por abordagens e técnicas que definissem bons modelos de previsão. A outra linha de pesquisa tratou de encontrar uma nova política de operação, para o problema de planejamento da operação, que fosse capaz de definir uma seqüência de decisões operativas mais estáveis, confiáveis e de menor custo operativo. Na primeira linha de pesquisa, investigou-se três aspectos importantes na definição de um modelo de previsão: técnicas de pré-processamento dos dados, definição automática do espaço de entrada e avaliação do desempenho de alguns modelos de redes neurais e sistemas Fuzzy como modelos de previsão. Nestes aspectos foram investigadas a utilização da análise dos componentes principais e o tratamento da série temporal de vazões afluentes como um sinal discreto, utilizando-se a representação \"analytic signal\". Para a definição do espaço de entrada de maneira automática utilizou-se a abordagem da \"dynamic modelling\", empregando-se a \"average mutual information\" e \"false nearest neighbors\". Para implementação dos modelos de previsão foram estudados e avaliados quatro modelos inteligentes: rede SONARX, rede SONARX-RBF, modelo ANFIS e a rede ESN. Já na outra linha de pesquisa, foi proposta uma política de operação que fosse capaz de estabilizar os despachos de geração termelétrica e conseqüentemente o custo marginal de operação. A política de operação preditiva estabilizada via termo inercial produziu excelentes resultados operativos, melhorando de forma significativa a performance da política preditiva. / This research work aimed at obtaining a new operation policy which could better describe the optimal behavior of hydropower systems, even when faced with the most varied hydrological conditions. This research had two lines of investigation. The first one dealt with the monthly water inflow forecasting problem, searching for approaches and techniques which could define efficient forecasting models. Three important aspects to define a forecasting model were investigated: data pre-processing techniques, automatic definition of the embedding and the performance assessment of some artificial neural networks and Fuzzy systems. Hence, the use of the principal components analysis was investigated and, considering the water inflow time series as a discrete signal, the analytic signal representation could be used to preprocess the data. Furthermore, the embedding was automatically defined using the dynamic modelling approach, by using the average mutual information and the false nearest neighbors techniques. The forecasting models were implemented by four intelligent models: SONARX network, SONARX-RBF network, ANFIS model and the ESN network. The other line of investigation came up with a new operation policy to solve the operation planning problem, defining a more stable, reliable and less costly operative decision sequence. It was proposed an approach to stabilize the thermoelectric generation dispatches and, as a result, the operative marginal cost. The predictive operation policy stabilized via inertial term produced excellent operation results, improving the performance of the predictive policy.

