This paper presents general considerations concerning the application of artificial neural networks algorithms, more specifically the back-propagation learning algorithm and feed-forward multi-layer networks, to several problems in power system. The main application in power systems is the load forecasting, and two solution methods are used to solve it. (author). 45 figs., 32 tabs., 144 refs.
Most of the information in this chapter is obtained from 1, 2, 3, 6, 8 and 9 annexes. It is worthwhile to clarify that in most of the cases they are estimations of the available resources, such as: trash, geothermal, solar energy, wind energy, biomass and mini and micro-hydro-energy. In several cases punctual measurements have been made, which do not represent a generic value of a national reserve. [Spanish] La mayor parte de la informacion en este capitulo es obtenida de los anexos 1, 2, 3, 6, 8 y 9. Cabe aclarar que en la mayoria de los casos son estimaciones de los recursos disponibles, tales como: la basura, la geotermia, la energia solar, la energia eolica, la biomasa y, la mini y la micro-hidroenergia. En varios casos se han realizado mediciones puntuales, las cuales no representan un valor generico de una reserva nacional.
One of the essential steps in the main scope Project Management is the Time Management made by the planning and control of the project schedule. In this work is presented the resource constrained scheduling problem beyond its mathematical formulation and a review of papers about this issue. In sequence is presented a practical example of this model considering a simplified model of an engineering project schedule of oil production equipment. The results obtained with the model application are shown and the conclusions about the work with resource constrained scheduling problems. (author)
This work describes a procedure for the adaptive time dependent Finite Element Method using an automatic mesh refinement (H-Version) that efficiently reduces estimated errors ( a posteriori) below pre-assigned limits. Classical model problem for steady-state heat transfer are investigated, and the results are compared with the analytical solution. Then some typical time-dependent problem are qualitatively analysed. (author) 10 refs., 7 figs.
This work presents a digital adaptive Power System Stabilizer (PSS) which operates in a gain scheduling scheme. It`s parameters are designed for a lot of different operating regions in a P x Q plane (active and reactive powers), and saved in a microcomputer real time control. During working, the PSS identifies the present region of operation, and synthesizes its damping signal in accordance with the parameters for that region. As an extension of the method, a neural PSS, which uses the set of parameters of each region as a standard set to train a neural network to form this PSS, is also proposed. The tests presented show good performance for both PSS, when compared to a conventional (non adaptive) one. (author) 10 refs., 5 figs., 1 tab.; e-mail: jalb at guama.cpgee.ufpa.br