Carlos Tavares da Costa

Técnica de controle preditivo baseado em modelo aplicada ao controle de tensão de um gerador síncrono - resultados experimentais

Neste trabalho serao apresentados os resultados da avaliacao experimental de uma metodologia de controle digital preditivo auto-ajustavel aplicada ao controle de tensao de um sistema de geracao de energia de escala reduzida. Um estimador recursivo baseado no conhecido metodo de minimos quadrados e utilizado na etapa de identificacao do controlador preditivo proposto. A etapa de calculo da lei de controle e realizada com o algoritmo Generalized Predictive Controller (GPC). A avaliacao experimental foi realizada com testes de resposta ao degrau e rastreamento aplicados em diferentes condicoes operacionais do sistema de potencia estudado. Para fins de comparacao, tambem serao apresentados os resultados da avaliacao de um controlador auto-ajustavel que utiliza o metodo de alocacao de polos para a sintese do sinal de controle e tres controladores digitais com parâmetros fixos.

Intelligent Analysis Program Applied to Production Logs in Oil and Gas Wells

Basic Production Logging can be interpreted in intermittent wells using the intelligent analysis program presents in this work, called PAINTWELL, which permits to use of same string configuration than a conventional Production Logging. This program can be performed in layered reservoir, determining the flowrate and type of the fluid for each production zone. The use of Artificial Neural Network as integrated part of this analysis program, provides a powerful tool to describe how downhole logging and advances in computer processing and interpretation data could spearhead the transformation that can take place in production logging with use of this new technique. This methodology should aid the production and reservoir engineering to solve recovery problems during the well life. The results of the production log using the PAINTWELL method is very similar to measurements using test separator in production facilities for each single perforated interval. This program was applied successfully for a typical intermittent well, which produces oil with high water cut using a gas-lift completion system.

Automation meta-system applied to smart grid convergence of low voltage distribution legacy grids

This paper presents the description of a system model (meta-system) that specifies structural and functional aspects of a middleware architecture, such as: composition elements, interfaces, interaction rules, protocols, etc. The implementation of such middleware architecture (middleware-layer) aim to make a retrofitting in the conventional low voltage circuits to convert them into smart circuits and, thus, enable the implementation of a set of functionalities compatible with (adhering to) the definition requirements of a smart grid (ex.: self-healing, energy efficiency, distributed generation compatibility, resilience, renewable energies support, distribution systems operation automation, load management, demand side management, demand response and generation curtailment). This process will be here referred to as Convergence Smart Grid. The meta-system proposed establishes a basic platform consisting of interoperable hardware and software units, which supervise and/or control operating variables in each of the different component segments of the distribution system and according to their particularities. This solution was developed under a systemic approach, which enables the implementation of a wide scope of functionalities and resources aligned to the smart grid paradigm; it also establishes a coherent and harmonious automation solution with structural and functional aspects of low voltage distribution grids that are known to be systemic; more specifically, those that can be submitted to processes tele-supervision, remote and electronic control.

A NEURAL POWER SYSTEM STABILIZER FROM LOCAL LINEAR CONTROLLERS

Abstract This paper presents the design of a Power System Stabilizer synthesized using an Artificial Neural Network. The patterns used in the network training are sets of controller parameters, previously calculated for several system operation points using the pole-placement method. The trained network presents, as its main characteristic, uniform values for all the stabilizers parameters when the system synchronous machine is generating reactive power, but these same parameters suffer great variations when reactive power is being absorbed by the machine. Simulation tests show very good performance for the proposed Neural PSS, when compared with a fixed-parameter stabilizer.

Adaptive controller adjusted by a fuzzy logic based supervisor from a local model network

Abstract This work presents a digital adaptive control method, where the controller is designed using a self-tuning structure. Parametric models are identified off line for several differently selected operating points of the plant to form a Local Model Network (LMN), and saved as a look-up table in a microcomputer real time control. During working, a supervisor identifies the actual condition of operation and uses a set of logic fuzzy rules to estimate a plant model as a function of the рrе-identified models for the nearest selected operating points. Then, using this plant model, the controller is designed. This method is employed in the design of a Power System Stabilizer (PSS) and the control obtained for a test with two severe faults showed good performance for this PSS, when compared to a conventional (non adaptive) one.

Tuning of a Adaptive Power System Stabilizer by a Fuzzy Logic Based Supervisor from a Local Model Network

Abstract This work presents a digital adaptive Power System Stabilizer (PSS) which operates in a self-tuning structure. The parametric models are identified off line for several differently selected operating points in a PxQ plane (active and reactive powers) to form a Local Model Network (LMN), and saved under the form of a look-up table in a microcomputer real time control. During working, a supervisor identifies the actual condition of operation and uses a set of logic fuzzy rules to estimate a plant model as a function of the pre-identified models for the nearest selected operating points. Then, using this plant model, the PSS is designed. The tests presented show good performance for this PSS, when compared to a conventional (non adaptive) one.