Este trabalho de pesquisa objetivou a obtenção de uma nova política de operação que melhor caracterizasse o comportamento ótimo dos sistemas hidrelétricos de potência, mesmo diante das mais variadas condições hidrológicas. Este trabalho teve duas linhas de investigação. Uma tratou do problema de previsão de vazões afluentes mensais, na busca por abordagens e técnicas que definissem bons modelos de previsão. A outra linha de pesquisa tratou de encontrar uma nova política de operação, para o problema de planejamento da operação, que fosse capaz de definir uma seqüência de decisões operativas mais estáveis, confiáveis e de menor custo operativo. Na primeira linha de pesquisa, investigou-se três aspectos importantes na definição de um modelo de previsão: técnicas de pré-processamento dos dados, definição automática do espaço de entrada e avaliação do desempenho de alguns modelos de redes neurais e sistemas Fuzzy como modelos de previsão. Nestes aspectos foram investigadas a utilização da análise dos componentes principais e o tratamento da série temporal de vazões afluentes como um sinal discreto, utilizando-se a representação \"analytic signal\". Para a definição do espaço de entrada de maneira automática utilizou-se a abordagem da \"dynamic modelling\", empregando-se a \"average mutual information\" e \"false nearest neighbors\". Para implementação dos modelos de previsão foram estudados e avaliados quatro modelos inteligentes: rede SONARX, rede SONARX-RBF, modelo ANFIS e a rede ESN. Já na outra linha de pesquisa, foi proposta uma política de operação que fosse capaz de estabilizar os despachos de geração termelétrica e conseqüentemente o custo marginal de operação. A política de operação preditiva estabilizada via termo inercial produziu excelentes resultados operativos, melhorando de forma significativa a performance da política preditiva. / This research work aimed at obtaining a new operation policy which could better describe the optimal behavior of hydropower systems, even when faced with the most varied hydrological conditions. This research had two lines of investigation. The first one dealt with the monthly water inflow forecasting problem, searching for approaches and techniques which could define efficient forecasting models. Three important aspects to define a forecasting model were investigated: data pre-processing techniques, automatic definition of the embedding and the performance assessment of some artificial neural networks and Fuzzy systems. Hence, the use of the principal components analysis was investigated and, considering the water inflow time series as a discrete signal, the analytic signal representation could be used to preprocess the data. Furthermore, the embedding was automatically defined using the dynamic modelling approach, by using the average mutual information and the false nearest neighbors techniques. The forecasting models were implemented by four intelligent models: SONARX network, SONARX-RBF network, ANFIS model and the ESN network. The other line of investigation came up with a new operation policy to solve the operation planning problem, defining a more stable, reliable and less costly operative decision sequence. It was proposed an approach to stabilize the thermoelectric generation dispatches and, as a result, the operative marginal cost. The predictive operation policy stabilized via inertial term produced excellent operation results, improving the performance of the predictive policy.

During 2002-2006 geological field work were carried out in Mozambique. The purpose was to check the preliminary geological interpretations and also to resolve the problems that arose during the compilation of preliminary geological maps and collect samples for laboratory studies. In parallel, airborne geophysical data were collected in many parts of the country to support the geological interpretation and compilation of geophysical maps. In the present work the aeromagnetic data collected in 2004 and 2005 in two small areas northwest of Niassa province and another one in eastern part of Tete province is analysed using GeosoftTM. The processing of aeromagnetic data began with the removal of diurnal variations and corrections for IGRF model of the Earth in the data set. The study of the effect of height variations on recorded magnetic field, levelling and interpolation techniques were also studied. La Porte interpolation showed to be a good tool for interpolation of aeromagnetic data using measured horizontal gradient. Depth estimation techniques are also used to obtain semi-quantitative interpretation of geological bodies. It was showed that many features in the study areas are located at shallow depth (less than 500 m) and few geological features are located at depths greater than 1000 m. This interpretation could be used to draw conclusions about the geology or be incorporated into further investigations in these areas.

Magnetic survey is a common geophysical exploration technique. By measuring the magnetic field strength at specific area, the characteristics and physical meaning of the target can be obtained through the analysis of the Earth's magnetic field anomalies within a stratigraphic zone or archaeological sites. In recent years, the marine magnetometer is employed to conduct underwater archaeological expedition at surrounding waters of Taiwan for ancient shipwrecks researching. The purpose of this study is to understand the relationship between the magnetic anomalies with the magnetic object via the various signal processing methods, included the calculation horizontal and vertical derivatives using fast Fourier transform (FFT) to eliminate the regional magnetic influence and gain the anomalies characteristics of the target itself, as well as highlight the location and boundaries of the magnetic source through the analytical signal. In addition, the Euler deconvolution implements as a tool for magnetic source inversion. The theory of Euler deconvolution was first proposed by Thompson (1982), this method is able to detect the magnetic source and estimate its locations by choosing the suitable structural index. Hsu (2002) proposed the enhanced Euler deconvolution, which is a combined inversion for structural index and source location through the use of the vertical derivative of measured data.In this study, we first generate various anomalies as testing models which are correspond with different geometric shape of magnetic source, the position and structural index for model is inversed by enhanced Euler deconvolution in both 2D and 3D.Moreover, the experiment was planned at offshore of Dalinpu in Kaohsiung, we took the CPC's pipelines as investigation objects which were buried under the seabed, than compare with sub-bottom profiler data to assess the feasibility of this method for underwater exploring applications.The most estimated results in 2D are correspond to the theory, but it does not have significant results in 3D due to the lack of observed data for the whole surface.In general, this method is concise and fast, it is fit for interpreting the magnetic data for exploring the underwater object.

