| 摘要 |
<p>PURPOSE:To properly monitor and diagnose the degrading of a sensor and parts whose properties change gradually with a passage of time by carrying out abnormality detection by fuzzy process data processing by which the process values sent from an abnormality detecting part are made fuzzy. CONSTITUTION:A sensor input part 7 sends out the water level of a nuclear reactor measured by a sensor part 6 as a process value 1 of the water level of the nuclear reactor. An abnormality detecting part 9 memorizes the value l and diagnoses whether it is abnormal or not and in the case it is normal, the signal is sent to a data processing part 10. If the water level of a nuclear reactor is out of the normal water level, the whole is defined as not necessarily normal by the fuzzy theory which can be applied to diagnose whether it is abnormal or not. Based on a membership function, the more the measured process value differs from the normal water level, the higher the possibility of the abnormality becomes and only when the measured process value reaches the low water level or the high water level, it is determined as complete abnormality. Further at the time of water level control near the low water level, the possibility obtained based on the membership function is integrated and at the time, for example, when the result of the integration exceeds 1, it is judged as control abnormality and reported and thus precise diagnosis and monitoring can be possible.</p> |
