No abstract

Cluster analysis is widely used in the machine diagnostic and monitoring field. This article discusses the issue of recognizing the states of rotary-support systems with fluid-friction bearings. An experiment was carried out to investigate the effect of tightening the bolts that connect rotor-support unit body to the frame; to investigate the effect of tightening the bolts that connect electric motor to the frame; to investigate the rotor imbalance, as well as a combination of these factors. Cluster analysis based on the K-means method was applied. The readings of the eddy- current transducer were used as an input data for training. Analysis of the results revealed two groups of defects. During testing, the accuracy of group identification was 100%.

No abstract

The paper deals with the problem of monitoring the complicated hydromechanical system of a rotary machine with fluid-film bearings. The prospect of accurate recognition of changes in the lubrication system, which are expressed in the appearance of air bubbles in the lubricant, is investigated. The monitoring of the state is carried out by means of high-speed measurements of shaft vibrations, lubricant pressure supply and other parameters. Measurement data are transmitted for analysis to an artificial neural network to recognize the state of the system. The developed neural network has demonstrated high recognition accuracy of more than 98 %. Some recommendations on the measurement results processing and the neural network settings are represented in the paper.

One of the most important characteristics taken into account when designing machines is their efficiency. Active fluid-film bearings are able to influence tribological processes and reduce frictional energy, which increases the efficiency of the machine. For experimental determination of friction and verification of simulation results, an experimental complex was created, which makes it possible to directly measure the moment of friction force in active journal bearings. The complex is based on a rotor-bearing system with a rigid rotor on one active and one passive bearing. The information-measuring system includes a set of sensors and has the ability to control servo valves of active hybrid bearings. The system interface is based on virtual devices and allows both monitoring the state of the rotor-bearing system in real time and recording signals for their subsequent processing. The article contains a description of design and the principle of operation of the complex, including the features of processing various sensor signals, as well as a description of measures to improve measurement quality.