Báo Cáo Identification Of The Control Objects Using Artification Neural Networks

Identification Of The Control Objects Using Artification Neural Networks

​ — This paper introduces the new ability of Artificial Neural Networks (ANNs) in defining the mathematical modern of the complicated control objects, in the context of missing the data base as well as performance function. The used Neural Networks is the Adaptive Neural Networks in which the activation function is the liner form - purelin, the late input; the Three layers feed-forward neural network with sigmoid activation functions in the input; and the Multi-input Adaptive Neural Networks with the hidden layers and purelin in the output layer. The training algorithm is Levenberg-Marquartd back propagation one. The result shows the advanced ability of the ANNs in defining of the mathematical model of non-liner control objects, especially the adaptive networks.