It is technically challenging to maintain stable tracking for multiple-input–multiple-output (MIMO) nonlinear systems with modeling uncertainties and actuation faults. The underlying problem becomes even more difficult if zero tracking error with guaranteed performance is pursued. In this work, by integrating filtered variables into the design process, we develop a neuroadaptive proportional-integral (PI) control with the following salient features: 1) the resultant control scheme is of the simple PI structure with analytical algorithms for auto-tuning its PI gains; 2) under a less conservative controllability condition, the proposed control is able to achieve asymptotic tracking with adjustable rate of convergence and bounded performance index collectively; 3) with simple modification, the strategy is applicable to square or nonsquare affine and nonaffine MIMO systems in the presence of unknown and time-varying control gain matrix; and 4) the proposed control is robust against nonvanishing uncertainties/disturbances, adaptive to unknown parameters and tolerant to actuation faults, with only one online updating parameter. The benefits and feasibility of the proposed control method are also confirmed by simulations.

中文翻译：

---

MIMO 系统具有有界性能指数的渐近跟踪控制：神经自适应容错比例积分解决方案


对具有建模不确定性和驱动故障的多输入多输出（MIMO）非线性系统保持稳定跟踪在技术上具有挑战性。如果追求性能保证的零跟踪误差，根本问题会变得更加困难。在这项工作中，通过将过滤变量集成到设计过程中，我们开发了一种神经自适应比例积分（PI）控制，具有以下显着特征：1）最终的控制方案是简单的 PI 结构，具有自动调整其分析算法的能力。 PI增益； 2）在不太保守的可控性条件下，所提出的控制能够实现具有可调收敛速度和有界性能指标的渐近跟踪； 3）通过简单的修改，该策略适用于存在未知且时变控制增益矩阵的方形或非方形仿射和非仿射MIMO系统； 4）所提出的控制对于非零不确定性/干扰具有鲁棒性，能够适应未知参数并且能够容忍驱动故障，并且仅需要一个在线更新参数。仿真也证实了所提出的控制方法的优点和可行性。