No abstract

Vibration analysis of rotating machinery can give an indication of possible faults, thus allowing maintenance before further damage occurs. Automating this analysis allows machinery to be run uattended for longer periods of time. This paper describes the use of neural networks as a method for automatically classifying the machine condition from the vibration time series. Several methods for the extraction of features to use as neural network inputs are described and compared. These methods are based upon measuring the zero lag higher-order statistics of the measured vibration time series. The time series for horizontal and vertical vibration signals are considered separately and combined to produce time series based upon the radius of the vibration displacement. The experimental set-up used for simulating unbalance and rub faults is described and classification success rates based upon each method reported. In particular, a classification success rate of over 99 per cent has been achieved.

We present the initial results from the FHPCA Supercomputer project at the University of Edinburgh. The project has successfully built a general-purpose 64 FPGA computer and ported to it three demonstration applications from the oil, medical and finance sectors. This paper describes in brief the machine itselfMaxwell -its hardware and software environment and presents very early benchmark results from runs of the demonstrators.

No abstract

Vibration analysis can give an indication of the condition of a rotating shaft highlighting potential faults such as unbalance and rubbing. Faults may however only occur intermittently and consequently to detect these requires continuous monitoring with real time analysis. This paper describes the use of artificial neural networks (ANNs) for classification of condition and compares these with other discriminant analysis methods. Moments calculated from time series are used as input features as they can be quickly computed from the measured data. Orthogonal vibrations are considered as a two-dimensional vector, the magnitude of which can be expressed as time series. Some simple signal processing operations are applied to the data to enhance the differences between signals and comparison is made with frequency domain analysis.

No abstract

No abstract

Model-based approaches to vibration monitoring can provide a means of detecting machine faults even if data are only available from the machine in its normal condition. The cyclostationary nature of rotating machine vibrations can be exploited by using periodic time-varying autoregressive models to model the signal better than time-invariant models. Experimental data collected from a small rotating machine set subjected to several bearing faults were used to compare time-varying and time-invariant models. Comparison is also made with a simple feature-based neural network fault diagnosis system.