| 摘要 |
PROBLEM TO BE SOLVED: To solve the problem in which although failure causes are classified by comparing operation data in machine failures with operation data of past cases, classification performance for failure causes depends upon the amount of operation data of the past cases, and operation data similar to the past cases are thereby needed to properly classify the operation data in machine failure occurrence.SOLUTION: A failure cause classification device 103 has an input unit which inputs operation data representing a state of a device, a storage unit which stores the operation data on the state of the device, and a classification unit which classifies the operation data input at the input part for each state of the device using the operation data stored in the storage part, and the classification has a region conversion part which converts operation data associated with a failure of a machine into region data according to learnt distribution regions, and a probability calculation unit which calculates a probability of classification into each failure cause according to a transition probability matrix of region data learnt for each failure cause.SELECTED DRAWING: Figure 1 |
