Stochastic nonlinear system identification based on HFC-GP

To identify structures and parameters of complex stochastic nonlinear systems with accuracy and efficiency, preventing premature convergence during the evolution, an improved multi-objective hierarchical fair competition (HFC) parallel genetic programming (GP) algorithm was employed. The improved HFC GP algorithm was used to identify an object system based on nonlinear autoregressive moving average with exogenous inputs (NARMAX)model, good identification results were achieved with simultaneous identification of both structures and parameters of the object system. In comparison with single population GP and traditional multi-population GP, HFC-GP showed a more competitive performance in preventing premature convergence. It can be concluded that HFC-GP is good at solving complex stochastic nonlinear system identification problems and is superior to other existing identification methods.