Accurate rainfall forecasting using weather radar imagery has always been a crucial and predominant task in the field of meteorology [1], [2], [3] and [4]. Competitive Radial Basis Function Neural Networks (CRBFNN) [5] is one of the methods used for weather radar image based forecasting. Recently, an alternative CRBFNN based approach [6] was introduced to model the precipitation events. The difference between the techniques presented in [5] and [6] is in the approach used to model the rainfall image. Overall, it was shown that the modified CRBFNN approach [6] is more computationally efficient compared to the CRBFNN approach [5]. However, both techniques [5] and [6] share the same prediction stage. In this thesis, a different GRBFNN approach is presented for forecasting Gaussian envelope parameters. The proposed method investigates the concept of parameter dependency among Gaussian envelopes. Experimental results are also presented to illustrate the advantage of parameters prediction over the independent series prediction.

Submitted by Amanda Silva (amanda.osilva2@ufpe.br) on 2015-03-03T14:16:54Z No. of bitstreams: 2 Dissertação Mariel Cadena da Matta.pdf: 2355898 bytes, checksum: 9c0530af680cf965137a2385d949b799 (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) / Made available in DSpace on 2015-03-03T14:16:54Z (GMT). No. of bitstreams: 2 Dissertação Mariel Cadena da Matta.pdf: 2355898 bytes, checksum: 9c0530af680cf965137a2385d949b799 (MD5) license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Previous issue date: 2013 / FACEPE / A Dosimetria citogenética empregando análise de cromossomos dicêntricos é o “padrão ouro” para estimativas da dose absorvida após exposições acidentais às radiações ionizantes. Todavia, este método é laborioso e dispendioso, o que torna necessária a introdução de ferramentas computacionais que dinamizem a contagem dessas aberrações cromossômicas radioinduzidas. Os atuais softwares comerciais, utilizados no processamento de imagens em Biodosimetria, são em sua maioria onerosos e desenvolvidos em sistemas dedicados, não podendo ser adaptados para microscópios de rotina laboratorial. Neste contexto, o objetivo da pesquisa foi o desenvolvimento do software ChromoSomeClassification para processamento de imagens de metáfases de linfócitos (não irradiados e irradiados) coradas com Giemsa a 5%. A principal etapa da análise citogenética automática é a separação correta dos cromossomos do fundo, pois a execução incorreta desta fase compromete o desenvolvimento da classificação automática. Desta maneira, apresentamos uma proposta para a sua resolução baseada no aprimoramento da imagem através das técnicas de mudança do sistema de cores, subtração do background e aumento do contraste pela modificação do histograma. Assim, a segmentação por limiar global simples, seguida por operadores morfológicos e pela técnica de separação de objetos obteve uma taxa de acerto de 88,57%. Deste modo, os cromossomos foram enfileirados e contabilizados, e assim, a etapa mais laboriosa da Dosimetria citogenética foi realizada. As características extraídas dos cromossomos isolados foram armazenadas num banco de dados para que a classificação automática fosse realizada através da Rede Neural com Funções de Ativação de Base Radial (RBF). O software proposto alcançou uma taxa de sensibilidade de 76% e especificidade de 91% que podem ser aprimoradas através do acréscimo do número de objetos ao banco de dados e da extração de mais características dos cromossomos.

The static synchronous series compensator (SSSC) is a series controller of Flexible AC Transmission Systems (FACTS). It can be controlled by Thyristors, it also has the ability of fast control adjustments and high frequency operation. Through impedance compensation, it is able to control the magnitude and directions of the real power flow in the transmission system. In order to achieve a fast and steady response for real power control in power systems, this thesis proposed a unified intelligent controller, which consists of RBFNN and GA for the SSSC to provide better control features for real power control in the dynamic operations of power systems. Finally, the simulation results of the proposed controllers is compared with the conventional proportional plus integral (PI) controllers to demonstrate the superiority and effectiveness of the unified intelligent controller.

