Neuronová sít jako klasifikátor a predikátor zbytkové životnosti VN izolačních materiálů.

Neural Network with Radial Basis Functions as Residual Lifetime Predictor and Classifier of High Voltage Insulating Materials.

článek ve sborníku ve WoS nebo Scopus

en

This contribution deals with the RBF neural network with radial basis transfer function improvement for residual lifetime prediction and classification of the Relanex insulating material which is used for electric rotary machine winding. This neural network is used for prediction of the residual lifetime of insulating material for the future on the basis of measured values which were measured on the electric machine in previous times. The output from neural network in this case is always a numeric value of magnitude characterizing the state of residual lifetime. The RBF neural network can be used as a classifier, too, where, on the basis of measured input data, the neural network verbally determines the residual lifetime for example and that means that the neural network classifies its state. The classification by neural network means to determine the residual lifetime and to classify those values into the predetermined number of categories which characterize the total state of the insulating system of the electrical rotary machine winding. This contribution describes the input and output magnitudes used, it presents the chosen RBF neural network and then, by means of testing procedures, it checks-up its quality. The RBF neural network for prediction and classification was programmed by the Matlab 6.0 mathematic software

Článek se zabývá diagnostikou izolačního materiálu elektrických strojů.

This contribution deals with the RBF neural network with radial basis transfer function improvement for residual lifetime prediction and classification of the Relanex insulating material which is used for electric rotary machine winding. This neural network is used for prediction of the residual lifetime of insulating material for the future on the basis of measured values which were measured on the electric machine in previous times. The output from neural network in this case is always a numeric value of magnitude characterizing the state of residual lifetime. The RBF neural network can be used as a classifier, too, where, on the basis of measured input data, the neural network verbally determines the residual lifetime for example and that means that the neural network classifies its state. The classification by neural network means to determine the residual lifetime and to classify those values into the predetermined number of categories which characterize the total state of the insulating system of the electrical rotary machine winding. This contribution describes the input and output magnitudes used, it presents the chosen RBF neural network and then, by means of testing procedures, it checks-up its quality. The RBF neural network for prediction and classification was programmed by the Matlab 6.0 mathematic software

neuronová síť, životnost

Neural Network,Predictor

01.07.2004

Petrosani (Romania)

Petrosani

973-718-026-7

3rd International Conference, ICPE-CA 2004

140–143

5