Analiza porównawcza wybranych neuroregulatorów prędkości dla napędu z połączeniem sprężystym – opis modeli neuronowych

In the paper the adaptive control problems of the drive system with elastic joint using neural models trained on-line have been presented. Perceptron-type MLP neural networks and networks with radial basis functions RBF have been proposed, which were trained with Back-Propagation Algorithm based on gradient algorithms. The structures of neurocontrollers and analytical expressions for weights and centers adaptation of the suitable neurocontrollers’ structures have been presented.