Stability of the power quality is one of the objectives that power companies always try to assure. With energy shortage and the increases of fuel cost over years, reduction of expenses in all areas is another effort of the power company. Dealing with the above problems, Taiwan Power Company sets up a standard voltage for secondary side of each primary substation. Standard voltage is a commitment of expected 69kV primary substation bus voltage. A proper setting of the standard voltage can reduce voltage variation, in the secondary substation, and reduce the operation frequencies of the on load tap changer. Besides, it can prolong the service life and the maintenance cycle, and it can also reduce maintenance cost of each main transformer. This study proposes a method to calculate the standard voltage to improve the shortcomings that the voltage used to be set up with experience rule. The load and voltage data were used to build a neural network model. Improved particle swarm optimizer was used to find the parameters of the radial basis function neural network in order to build an efficient network. This network uses improved particle swarm optimizer again to the standard voltage. The proposed approach has been verified by the comparison of winter and summer standard voltages on the Tainan primary substation of taipower with accurate results.

Weather radar imagery is important for several remote sensing applications including tracking of storm fronts and radar echo classification. In particular, tracking of precipitation events is useful for both forecasting and classification of rain/non-rain events since non-rain events usually appear to be static compared to rain events. Recent weather radar imaging-based forecasting approaches [3] consider that precipitation events can be modeled as a combination of localized functions using Radial Basis Function Neural Networks (RBFNNs). Tracking of rain events can be performed by tracking the parameters of these localized functions. The RBFNN-based techniques used in forecasting are not only computationally expensive, but also moderately effective in modeling small size precipitation events. In this thesis, an existing RBFNN technique [3] was implemented to verify its computational efficiency and forecasting effectiveness. The feasibility of modeling precipitation events using RBFNN effectively was evaluated, and several modifications to the existing technique have been proposed.

<p>This study is based on three models, Markov model, Hidden Markov model and the Radial basis function neural network. A number of work has been done before about application of these three models to the stock market. Though, individual researchers have developed their own techniques to design and test the Radial basis function neural network. This paper aims to show the different ways and precision of applying these three models to predict price processes of the stock market. By comparing the same group of data, authors get different results. Based on Markov model, authors find a tendency of stock market in future and, the Hidden Markov model behaves better in the financial market. When the fluctuation of the stock price index is not drastic, the Radial basis function neural network has a nice prediction.</p>

This study is based on three models, Markov model, Hidden Markov model and the Radial basis function neural network. A number of work has been done before about application of these three models to the stock market. Though, individual researchers have developed their own techniques to design and test the Radial basis function neural network. This paper aims to show the different ways and precision of applying these three models to predict price processes of the stock market. By comparing the same group of data, authors get different results. Based on Markov model, authors find a tendency of stock market in future and, the Hidden Markov model behaves better in the financial market. When the fluctuation of the stock price index is not drastic, the Radial basis function neural network has a nice prediction.

