METHOD AND INTELLIGENT COMPUTERIZED SYSTEM FOR PREDICTION, ANALYSIS AND MONITORING OF PRODUCTION AND CONSUMPTION OF ELECTRIC POWER FROM WIND FARMS

摘要

The invention relates to a method and an intelligent computerized system meant to predict, analyze and monitor the production and consumption of electric power from wind power plants. According to the invention, the method consists of a first stage in which steps are made for training an artificial neural weather-forecast network and for forecasting using the said network, followed by a second stage in which steps are made for training an artificial neural energy-predicting network and for forecasting using the said network. The intelligent computerized system, as claimed, comprises an artificial neural weather-forecast network (RNAmeteo) using the following architecture: one neuron for each of the input data (temperature, absolute wind direction, average wind velocity), 15 neurons for the hidden layer, 3 neurons for each of the output layer and the output data (temperature, absolute wind direction, average wind velocity, as forecasted), respectively, an artificial neural network (RNAenerg) predicting the wind power production and the consumed power, using the following architecture: one neuron for each of the input data (temperature, absolute wind direction, average wind velocity) for each turbine of the wind power plant, 12 neurons for the hidden layer and 2 neurons for each of the output layer and the output data (energy output and consumed energy), Levenberg-Marquardt algorithm (LM), Bayesian regularization algorithm and scaled conjugate gradient algorithm (SCG), MATLAB Compiler SDK tool, a shared library of functions for C language (Bibl-C), a shared library of functions for C++ language (Bibl-C++), a developing framework Microsoft.NET, a Java packet (PJFpred) offering access to predictive functions and a reusable component of the Component Object Model type.