1996 年 62 巻 2 号 p. 205-209
In the 1st report, the conceptual design and the sound information failure detection method for the plant inspection and diagnosis robot had been discussed. The inspection and diagnosis robot must monitor the condition of many plant machines at the same time. So it must not only inspect machine condition at prescribed inspection points, but also automatically move to a faulty machine after an abnormal state has been detected, in order to precisely identify faulty parts of the machine and failure types. This paper discusses (1)the calculation method of failure signal power from sound spectrum, (2)fuzzy navigation method for leading the inspection and diagnosis robot to the faulty machine by using the failure signal power. Furthermore, in order to verify whether the methods discussed in this report are effective, two examples of practical applications are shown.