The present work focuses on the two areas of investigation: system identification of helicopter and design of controller for the helicopter. Helicopter system identification, the first subject of investigation in this thesis, can be described as the extraction of system characteristics/dynamics from measured flight test data. Wind tunnel experimental data suffers from scale effects and model deficiencies. The increasing need for accurate models for the design of high bandwidth control system for helicopters has initiated a renewed interest in and a more active use of system identification. Besides, system identification is likely to become mandatory in the future for model validation of ground based helicopter simulators. Such simulators require accurate models in order to be accepted by pilots and regulatory authorities like Federal Aviation Regulation for realistic complementary helicopter mission training. Two approaches are widely used for system identification, namely, black box and gray box approach. In the black-box approach, the relationship between input-output data is approximated using nonparametric methods such as neural networks and in such a case, internal details of the system and model structure may not be known. In the gray box approach, parameters are estimated after defining the model structure. In this thesis, both black box and gray box approaches are investigated. In the black box approach, in this thesis, a comparative study and analysis of different Recurrent Neural Networks(RNN) for the identification of helicopter dynamics using flight data is investigated. Three different RNN architectures namely, Nonlinear Auto Regressive eXogenous input(NARX) model, neural network with internal memory known as Memory Neuron Networks(MNN)and Recurrent MultiLayer perceptron (RMLP) networks are used to identify dynamics of the helicopter at various flight conditions. Based on the results, the practical utility, advantages and limitations of the three models are critically appraised and it is found that the NARX model is most suitable for the identification of helicopter dynamics. In the gray box approach, helicopter model parameters are estimated after defining the model structure. The identification process becomes more difficult as the number of degrees-of-freedom and model parameters increase. To avoid the drawbacks of conventional methods, neural network based techniques, called the delta method is investigated in this thesis. This method does not require initial estimates of the parameters and the parameters can be directly extracted from the flight data. The Radial Basis Function Network(RBFN)is used for the purpose of estimation of parameters. It is shown that RBFN is able to satisfactorily estimate stability and control derivatives using the delta method. The second area of investigation addressed in this thesis is the control of helicopter in flight. Helicopter requires use of a control system to achieve satisfactory flight. Designing a classical controller involves developing a nonlinear model of the helicopter and extracting linearized state space matrices from the nonlinear model at various flight conditions. After examining the stability characteristics of the helicopter, the desired response is obtained using a feedback control system. The scheduling of controller gains over the entire envelope is used to obtain the desired response. In the present work, a helicopter having a soft inplane four bladed hingeless main rotor and a four-bladed tail rotor with conventional mechanical controls is considered. For this helicopter, a mathematical model and also a model based on neural network (using flight data) has been developed. As a precursor, a feed back controller, the Stability Augmentation System(SAS), is designed using linear quadratic regulator control with full state feedback and LQR with out put feedback approaches. SAS is designed to meet the handling qualities specification known as Aeronautical Design Standard ADS-33E-PRF. The control gains have been tuned with respect to forward speed and gain scheduling has been arrived at. The SAS in the longitudinal axis meets the requirement of the Level1 handling quality specifications in hover and low speed as well as for forward speed flight conditions. The SAS in the lateral axis meets the requirement of the Level2 handling quality specifications in both hover and low speed as well as for forward speed flight conditions. Such conventional design of control has served useful purposes, however, it requires considerable flight testing which is time consuming, to demonstrate and tune these control law gains. In modern helicopters, the stringent requirements and non-linear maneuvers make the controller design further complicated. Hence, new design tools have to be explored to control such helicopters. Among the many approaches in adaptive control, neural networks present a potential alternative for modeling and control of nonlinear dynamical systems due to their approximating capabilities and inherent adaptive features. Furthermore, from a practical perspective, the massive parallelism and fast adaptability of neural network implementations provide more incentive for further investigation in problems involving dynamical systems with unknown non-linearity. Therefore, adaptive control approach based on neural networks is proposed in this thesis. A neural network based Feedback Error Neural adaptive Controller(FENC) is designed for a helicopter. The proposed controller scheme is based on feedback error learning strategy in which the outer loop neural controller enhances the inner loop conventional controller by compensating for unknown non-linearity and parameter un-certainties. Nonlinear Auto Regressive eXogenous input(NARX)neural network architecture is used to approximate the control law and the controller network parameters are adapted using updated rules Lyapunov synthesis. An offline (finite time interval)and on-line adaptation strategy is used to approximate system uncertainties. The results are validated using simulation studies on helicopter undergoing an agile maneuver. The study shows that the neuro-controller meets the requirements of ADS-33 handling quality specifications. Even though the tracking error is less in FENC scheme, the control effort required to follow the command is very high. To overcome these problems, a Direct Adaptive Neural Control(DANC)scheme to track the rate command signal is presented. The neural controller is designed to track rate command signal generated using the reference model. For the simulation study, a linearized helicopter model at different straight and level flight conditions is considered. A neural network with a linear filter architecture trained using back propagation through time is used to approximate the control law. The controller network parameters are adapted using updated rules Lyapunov synthesis. The off-line trained (for finite time interval)network provides the necessary stability and tracking performance. The on-line learning is used to adapt the network under varying flight conditions. The on-line learning ability is demonstrated through parameter uncertainties. The performance of the proposed direct adaptive neural controller is compared with feedback error learning neural controller. The performance of the controller has been validated at various flight conditions. The theoretical results are validated using simulation studies based on a nonlinear six degree-of-freedom helicopter undergoing an agile maneuver. Realistic gust and sensor noise are added to the system to study the disturbance rejection properties of the neural controllers. To investigate the on-line learning ability of the proposed neural controller, different fault scenarios representing large model error and control surface loss are considered. The performances of the proposed DANC scheme is compared with the FENC scheme. The study shows that the neuro-controller meets the requirements of ADS-33 handling quality specifications.

