Использование нейронных сетей прямого распространения для прогнозирования технического состояния судовых дизель-генераторов

Abstract

The efficiency of use, together with the traditional methods of technical diagnostics of direct distribution neural networks to predict technical condition of marine diesel generators in operation. To model the neural network algorithms for predicting operational parameters of diesel generators proposed two-layer neural network. Neural network training was conducted two algorithms: gradient descent (traingdx) and Levenberg-Marquardt (trainglm)