针对深空探测任务由单探测器向多节点协同、由程控指令向自主智能演进的发展趋势，传统全向通信和固定路由模式已难以适应超长时延、大尺度多普勒、天体遮挡以及拓扑强时变等深空环境特征，因而在链路增益、邻居发现、资源调度和安全防护等方面均面临明显限制。定向天线在提高信噪比、提升网络容量、降低能耗以及增强抗截获能力方面具有优势，其与智能算法相结合的定向组网技术已成为应对深空复杂组网问题的重要研究方向。本文围绕深空探测场景下的智能定向自组网技术开展综述，系统梳理相关关键技术及研究进展。首先，分析深空定向传播特性、天线模型及其与全向天线的性能差异，并引入相关智能算法基础模型；其次，围绕定向网络中的时空同步、邻居发现和智能接入机制，重点综述智能算法在同步、发现与接入中的应用；在此基础上，进一步总结高动态拓扑维护、智能路由决策以及多维资源联合调度等技术，讨论人工智能与组网管控融合对网络鲁棒性和传输效率的提升作用；随后，面向深空定向组网的安全需求，综述认证鉴权、AI赋能网络安全等方面的研究进展，以提升网络生存能力；最后，结合我国深空探测任务及智能组网的发展现状，总结当前技术水平，并对未来研究方向进行展望。

With deep-space exploration evolving from single-probe missions toward multi-node coordination and from preprogrammed instructions to autonomous intelligence, traditional omnidirectional communication and fixed routing paradigms struggle to cope with harsh deep-space characteristics including ultra-long propagation delay, large-scale Doppler shift, celestial occultation and drastically time-varying topologies, imposing prominent constraints on link gain optimization, neighbor discovery, resource scheduling and security protection. Benefiting from superior performance in signal-to-noise ratio improvement, network capacity expansion, power consumption reduction and anti-interception capability enhancement, directional antennas combined with intelligent algorithms have emerged as a vital research solution to tackle complicated deep-space networking challenges. With deep-space exploration evolving from single-probe missions toward multi-node coordination and from preprogrammed instructions to autonomous intelligence, traditional omnidirectional communication and fixed routing paradigms struggle to cope with harsh deep-space characteristics including ultra-long propagation delay, large-scale Doppler shift, celestial occultation and drastically time-varying topologies, imposing prominent constraints on link gain optimization, neighbor discovery, resource scheduling and security protection. Benefiting from superior performance in signal-to-noise ratio improvement, network capacity expansion, power consumption reduction and anti-interception capability enhancement, directional antennas combined with intelligent algorithms have emerged as a vital research solution to tackle complicated deep-space networking challenges.This paper presents a comprehensive survey on intelligent directional ad hoc networking for deep-space exploration and systematically reviews key technologies and recent advances. It first analyzes the propagation features of directional deep-space links, antenna models and their performance gaps against omnidirectional counterparts, together with fundamental intelligent algorithm frameworks. Next, it elaborates on the deployment of intelligent algorithms for temporal-spatial synchronization, neighbor discovery and intelligent access in directional networks. On this basis, the work summarizes technologies covering highly dynamic topology maintenance, intelligent routing decision-making and multi-dimensional joint resource scheduling, and analyzes how the integration of artificial intelligence and network management improves network robustness and transmission efficiency. From the perspective of deep-space directional networking security, recent progress in authentication and AI-enabled network defense is reviewed to strengthen network survivability. Finally, grounded on China’s ongoing deep-space programs and intelligent networking developments, the paper concludes existing technical limitations and outlines prospective research directions.