A systems-engineered intuitive adaptive failure prediction system

Current methods of failure prediction have an incomplete insight into the operation of monitored equipment. Using systems engineering to process initial requirements, a new method has been developed which aims to mimic a human approach to the examination of waveforms measured by condition monitoring equipment. This method has been designed to work with a minimum of prior training so that it is suitable for use on a large set of single-throw actuators, without a need for excessive run-in training. The method has been tested on data measured during fault simulations on a HW railway switch actuator, and the test results show that the system is capable of detecting the early onset of faults.