Efficient and adaptive monitoring of remote power transmission lines is crucial for grid reliability. This paper proposes a novel UAV-assisted monitoring solution leveraging Integrated Sensing and Communication (ISAC) to improve resource utilization and performance. We introduce a hierarchical, rule-guided Deep Reinforcement Learning (DRL) framework, specifically employing a Dueling Deep Q-Network (JTS-DQN), designed for this complex task. The JTS-DQN agent learns to dynamically co-optimize the UAV's flight trajectory and essential ISAC sensing parameters, notably sensing duration and power allocation, aiming to maximize the overall service energy efficiency. This optimization intelligently balances sensing quality, communication requirements, and energy expenditure based on real-time environmental conditions and operational needs. Through simulations, we demonstrate that the proposed JTS-DQN approach significantly enhances service energy efficiency compared to standard DRL and conventional optimization baselines, while simultaneously maintaining high sensing quality and ensuring robust constraint satisfaction. This work underscores the potential of DRL-driven ISAC for developing intelligent, autonomous monitoring systems for critical infrastructure.