A dynamic protection system in power systems should be able of predicting/preventing device failures such an transformers, capacitor banks, sectionalizers, reclosers, etc.
What kind of failures made by one of these can predicted in a modern protection system? And which kinds can be prevented? How?
Conventional protection devices are more reactive rather than pro-active. Their primary purpose is to monitor and isolate healthy equipment from any change from set value of voltage / current / frequency. With the exception of df/dt relay and power swing relay, I cannot think of other conventional protections to proactively isolate a system before it actually hits your equipment.
Modern power stations use dedicated monitoring systems such as online partial discharge monitoring system, online Dissolved Gas analysis system, etc., which are highly effective in predicting failures before they actually occur. The big difference between conventional and pro-active protections such as these, is that there is sufficient time for the operator to make a judicious decision to isolate the particular equipment before failure takes place. Dedicated software are in-built in PD monitoring systems which compare the PD activity with library patterns and alert the operator when there is an impending failure. To prevent false alarms, most of such systems wait for two consecutive patterns in a set period before giving an alarm. Likewise, in case of online DGA monitoring systems, the alarm limit of standard gases are pre-loaded in the vendors' software. Modern systems have a trending facility which help in monitoring the rate of rise of a particular parameter.
Hope this helps.
Most electrical failures can be eliminated by proactive measures. These include proper studies for coordiantion and protection, proper pm and testing of all devices at set periodicities and the utilization of predictive technologies such as UE, IR and PD monitoring of the Distribution network.
Consider the components of a power system which made up of different conducting, semi-conducting and insulating materials.
The occurrence of fault in the power system means that the insulators have failed, or the conductors can not pass their nominal current safely. As noted by Raju, insulation defects can be predicted by PD and DGA analysis. However, some of insulation defects are due to the conductors degrading and overheating. So techniques such as thermography, temperature rise measurement or analysing of emmited ultrasound spectrum of the electrical contacts also can be used for fault prediction.