DataInsight
Extracting Knowledge From Data
Home
About
Competencies
Projects
Patents and Talks
Research
Publications
Press
Consulting
Recent News
Remaining Lifetime Estimation
When a machine or other device fails unexpectedly, the costs can be tremendous. The very worst case would be a device such as the gear box of a helicopter, where failure could easily cause the death of everyone aboard. In an industrial context, loss of life is more rare, but unexpected device failure can result in tremendous financial costs. These costs do not only include the obvious ones: the labor and material cost to repair or replace the device. Depending upon the use of the device, the costs can extend to collateral damage to product or devices, lost business because the device is unavailable, loss of the use of other devices whose use depend in some way on the broken one, as well as lateness penalties and future business lost because of customer dissatisfaction.
We advise our clients on how to move from preventive maintenance to condition-based, or "predictive'' maintenance, i.e. servicing or replacement only when called for by the actual condition of the device in question. The figure displays a map of the "wear surface" of one of our former client's devices. Such a map enables the user to estimate the remaining lifetime of a device given its past use.
Pattern RecognitionAnomaly and Impending Nonlinear ProcessingData Mining Noise Sustained Patterns
Network Dynamics Remaining Lifetime More PublicationsStatistical Learning Multivariate Modeling