Improved Identification of Hammerstein plant using a non-linear model trained with symbiotic organisms search

A Hammerstein plant represents characteristic of a practical nonlinear dynamic system. Identification of such complex plant finds numerous applications in accurate modeling of nonlinear controller, designing actuator, developing models for hysteresis analysis etc. Literature reveals the process to identify such a nonlinear dynamic plants is to develop a nonlinear neural network based model combined with other associated FIR and IIR structures. For effective learning of the weights nature inspired algorithms are potential candidates. In this manuscript the nonlinear neural model is based on Wavelet Neural Network (WNN). Cheng and Prayogo in 2014, proposed a new nature inspired algorithm Symbiotic Organisms Search (SOS). It is influenced by the living strategies and interaction among various organism to survive in an ecosystem. The goal is to survive in the environment while taking care about the benefit and harm caused by the other organisms. The SOS is used to simultaneously train the weights of the WNN, FIR and IIR blocks of the model. Simulation study reveals that the proposed SOS model has superior accuracy than the same model trained with Colliding Bodies Optimization (CBO).