The integration of Digital Twins (DT) and Artificial Intelligence (AI) has significantly advanced predictive maintenance (PdM) in cyber-physical systems. However, existing frameworks primarily function as passive diagnostic tools, relying heavily on black-box machine learning models that often generate high false-positive rates due to a lack of physical domain knowledge. This paper introduces a novel Neuro-Symbolic Digital Twin with Deep Reinforcement Learning Control (NSDT-DRL) framework.