Probabilistic approach for reference regimes development for expert systems on electrical distribution network diagnosis

The expert systems are more and more an essential tool for carrying out the electric “smart grids” of the future. The development of expert systems requires the use of electrical test systems and reference regimes with special characteristics to identify how expert systems will behave not only for quite normal or even trivial schemes and operating condition, but even in cases of regimes and / or schemes with very particular characteristics or exceptional ones. In this paper, the authors propose an original procedure to develop reference regimes, based on a probabilistic approach. The core of the procedure is the probabilistic consideration as independent random variable for each consumer. A set of minimum 500 to 1000 regimes are generated based of probabilistic values of each bus load. In the following an original procedure to classification of these regimes is applied in order to select so called reference regimes. As a consequence the selected regimes are more realistic reference regimes for the competitive environment that characterizes the energy sector at the moment and obviously in the future. The presented case study and the other results demonstrate the validity of the procedure adopted by the authors.