Stochastic configuration networks for adaptive inverse dynamics modeling

Our previous works have shown that an enhanced plant Jacobian is helpful to improve the control performance of the inverse dynamic neural controller. This paper further studies this control scheme by using stochastic configuration networks (SCNs) and Savitzky–Golay (SG) filter, which can be used to produce higher quality of Jacobian teaching signals. It is observed SCN-based modeling techniques with reduced noise of estimated Jacobian can make the tracking performance favorably. Convergence and stability analysis of the closed-loop system are given. Comprehensive simulations are carried out, and results clearly demonstrate the effectiveness of our proposed method.