| 摘要 |
<p>PURPOSE:To quickly and accurately perform the diagnosis of a fault by holding the alarm data to divide the data into groups and then estimating the candidate points for the factor of the fault to each group by means of an adverse searching method. CONSTITUTION:The measurement and non-measurement points included in a process 1 are shown by F1 - Fn and serve as the nodes forming a graph. An alarm data collection means 2 collects periodically the alarm data at each of those measurement and non-mesurement points of the process 1. Then the diagnostic processing is started when the difference is confirmed between the number of alarm data collected in the precedent time and that collected this time. A grouping means 3 divides those effective measurement points of the alarm data collected and held by the means 2 into groups. A fault factor estimating means 4 estimates the node having the coincidence between the number of effective points and the counted number of estimated codes in the same group as a candidate point for the factor of the fault to each group divided by the means 3 by means of an adverse searching method. These estimated candidate points are displayed on a diagnostic result display means 5.</p> |
