控制方向未知的不确定系统预设性能自适应神经网络反演控制

对一类控制方向未知的不确定严格反馈非线性系统的预设性能自适应神经网络反演控制问题进行了研究.系统中含有时变非匹配不确定项且控制方向未知.首先,提出了一种新的误差转化方法,放宽了对初始误差已知的限制;随后,利用径向基函数(radial basis function,RBF)神经网络及跟踪微分器分别实现了对未知函数和虚拟控制量导数的逼近,并综合运用Nussbaum函数和反演控制技术设计了控制器.所设计的控制器能保证系统内所有信号有界且输出误差满足预设的瞬态和稳态性能要求.最后的仿真研究验证了控制器设计方法的有效性.

Prescribed performance adaptive neural backstepping control for nonlinear system with uncertainties and unknown control directions

We investigate prescribed performance adaptive neural backstepping control for a class of strict-feedback nonlinear systems with time-varying uncertainties and unknown control directions. Firstly, a novel error transformation is proposed to eliminate the limitation that initial error must be known. Subsequently, radial basis function (RBF) neural networks and track differentiators are proposed to approximate unknown functions and derivatives of virtual controls respectively. At the same time, Nussbaum function and backstepping technique are combined to design the controller. The controller guarantees that all state variables are bounded and the prescribed transient and steady state error bounds are satisfied. Finally, the effectiveness of proposed scheme is validated by simulation research.