| 摘要 |
The present invention relates to an abnormal sound diagnostic device. In the past, the abnormal sound is determined by comparing the reference sound data of a normal sound and the sound data of a detected object device, however, because the characteristic change caused by the environment conditions or operation conditions of the device is not considered, the change of the sound data of the detected object device caused by the difference of the many conditions, is determined to be abnormal mistakenly. According to the sound data of the device synchronously obtained with the action of the detected object device, a sample sequence composed of time intensities is calculated, and the sample at learning is used as a reference sample sequence. According to a correction amount of a correction amount sequence estimated based on an object sample sequence and the reference sample diagnosis at diagnosis, at least one of the object sample sequence and the reference sample sequence is corrected; by comparing the corrected object sample sequence or the reference sample sequence with the corresponding the reference sample sequence or the object sample sequence, an abnormality is calculated; and by comparing the calculated abnormality with a threshold value, the abnormality is outputted and determined. |
