Currently, the NIS is segmented into four interconnected systems: South, Southeast/Midwest, Northeast and North. Since it is a huge system, for medium and long term planning an aggregation occurs from the plants reservoirs into power equivalent reservoirs, one for each subsystem. There is also the aggregation of inflows to energy plants in affluent natural energy (ANE), which corresponds to the estimate of the energy that can be generated with all the inflows to each reservoir, under a given operational policy [2].

Each ANE is a non-stationary series, arising from periods of flooding and dry season in the year, for periods of twelve months. These series exhibit temporal correlation and spatial correlation.

To model the intrinsically non-linear and stochastic behavior of the inflow series, a new generic model of a stochastic process was proposed [3], called neural stochastic process (NSP). In its original formation this model is capable of capturing the time correlation but does not incorporate the spatial correlation of the ANE series.

The aim of this study was to elaborate a new version of the NSP model with the objective of incorporating the spatial correlation existing in the ANE series. The new model is a single neural stochastic process for all NIS subsystems and its ANNs have four outputs, where each output provides values to define the synthetic series, one for each subsystem. Then, the series of four NIS subsystems are simultaneously generated, so this new model is called the NSP4out. The goal of this model is to generate a synthetic time series as probable as the ANE historical series, covering any period of time.

The results obtained have shown that the synthetic series generated by using the NSP4out model actually provide the same behavior as the historical series set, for the next five years. In the model evaluated in this work, all subsystems use the same amount of past terms series. By allowing a different order for each specific subsystem, we believe that the performance can be further improved.

1

"ONS - Brazilian Operator of the Electric System", 2011. http://www.ons.com.br

2

M.V.F. Pereira, L.M.V.G. Pinto, "Operation planning of large scale hydroelectric systems", Proc.8th Power System Computation Cont., 1984.

3

L.C.D. Campos, M.M.B.R. Vellasco, J.G.L. Lazo, "A Stochastic Model based on Neural Networks", International Joint Conference on Neural Networks, 2011. doi:10.1109/IJCNN.2011.6033399

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £70 +P&P)