Adaptive NN Control for Nonlinear Multi-Agent Systems with Unknown Control Direction and Full State Constraints

Abstract

This paper proposes an adaptive full state constrained consensus control strategy for a class of non-affine multi-agent systems with partially unknown control directions. Such non-affine systems commonly appear in practical applications and are difficult to control, especially with unknown control directions. Neural network is an excellent approximation tool to solve unknown parameters in the aforementioned consensus systems. The method of one-to-one mapping and mean value theorem can eliminate coupling terms and guarantee that the states of each agent are not violated the predetermined time-varying dynamic constraint boundary. In this way, the original systems can be transformed into an equivalent unconstrained systems, and the transformed systems obtained has the same consensus with the original systems. Considering that directions of control are partially unknown, the Nussbaum function can solve this problem in a novel way. Then the stability of the systems can be proved by Lyapunov function and all signals in the closed-loop systems are bounded. In the end, the simulation demonstrates the effectiveness of the proposed method.