This project considers methods of construction QRS detectors. It focus in detection complexes of QRS single leads and space speed, which are calculated from three orthogonal leads. In theory was refer to various methods, which lead to design detector. It were designed two algoritms (constant and adaptive detecting threshold), which were implemented into detector and the signal was preprocessed by Hilbert transformation. Toward algoritms were completed by modification, which improved detection effectivity. Function of algoritms were tested in all signals of CSE (V2,V5,aVF).

Most signal processing methods were developed for continuous signals. Digital devices, such as the computer, process only discrete signals. This dissertation proposes new techniques to accurately define and efficiently implement an important signal processing method---the time--frequency distribution (TFD)---using discrete signals. The TFD represents a signal in the joint time--frequency domain. Because these distributions are a function of both time and frequency they, unlike traditional signal processing methods, can display frequency content that changes over time. TFDs have been used successfully in many signal processing applications as almost all real-world signals have time-varying frequency content. Although TFDs are well defined for continuous signals, defining and computing a TFD for discrete signals is problematic. This work overcomes these problems by making contributions to the definition, computation, and application of discrete TFDs. The first contribution is a new discrete definition of TFDs. A discrete TFD (DTFD) should be free from the sampling-related distortion known as aliasing and satisfy all the important mathematical properties that the continuous TFD satisfies. Many different DTFD definitions exist but none come close to attaining this ideal. I propose three new components which make up the DTFD: 1) a new discrete Wigner--Ville distribution (DWVD) definition which satisfies all properties, 2) a new discrete analytic signal which minimises aliasing in the DWVD, and 3) a new method to define and convolve the discrete kernel with the DWVD to produce the DTFD. The result: a DTFD definition that, relative to the existing definitions, better approximates the ideal DTFD. The second contribution is two sets of computationally efficient algorithms to compute the proposed DTFD. The first set of algorithms computes the DTFD exactly; the second set requires less memory than the first set by computing time- and, or frequency-decimated versions of the DTFD. Both sets of algorithms reduce the computational load by exploiting symmetries in the DTFD and by constructing kernel-specific algorithms for four different kernel types. The third, and final, contribution is a biomedical application for the proposed DTFD and algorithms. This application is to accurately detect seizure events in newborn electroencephalogram (EEG) signals. Existing detection methods do not perform well enough for use in a clinical setting. I propose a new method which is more robust than existing methods and show how using the proposed DTFD, comparative to an existing DTFD, improves detection performance for this method. In summary, this dissertation makes practical contributions to the area of time--frequency signal processing by proposing an improved DTFD definition, efficient DTFD algorithms, and an improved newborn EEG seizure detection method using DTFDs.

