science-review.ru
Scientific journal
Scientific Review. Technical science
ISSN 2500-0799
ПИ №ФС77-57440
ISSN 2500-0799
ПИ №ФС77-57440
TO THE QUESTION OF SHORT-TERM FORECASTING OF ELECTRIC LOADINGS WITH APPLICATION OF INDISTINCT NEURAL NETWORKS
Sakhno E.P. 1
Dyachenko R.A. 1
Reshetnyak M.G. 1
Kapustin K.Yu. 1
1 Kuban State Technological University
179 KB
The technique of short-term forecasting of electric loadings with use of model of an indistinct logical conclusion is presented we will to use conclusion model like Sugeno. The received technique of forecasting includes the following stages: creation of training selection of the electric loadings consisting of four entrance variables and one output variable, loading of training selection in the ANFIS editor, a task for each entrance variable on 3 linguistic terms, like function of accessory of entrance data (triangular), and like function of accessory for output given (linear), training of a hybrid network with level of a mistake 0 and quantity of cycles of training equal 60, check of adequacy of the constructed indistinct model of a hybrid network. The created indistinct neural network allows to receive the short-term forecast of loading with an accuracy of 0,003.