Predefined‐Time Adaptive Control for Nonlinear Multi‐Agent Systems With Prescribed Performance via Command Filtering and Dynamic Event Triggering

Abstract

This paper investigates three fundamental challenges in cooperative control of nonlinear multi‐agent systems, namely, unknown and coupled nonlinear dynamics, initial‐condition sensitivity of transient performance inherent in classical prescribed performance control, and limited communication resources in networked environments. To simultaneously address these issues, a unified adaptive neural control framework is developed that synergistically combines predefined‐time stabilization, fixed‐time prescribed performance (FTPP) constraints, command‐filtered backstepping, and a dynamic event‐triggering mechanism (DETM). A piecewise fixed‐time performance function, in conjunction with a barrier transformation, is constructed to convert the tracking error into a transformed variable independent of initial conditions, thereby guaranteeing that both transient and steady‐state behaviors rigorously remain within the pre‐specified boundaries for arbitrary bounded initial errors. Radial basis function neural networks are employed to approximate the uncertain system dynamics, while command‐filtered backstepping is adopted to eliminate the complexity explosion problem that plagues conventional recursive designs. The virtual controllers and adaptive update laws are synthesized within the framework of predefined‐time stability theory, ensuring practical stabilization of the closed‐loop system within a designer‐prescribed settling time that is entirely independent of initial states. A distinctive two‐timescale decoupling is achieved, in which governs the convergence deadline for stability, while the fixed‐time performance parameter (with ) independently prescribes the transient response envelope, thereby affording flexible and decoupled tuning of stability and performance specifications. Moreover, the proposed DETM dynamically adjusts the triggering threshold using auxiliary states, markedly curtailing the controller update frequency and communication overhead, while rigorously precluding Zeno behavior via theoretical analysis. Numerical simulations substantiate the superiority of the proposed methodology in terms of convergence rapidity, tracking precision, and communication efficiency.