The predictive maintenance is a specific type of condition-based maintenance (CBM). Generally speaking, condition-based maintenance strategy continuously monitors various parameters or health indicators of an asset during its normal operation, detects conditional changes that may indicate a developing fault, diagnoses the root cause, then monitors the asset more closely while schedules a repair or correction in order to avoid the consequences of failure. Compare to the preventive maintenance, CBM prevents catastrophic breakdowns, reduces unplanned downtime, and optimizes maintenance schedule and spare parts inventory.
In contrast to diagnostics which is a reactive task, condition monitoring is about being proactive: measuring, trending, and assessing the machine health indicators continuously over a period of time, mid-term or long-term depending on the applications. For rotating machines, vibration is especially effective in diagnosing faults, and one of the most developed standards for vibration evaluation of machinery is ISO 10816, which provides allowable vibrations and alarm conditions for various machineries based on statistical analysis of historical data. When an alarm is triggered, notifications are sent automatically for rapid countermeasures. The pre-trigger time signal can be recorded thanks to the Time to Stop feature, and it is often post-processed to extract the wealth of information it contains, especially right before the event of interest occurs.