<p>This paper investigates the bipartite consensus control problem for uncertain heterogeneous unmanned systems subject to communication link failures under directed and undirected signed topologies. A novel fully distributed globally adaptive neural network fault-tolerant control strategy is proposed, consisting of two main components: (i) a distributed adaptive fault-tolerant communication policy that compensates in real time for unknown and time-varying link weight perturbations induced by communication failures; (ii) a globally adaptive neural network control scheme that performs online estimation and compensation for system uncertainties arising from unmodeled dynamics and external disturbances. The proposed strategy is applicable to both directed and undirected signed graphs and can be implemented in a fully distributed manner without relying on global network topology information. The effectiveness and robustness of the control strategy are validated through two simulation case studies involving heterogeneous quadrotor formations under faulty communication scenarios.</p>