Most signal processing methods were developed for continuous signals. Digital devices, such as the computer, process only discrete signals. This dissertation proposes new techniques to accurately define and efficiently implement an important signal processing method---the time--frequency distribution (TFD)---using discrete signals. The TFD represents a signal in the joint time--frequency domain. Because these distributions are a function of both time and frequency they, unlike traditional signal processing methods, can display frequency content that changes over time. TFDs have been used successfully in many signal processing applications as almost all real-world signals have time-varying frequency content. Although TFDs are well defined for continuous signals, defining and computing a TFD for discrete signals is problematic. This work overcomes these problems by making contributions to the definition, computation, and application of discrete TFDs. The first contribution is a new discrete definition of TFDs. A discrete TFD (DTFD) should be free from the sampling-related distortion known as aliasing and satisfy all the important mathematical properties that the continuous TFD satisfies. Many different DTFD definitions exist but none come close to attaining this ideal. I propose three new components which make up the DTFD: 1) a new discrete Wigner--Ville distribution (DWVD) definition which satisfies all properties, 2) a new discrete analytic signal which minimises aliasing in the DWVD, and 3) a new method to define and convolve the discrete kernel with the DWVD to produce the DTFD. The result: a DTFD definition that, relative to the existing definitions, better approximates the ideal DTFD. The second contribution is two sets of computationally efficient algorithms to compute the proposed DTFD. The first set of algorithms computes the DTFD exactly; the second set requires less memory than the first set by computing time- and, or frequency-decimated versions of the DTFD. Both sets of algorithms reduce the computational load by exploiting symmetries in the DTFD and by constructing kernel-specific algorithms for four different kernel types. The third, and final, contribution is a biomedical application for the proposed DTFD and algorithms. This application is to accurately detect seizure events in newborn electroencephalogram (EEG) signals. Existing detection methods do not perform well enough for use in a clinical setting. I propose a new method which is more robust than existing methods and show how using the proposed DTFD, comparative to an existing DTFD, improves detection performance for this method. In summary, this dissertation makes practical contributions to the area of time--frequency signal processing by proposing an improved DTFD definition, efficient DTFD algorithms, and an improved newborn EEG seizure detection method using DTFDs.

Most signal processing methods were developed for continuous signals. Digital devices, such as the computer, process only discrete signals. This dissertation proposes new techniques to accurately define and efficiently implement an important signal processing method---the time--frequency distribution (TFD)---using discrete signals. The TFD represents a signal in the joint time--frequency domain. Because these distributions are a function of both time and frequency they, unlike traditional signal processing methods, can display frequency content that changes over time. TFDs have been used successfully in many signal processing applications as almost all real-world signals have time-varying frequency content. Although TFDs are well defined for continuous signals, defining and computing a TFD for discrete signals is problematic. This work overcomes these problems by making contributions to the definition, computation, and application of discrete TFDs. The first contribution is a new discrete definition of TFDs. A discrete TFD (DTFD) should be free from the sampling-related distortion known as aliasing and satisfy all the important mathematical properties that the continuous TFD satisfies. Many different DTFD definitions exist but none come close to attaining this ideal. I propose three new components which make up the DTFD: 1) a new discrete Wigner--Ville distribution (DWVD) definition which satisfies all properties, 2) a new discrete analytic signal which minimises aliasing in the DWVD, and 3) a new method to define and convolve the discrete kernel with the DWVD to produce the DTFD. The result: a DTFD definition that, relative to the existing definitions, better approximates the ideal DTFD. The second contribution is two sets of computationally efficient algorithms to compute the proposed DTFD. The first set of algorithms computes the DTFD exactly; the second set requires less memory than the first set by computing time- and, or frequency-decimated versions of the DTFD. Both sets of algorithms reduce the computational load by exploiting symmetries in the DTFD and by constructing kernel-specific algorithms for four different kernel types. The third, and final, contribution is a biomedical application for the proposed DTFD and algorithms. This application is to accurately detect seizure events in newborn electroencephalogram (EEG) signals. Existing detection methods do not perform well enough for use in a clinical setting. I propose a new method which is more robust than existing methods and show how using the proposed DTFD, comparative to an existing DTFD, improves detection performance for this method. In summary, this dissertation makes practical contributions to the area of time--frequency signal processing by proposing an improved DTFD definition, efficient DTFD algorithms, and an improved newborn EEG seizure detection method using DTFDs.

