Hybrid adaptive feedforward cancellation against time-varying discrete systems with non-stationary random periodic disturbance

This study presents a novel hybrid adaptive feedforward cancellation (AFC) methodology in particular for time-varying discrete systems with stochastic periodic disturbance. The proposed hybrid AFC framework is composed of conventional AFC modules and auxiliary PID control variables. The former is traditionally constructed with sinusoidal functions and parameters of the AFC system, and the latter is added for compensating perturbation because of non-stationary random disturbance in practice. The authors derive an estimation algorithm of parameters in the hybrid AFC system by using a gradient descent based optimisation method. In addition, a theoretical investigation is conducted for verifying convergence property of the parameter estimation by means of a well-known Lyapunov stability theory. Lastly, the authors achieve numerical simulation to prove reliability and superiority of the proposed AFC algorithm by comparing to a conventional AFC approach.