Ultra-reliable low latency communication (URLLC) is a critical use case in 5G networks, particularly in the context of wireless industrial IoT networks. This paper delves into the realm of radio resource management (RRM) techniques with the primary objective of minimizing the age-of-information, a crucial metric for real-time applications. We propose a comprehensive framework that addresses the joint optimization of scheduling, power control, and rate adaptation to enhance data freshness. To achieve this, we introduce a novel, low-complexity algorithm that leverages Lyapunov optimization and deep reinforcement learning, explicitly taking into consideration factors such as reliability, latency, and the dynamic nature of traffic patterns. Through extensive simulations, we demonstrate that our proposed RRM scheme significantly outperforms conventional approaches, reducing the peak age-of-information by more than 2 times. These findings underscore the profound significance of age-aware radio resource optimization for applications in industrial wireless control, where stringent demands for reliability and low latency are paramount.