Most signal processing methods were developed for continuous signals. Digital devices, such as the computer, process only discrete signals. This dissertation proposes new techniques to accurately define and efficiently implement an important signal processing method---the time--frequency distribution (TFD)---using discrete signals. The TFD represents a signal in the joint time--frequency domain. Because these distributions are a function of both time and frequency they, unlike traditional signal processing methods, can display frequency content that changes over time. TFDs have been used successfully in many signal processing applications as almost all real-world signals have time-varying frequency content. Although TFDs are well defined for continuous signals, defining and computing a TFD for discrete signals is problematic. This work overcomes these problems by making contributions to the definition, computation, and application of discrete TFDs. The first contribution is a new discrete definition of TFDs. A discrete TFD (DTFD) should be free from the sampling-related distortion known as aliasing and satisfy all the important mathematical properties that the continuous TFD satisfies. Many different DTFD definitions exist but none come close to attaining this ideal. I propose three new components which make up the DTFD: 1) a new discrete Wigner--Ville distribution (DWVD) definition which satisfies all properties, 2) a new discrete analytic signal which minimises aliasing in the DWVD, and 3) a new method to define and convolve the discrete kernel with the DWVD to produce the DTFD. The result: a DTFD definition that, relative to the existing definitions, better approximates the ideal DTFD. The second contribution is two sets of computationally efficient algorithms to compute the proposed DTFD. The first set of algorithms computes the DTFD exactly; the second set requires less memory than the first set by computing time- and, or frequency-decimated versions of the DTFD. Both sets of algorithms reduce the computational load by exploiting symmetries in the DTFD and by constructing kernel-specific algorithms for four different kernel types. The third, and final, contribution is a biomedical application for the proposed DTFD and algorithms. This application is to accurately detect seizure events in newborn electroencephalogram (EEG) signals. Existing detection methods do not perform well enough for use in a clinical setting. I propose a new method which is more robust than existing methods and show how using the proposed DTFD, comparative to an existing DTFD, improves detection performance for this method. In summary, this dissertation makes practical contributions to the area of time--frequency signal processing by proposing an improved DTFD definition, efficient DTFD algorithms, and an improved newborn EEG seizure detection method using DTFDs.

Most signal processing methods were developed for continuous signals. Digital devices, such as the computer, process only discrete signals. This dissertation proposes new techniques to accurately define and efficiently implement an important signal processing method---the time--frequency distribution (TFD)---using discrete signals. The TFD represents a signal in the joint time--frequency domain. Because these distributions are a function of both time and frequency they, unlike traditional signal processing methods, can display frequency content that changes over time. TFDs have been used successfully in many signal processing applications as almost all real-world signals have time-varying frequency content. Although TFDs are well defined for continuous signals, defining and computing a TFD for discrete signals is problematic. This work overcomes these problems by making contributions to the definition, computation, and application of discrete TFDs. The first contribution is a new discrete definition of TFDs. A discrete TFD (DTFD) should be free from the sampling-related distortion known as aliasing and satisfy all the important mathematical properties that the continuous TFD satisfies. Many different DTFD definitions exist but none come close to attaining this ideal. I propose three new components which make up the DTFD: 1) a new discrete Wigner--Ville distribution (DWVD) definition which satisfies all properties, 2) a new discrete analytic signal which minimises aliasing in the DWVD, and 3) a new method to define and convolve the discrete kernel with the DWVD to produce the DTFD. The result: a DTFD definition that, relative to the existing definitions, better approximates the ideal DTFD. The second contribution is two sets of computationally efficient algorithms to compute the proposed DTFD. The first set of algorithms computes the DTFD exactly; the second set requires less memory than the first set by computing time- and, or frequency-decimated versions of the DTFD. Both sets of algorithms reduce the computational load by exploiting symmetries in the DTFD and by constructing kernel-specific algorithms for four different kernel types. The third, and final, contribution is a biomedical application for the proposed DTFD and algorithms. This application is to accurately detect seizure events in newborn electroencephalogram (EEG) signals. Existing detection methods do not perform well enough for use in a clinical setting. I propose a new method which is more robust than existing methods and show how using the proposed DTFD, comparative to an existing DTFD, improves detection performance for this method. In summary, this dissertation makes practical contributions to the area of time--frequency signal processing by proposing an improved DTFD definition, efficient DTFD algorithms, and an improved newborn EEG seizure detection method using DTFDs.