Abstract
Spectral analysis as well as parametric identification techniques have been applied to data from boiling water reactor noise measurements to estimate the process dynamics. Three different parametric model structures were tried. Least squares and maximum likelihood methods were tried to estimate the parameters of the models. Low order models were obtained in most cases.
The results show the difficulties to identify the dynamics and the static gain from normal operating data when strong feedback and correlation between the inputs exist.
The results also show how sharp resonant peaks will be smoothed in long-time analysis probably due to time-varying parameters. The resonance frequency in average power range monitor (APRM) and main circulation flow can be well predicted by rather simple models.
The results show the difficulties to identify the dynamics and the static gain from normal operating data when strong feedback and correlation between the inputs exist.
The results also show how sharp resonant peaks will be smoothed in long-time analysis probably due to time-varying parameters. The resonance frequency in average power range monitor (APRM) and main circulation flow can be well predicted by rather simple models.
| Original language | English |
|---|---|
| Pages (from-to) | 77-88 |
| Journal | Journal of Nuclear Science and Technology |
| Volume | 16 |
| Issue number | 2 |
| DOIs | |
| Publication status | Published - 1979 |
Subject classification (UKÄ)
- Control Engineering