Face recognition is a field of biometrics that deals with identification of subjects based on features present in the images of their faces. The factors that make face recognition popular and favorite as compared to other biometric methods are easier operation and ability to identify subjects without their knowledge. With these features, face recognition has become an integral part of the present day security systems, targeting a smart and secure world. There are various factors that de ne the performance of a face recognition system. The most important among them are recognition accuracy of algorithm used and time taken for recognition. Recognition accuracy of the face recognition algorithm gets affected by changes in pose, facial expression and illumination along with occlusions in the images. There have been a number of algorithms proposed to enable recognition under these ambient changes. However, it has been hard to and a single algorithm that can efficiently recognize faces in all the above mentioned conditions. Moreover, achieving real time performance for most of the complex face recognition algorithms on embedded platforms has been a challenge. Real-time performance is highly preferred in critical applications such as identification of crime suspects in public. As available software solutions for FR have significantly large latency in recognizing individuals, they are not suitable for such critical real-time applications. This thesis focuses on real-time aspect of FR, where acceleration of the algorithms is achieved by means of parallel hardware architectures. The major contributions of this work are as follows. We target to design a face recognition system that can identify at most 30 faces in each frame of video at 15 frames per second, which amounts to 450 recognitions per second. In addition, we target to achieve good recognition accuracy along with scalability in terms of database size and input image resolutions. To design a system with these specifications, as a first step, we explore algorithms in literature and come up with a hybrid face recognition algorithm. This hybrid algorithm shows good recognition accuracy on face images with changes in illumination, pose and expressions, and also with occlusions. In addition the computations in the algorithm are modular in nature which are suitable for real-time realizations through parallel processing. The face recognition system consists of a face detection module to detect faces in the input image, which is followed by a face recognition module to identify the detected faces. There are well established algorithms and architectures for face detection in literature which can perform detection at 15 frames per second on video frames. Detected faces of different sizes need to be scaled to the size specified by the face recognition module. To meet the real-time constraints, we propose a hardware architecture for real-time bi-cubic convolution interpolation with dynamic scaling factors. To recognize the resized faces in real-time, a scalable parallel pipelined architecture is designed for the hybrid algorithm which can perform 450 recognitions per second on a database containing grayscale images of at most 450 classes on Virtex 6 FPGA. To provide flexibility and programmability, we extend this design to REDEFINE, a multi-core massively parallel reconfigurable architecture. In this design, we come up with FR specific programmable cores termed Scalable Unit for Region Evaluation (SURE) capable of performing modular computations in the hybrid face recognition algorithm. We replicate SUREs in each tile of REDEFINE to construct a face recognition module termed REDEFINE for Face Recognition using SURE Homogeneous Cores (REFRESH). There is a need to learn new unseen faces on-line in practical face recognition systems. Considering this, for real-time on-line learning of unseen face images, we design tiny processors termed VOP, Processor for Vector Operations. VOPs function as coprocessors to process elements under each tile of REDEFINE to accelerate micro vector operations appearing in the synaptic weight computations. We also explore deep neural networks which operate similar to the processing in human brain and capable of working on very large face databases. We explore the field of Random matrix theory to come up with a solution for synaptic weight initialization in deep neural networks for better classification . In addition, we perform design space exploration of hardware architecture for deep convolution networks and conclude with directions for future work.

