Artificial neural network for diagnosis of defects of hydro unit

Abstract

The hydrounit is very difficult dynamic hydroelectromechanical system and adequate mathematical description of dependence vibrating signal by all factors causing vibration, practically is impossible. Therefore in this work it is offered for the automated diagnosing of existing defects of the hydrounit to use three-layer artificial neurosimilar network, that will allow to consider the hydrounit as «black box». The structure of this network and its mathematical model is given. Some sections peak-frequency-hour of spectra of vibrating signals of the hydrounit are considered and on their basis the preliminary diagnostic conclusions are made.