In this paper, a finite-time neural funnel control(FTNFC) scheme is proposed for motor servo systems with unknown input constraint. To deal with the non-smooth input saturation constraint problem, a smooth non-affine ...In this paper, a finite-time neural funnel control(FTNFC) scheme is proposed for motor servo systems with unknown input constraint. To deal with the non-smooth input saturation constraint problem, a smooth non-affine function of the control input signal is employed to approximate the saturation constraint, which is further transformed into an affine form according to the mean-value theorem. A fast terminal sliding mode manifold is constructed by using a novel funnel error variable to force the tracking error falling into a prescribe boundary within a finite time. Then, a simple sigmoid neural network is utilized to approximate the unknown system nonlinearity including the saturation.Different from the prescribed performance control(PPC), the proposed finite-time neural funnel control avoids using the inverse transformed function in the controller design, and could guarantee the prescribed tracking performance without knowing the saturation bounds in prior. The effectiveness and superior performance of the proposed method are verified by comparative simulation results.展开更多
基金supported by the National Natural Science Foundation of China under Grant Nos.61403343 and 61433003Zhejiang Provincial Natural Science Foundation of China under Grant No.Y17F030063the China Postdoctoral Science Foundation Funded Project under Grant No.2015M580521
文摘In this paper, a finite-time neural funnel control(FTNFC) scheme is proposed for motor servo systems with unknown input constraint. To deal with the non-smooth input saturation constraint problem, a smooth non-affine function of the control input signal is employed to approximate the saturation constraint, which is further transformed into an affine form according to the mean-value theorem. A fast terminal sliding mode manifold is constructed by using a novel funnel error variable to force the tracking error falling into a prescribe boundary within a finite time. Then, a simple sigmoid neural network is utilized to approximate the unknown system nonlinearity including the saturation.Different from the prescribed performance control(PPC), the proposed finite-time neural funnel control avoids using the inverse transformed function in the controller design, and could guarantee the prescribed tracking performance without knowing the saturation bounds in prior. The effectiveness and superior performance of the proposed method are verified by comparative simulation results.
