Intelligent models based on neural networks with long short-term memory for diagnostics of the state of machines in machine-building

Abstract

In paper we offer the architecture of intellectual models for the decision of problems of monitoring and diagnostics of machine tools, including in a mode of real. As the main components of intelligent models, it is proposed to use an ensemble of convolutional neural networks and recurrent neural networks with long short-term memory. Convolutional neural networks are used to derive local characteristics and data compression, processing information that is obtained directly from sensors. Recurrent neural networks with long short-term memory are used to encode time information, fixing long-term dependencies, taking the consistent nature of the data and revealing abstract characteristics.