Data Filtering‐Based Maximum Likelihood Gradient Iterative Identification for Hammerstein Systems With Asymmetric Preload Nonlinearity

This paper focuses on the parameter identification problems of Hammerstein systems with asymmetric preload nonlinearity. The nonlinear part, which is described by a piecewise function, poses a challenge for identification. The input nonlinearity is represented as a linear parameter expression with unknown preload points and slopes by introducing a switch function. Then, the identification model of the system is obtained using the key item separation technique. Based on the identification model and the data filtering technique, a filtering‐based sliding‐data‐window maximum likelihood gradient iterative algorithm is derived. This algorithm effectively reduces the model complexity and improves the identification accuracy. In addition, a sliding‐data‐window maximum likelihood gradient iterative algorithm is derived for comparison. Finally, the feasibility and superiority of the proposed algorithms are demonstrated through the numerical simulation.