Controlled Autoregressive Moving Average Systems Identification Using Salp Swarm Algorithm

Abstract

In this study, an advanced bio-inspired evolutionary technique named as salp swarm algorithm (SSA) is employed to determine the coefficients of the unknown controlled autoregressive moving average (CARMA) systems. SSA imitates the behavior of the salps navigation in the deep oceans. Moreover, it incorporates an efficient control parameter to acquire the global optimum results with greater convergence and low steady state error. The cost function for CARMA model estimation is constructed based on normalized mean squared sense. In this paper, three distinct most popular CARMA systems are identified by using SSA, biogeography-based optimization (BBO) and a classical optimizer called basic Kalman filter (KF). The experimental results show that SSA accurately identified the CARMA systems than other employed methods which have been proved by considering various performance metrics namely parameter estimation error, fitness percentage and mean squared deviation.