Non-fragile l 2-l ∞ state estimation for time-delayed artificial neural networks: an adaptive event-triggered approach