Today engineering companies expend a large amount of resources on the detection and correction of the bugs (defects) in their software. These bugs are usually due to errors and mistakes made by programmers while writing the code or writing the specifications. No tool is able to detect all of these bugs. Some of these bugs remain undetected despite testing of the code. For these reasons, many researchers have tried to find indicators in the software’s source codes that can be used to predict the presence of bugs. Every bug in the source code is a potentially failure of the program to perform as expected. Therefore, programs are tested with many different cases in an attempt to cover all the possible paths through the program to detect all of these bugs. Early prediction of bugs informs the programmers about the location of the bugs in the code. Thus, programmers can more carefully test the more error prone files, and thus save a lot of time by not testing error free files. This thesis project created a tool that is able to predict error prone source code written in C++. In order to achieve this, we have utilized one predictor which has been extremely well studied: software metrics. Many studies have demonstrated that there is a relationship between software metrics and the presence of bugs. In this project a Neuro-Fuzzy hybrid model based on Fuzzy c-means and Radial Basis Neural Network has been used. The efficiency of the model has been tested in a software project at Ericsson. Testing of this model proved that the program does not achieve high accuracy due to the lack of independent samples in the data set. However, experiments did show that classification models provide better predictions than regression models. The thesis concluded by suggesting future work that could improve the performance of this program. / Idag spenderar ingenjörsföretag en stor mängd resurser på att upptäcka och korrigera buggar (fel) i sin mjukvara. Det är oftast programmerare som inför dessa buggar på grund av fel och misstag som uppkommer när de skriver koden eller specifikationerna. Inget verktyg kan detektera alla dessa buggar. Några av buggarna förblir oupptäckta trots testning av koden. Av dessa skäl har många forskare försökt hitta indikatorer i programvarans källkod som kan användas för att förutsäga förekomsten av buggar. Varje fel i källkoden är ett potentiellt misslyckande som gör att applikationen inte fungerar som förväntat. För att hitta buggarna testas koden med många olika testfall för att försöka täcka alla möjliga kombinationer och fall. Förutsägelse av buggar informerar programmerarna om var i koden buggarna finns. Således kan programmerarna mer noggrant testa felbenägna filer och därmed spara mycket tid genom att inte behöva testa felfria filer. Detta examensarbete har skapat ett verktyg som kan förutsäga felbenägen källkod skriven i C ++. För att uppnå detta har vi utnyttjat en välkänd metod som heter Software Metrics. Många studier har visat att det finns ett samband mellan Software Metrics och förekomsten av buggar. I detta projekt har en Neuro-Fuzzy hybridmodell baserad på Fuzzy c-means och Radial Basis Neural Network använts. Effektiviteten av modellen har testats i ett mjukvaruprojekt på Ericsson. Testning av denna modell visade att programmet inte Uppnå hög noggrannhet på grund av bristen av oberoende urval i datauppsättningen. Men gjordt experiment visade att klassificering modeller ger bättre förutsägelser än regressionsmodeller. Exjobbet avslutade genom att föreslå framtida arbetet som skulle kunna förbättra detta program. / Actualmente las empresas de ingeniería derivan una gran cantidad de recursos a la detección y corrección de errores en sus códigos software. Estos errores se deben generalmente a los errores cometidos por los desarrolladores cuando escriben el código o sus especificaciones.  No hay ninguna herramienta capaz de detectar todos estos errores y algunos de ellos pasan desapercibidos tras el proceso de pruebas. Por esta razón, numerosas investigaciones han intentado encontrar indicadores en los códigos fuente del software que puedan ser utilizados para detectar la presencia de errores. Cada error en un código fuente es un error potencial en el funcionamiento del programa, por ello los programas son sometidos a exhaustivas pruebas que cubren (o intentan cubrir) todos los posibles caminos del programa para detectar todos sus errores. La temprana localización de errores informa a los programadores dedicados a la realización de estas pruebas sobre la ubicación de estos errores en el código. Así, los programadores pueden probar con más cuidado los archivos más propensos a tener errores dejando a un lado los archivos libres de error. En este proyecto se ha creado una herramienta capaz de predecir código software propenso a errores escrito en C++. Para ello, en este proyecto se ha utilizado un indicador que ha sido cuidadosamente estudiado y ha demostrado su relación con la presencia de errores: las métricas del software. En este proyecto un modelo híbrido neuro-disfuso basado en Fuzzy c-means y en redes neuronales de función de base radial ha sido utilizado. La eficacia de este modelo ha sido probada en un proyecto software de Ericsson. Como resultado se ha comprobado que el modelo no alcanza una alta precisión debido a la falta de muestras independientes en el conjunto de datos y los experimentos han mostrado que los modelos de clasificación proporcionan mejores predicciones que los modelos de regresión. El proyecto concluye sugiriendo trabajo que mejoraría el funcionamiento del programa en el futuro.