Simulated Annealing Algorithm applied as a Fault Identification Method for Electromechanical Actuators affected by Multiple Failures

The approaches used in prognostics can be varied, all with the aim to recognize incipient failures. The considered case is one primary flight command electromechanical actuator (EMA), with multiple failures originating from progressive wear. The capability to anticipate incoming faults is a consequence of the possibility to identify early clues on monitored parameters. Early alerts can avoid the occurrence of the failure, triggering opportune out-of-schedule maintenance activities. This paper proposes a fault detection approach, based on the simulated annealing optimization algorithm, capable to identify symptoms of EMA degradation before the behavior becomes anomalous. In particular, the work puts a focus on the effects of multiple failures, analyzing the performance of this prognostic tool in three different fault configurations. The results originated from the model are validated through experimental data obtained from a test bench developed for the purpose; such comparison shows that the method is robust and has a high degree of confidence in the ability to identify faults before they occur, minimizing the possibility of false alarms or unexpected failures.