Statistical finger-printing and prediction of plant vibration signals

The usefulness of vibration monitoring is widely recognised in the electricity supply industry. This paper discusses a system which statistically analyses vibration data online from power generating plant (turbogenerator) and presents the information in a way which assists the plant operator to identify possible plant faults and aids the `expert´ in quickly arriving at a fault diagnosis. One of the reasons why this system is particularly attractive arises from the fact that a common approach can be successfully adopted for a wide range of plant without the need for plant-specific mathematical modelling