| 摘要 |
<p>PURPOSE:To save storage capacity of a memory for storing a normal and/or abnormal pattern table. CONSTITUTION:This method has a failure diagnosing mode in which a pattern of an input layer data made of binary data obtained from sensors S1-S4 for detecting condition of an object 2 to be monitored sample and stored at each specified cycle is compared with a normal and/or abnormal pattern registered in pattern tables 5a and 5b to perform a failure diagnosis and a learning mode in which a novel pattern is registered in a pattern table. In the learning mode, the novel pattern is compared with a normal and/or abnormal pattern registered and when one piece of data exists with non-coincidence between both the patterns, the non-coincidence data is rewritten to a specified common-oriented data so that both patterns are converted to one common oriented pattern to register the common-oriented pattern. When a new pattern is registered into a normal and/or abnormal table, the storage capacity of a memory can be saved by a portion corresponding to the common oriented pattern.</p> |
