The Filtering Based Maximum Likelihood Recursive Least Squares Parameter Estimation Algorithms for a Class of Nonlinear Stochastic Systems with Colored Noise

This paper focuses on the maximum likelihood estimation for bilinear systems in the presence of colored noise. The state variables in the model is eliminated and an input-output expression is provided. The input-output data of the system is filtered by an estimated noise transfer function, and the system is transformed into two subsystems. A filtering based maximum likelihood recursive least squares algorithm is proposed to strengthen the identification accuracy and improve computational efficiency. The superior performance of the developed methods are demonstrated by numerical simulations.

Automated summary

This paper focuses on maximum likelihood estimation for bilinear systems with colored noise. It eliminates the state variables in the model and provides an input-output expression. The input-output data of the system is filtered using an estimated noise transfer function, and the system is transformed into two subsystems. A filtering-based maximum likelihood recursive least squares algorithm is proposed to improve identification accuracy and computational efficiency. Numerical simulations demonstrate the superior performance of the developed methods.