Prognostics and Health Management (PHM) is an emerging discipline aiming to exploit a continuous estimation of a system Remaining Useful Life (RUL) to optimize the planning of maintenance interventions. The RUL estimation is based on the detection of the current health status of the system, performed in the Fault Detection and Identification (FDI) phase of the prognostic analysis. The FDI task can be structured as an optimization problem, to minimize the error between an output measured from the system and one computed by a model, which accounts for multiple possible failure modes. Many different optimization algorithms are available in literature, but none is universally suitable for all problems. The choice of an acceptable optimization strategy is in fact strongly problem-dependent. In this paper, we focus on the FDI task for an Electromechanical Actuator (EMA), and we consider bio-inspired meta-heuristic algorithms. Those are characterized by a great robustness when the objective function is poorly known, although convergence is usually quite slow compared to deterministic gradient-based algorithms. We compare four different algorithms, namely Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Differential Evolution (DE) and Grey Wolf Optimization (GWO). Their applicability for our application and their performance, both in terms of computational time and accuracy, are investigated and compared to each other.