Infusing PID Tracking Control With Intelligence-Like Elements/Actions
针对多输入多输出非线性动态系统，提出一种嵌入惩罚、经验、预测等智能元素的PID跟踪控制策略，能在建模不确定、耦合和故障下实现渐近跟踪。
Developing structurally simple and functionally trustworthy control strategies for multi-input multi-output (MIMO) nonlinear dynamic systems has always been an interesting yet challenging research topic in the control community. In this note, we present a tracking control design approach embedded with the key intelligent elements/actions (IEs/ICs). More specifically, by properly exploiting and processing fundamental IEs/ICs, such as “penalty/punishment”, “experience/memory”, and “forecasting/prediction” often observed from and utilized in human decision making, we develop an interpretable PID-like control strategy capable of ensuring asymptotic tracking for nonaffine systems in the presence of modeling uncertainties, MIMO couplings, and unexpected actuation faults. The key design steps consist of analytically characterizing the fundamental IEs/ICs via certain mathematical representations, introducing generalized error, selecting and converting the related IEs/ICs into a signal carrying intelligence ingredients, and adaptively weighting such a signal to eventually produce the control action. The proposed framework of blending intelligence-like ingredients into control synthesis proves promising and is expected to stimulate interest in developing explainable IEs/ICs-driven control strategies for nonlinear dynamic systems.