Abstract: With the explosive growth of mobile communication demands, large-scale and highly spatiotemporally dynamic scenarios such as smart campuses pose severe challenges to the capacity and coverage performance of wireless network planning. In such environments, base station (BS) site selection and antenna parameter configuration exhibit strong coupling and multi-scale characteristics, which necessitate coordinated optimization from a global perspective. However, conventional and deep learning methods often suffer from inefficient exploration and unstable convergence in large-scale, high-dimensional decision spaces, making it difficult to effectively address the joint optimization of BS locations and engineering parameters. To tackle these challenges, this paper proposes a Conditional Diffusion Model-enabled Hierarchical Reinforcement Learning (HRL-DiffNet) framework. Specifically, a conditional diffusion model learns high-quality deployment distributions to generate candidates, serving as structured priors to shrink the reinforcement learning (RL) exploration space. A hierarchical reinforcement learning architecture is developed, where the high-level policy is responsible for BS site selection, while the low-level policy performs fine-grained optimization of antenna parameters such as downtilt angles, thereby enabling multi-scale decision decoupling and coordinated optimization. In addition, a hybrid neural architecture that integrates convolutional neural networks and graph convolutional networks is designed to model complex topological dependencies and environmental features, providing conditional constraints for the diffusion model and serving as a unified state representation for reinforcement learning. This establishes an end-to-end coupled framework that jointly integrates topological representation, conditional diffusion-based generation, and optimization-driven decision making. Simulation results demonstrate that, in large-scale smart campus network planning tasks, HRL-DiffNet achieves significant performance improvements over existing state-of-the-art methods in terms of key performance metrics such as coverage probability and system throughput. Specifically, the proposed approach attains a network coverage rate of 99.1% and an average throughput of 1.766 Mbps, validating its effectiveness for fine-grained network planning in complex large-scale environments.