In this study, dynamical neural networks (DNNs) are used as an online identifier for a class of non‐linear systems with unknown actuator faults. By integrating the designed non‐linear FD observer with adaptive regulation algorithm, the parameter coupling problem in DNNs can be successfully solved and the unknown actuator fault can be rejected simultaneously. Based on Lyapunov theory and convex optimisation algorithm, both the observation error and the identification error can be proved to convergent to zero. Finally, simulation examples for the non‐linear systems with actuator fault are given to illustrate the effectiveness of the proposed approach.