摘要：An inter-agent learning adaptive control framework is proposed for mass production by using multi-agent system approach to enhance the convergence performance over the single agent control. The idea is to invoke agent-wise differences on estimated parameters of the online estimator in adaptive control. Each agent’s estimator selects estimated parameters from a corresponding ‘best’ agent among them for next iteration according to ‘State Déjà vu’ criterion mimicking psychological phenomenon of adopting experience from state similar to current one. The application of proposed framework on model free adaptive control shows to have robust convergence, good stability, and effective performances enhancement.
