RESEARCH METHOD FOR PERIODICAL PROCESS MODEL PARAMETERS IN NEURAL NETWORKS

摘要

The research method for periodical process model parameters in neural networks consists in directing sequentially argument numerical values to a neural network input and a function of any process, thereafter neural network learning is implemented and investigated parameters of received a process model. A signal integrated function is supplied to a neural network input besides argument signal and function of investigated process received during tests of a function argument value and integrated quantity of this function is directed to the second neural network input.