AREVA > AREVA Diagnostics > Loose Parts Monitoring > Signal Processing

KÜS - Signal Processing

The Principles
of Signal Processing

The signals from the maximum of 24 accelerometers in the 1 – 10 kHz frequency range are digitized at a pre-selected sampling rate of 50 or 100 kHz.

The short-time RMS value (time constant 5 ms) and the long-time RMS value (time constant selectable from 0.1 to 10 s) are specified for each channel.

All channels are monitored simultaneously by comparing the short-time RMS value with two thresholds that can be individually specified and adjusted for each channel: one threshold can be permanently set to a specified value, the other is achieved by multiplying the long-time RMS value by a selectable factor. When an event is detected, the signals from all the channels, including the previous history, are stored (recording time of 50 or 100 ms).

(For a better look at the graphic click on the picture or here.)

By selecting the short-time RMS value as a monitoring variable, it can be ensured that power spikes are not recorded as events.

The variable thresholds, based on the long-time RMS value, ensure continuous adaptation to the background noise level. An external blocking input blocks the transient recorder when deterministic events occur (e.g. movement of control rods). The diagnostic stage of KÜS also differentiates between known and unknown events. The known events are further subdivided into relevant and irrelevant events. Irrelevant events do not initiate a KÜS alarm.

The loose part impact events recorded are classified (i.e. compared with reference noise patterns) fully automatically. With the aid of a wide range of functions, the user can carry out further evaluations without the monitoring function being interrupted.