This article investigates the convergence analysis of networked iterative learning control for nonlinear nonaffine systems firstly by considering stochastic noise introduced by the network channels. The convergence analysis is under a data-driven framework, which does not rely on any mechanism model information. To deal with the nonlinearity, both the state transition technique and the differential mean value principle are used to formulate the iterative dynamics of system states, tracking errors and input signals using a lifted matrix expression, respectively. In terms of the contraction mapping principle, the tracking error is shown to be iteratively convergent under the sense of mathematical expectation. Since the $\lambda$-norm is not used in the analysis, the convergence property of the tracking error is not affected by the operation interval and a good transient performance can be ensured in theory. Simulation studies test the theoretical results.

本文首先通过考虑网络通道引入的随机噪声，研究非线性非仿射系统的网络化迭代学习控制的收敛性分析。收敛分析是在数据驱动的框架下进行的，不依赖于任何机制模型信息。为了处理非线性，状态转移技术和微分平均值原理都被用来分别使用提升矩阵表达式来制定系统状态、跟踪误差和输入信号的迭代动力学。从收缩映射原理来看，跟踪误差在数学期望的意义下是迭代收敛的。由于$\lambda$分析中不使用-范数，跟踪误差的收敛性不受操作区间的影响，理论上可以保证良好的瞬态性能。模拟研究检验了理论结果。