用于巡天及机会源观测的小口径望远镜智能化控制技术研究

Small aperture ground-based optical telescopes, especially small aperture telescope systems from 30cm to 60cm, play an important role in many fields of astronomical observations, especially in the study of sky survey and target of opportunity. This puts new requests to the response speed, continuous observation capability, remote observation capability and automatic observation capability of small aperture telescopes. Through the preliminary research, the photometric telescope and the 30cm telescope of the Xinglong Observation have preliminarily achieved the capability of remote observation. This project proposes to introduce the artificial intelligence to astronomical observations. Based on the preliminary work, combining information of observational requests, observation plans, equipment status, environmental parameters, etc., intelligent decision-making and the real-time adjustment of the observational strategy of small telescope for sky survey and target of opportunity will be studied, to realize the full automation of the observation process. At the same time, online monitoring and fault warning of equipment status will be realized through machine learning. Finally, relying on the photometric telescope and the 30cm telescope of the Xinglong Observation, this project will complete the artificial intelligence based automatic observation system under the unified interface standard and control mode to improve the observation efficiency of the small aperture telescope.

小口径地基光学望远镜，特别是30cm到60cm左右的小口径望远镜系统，在天文观测的很多领域，尤其是巡天及机会源观测过程中发挥着重要作用。这对小口径望远镜的响应速度、连续观测能力、远程观测能力、自动观测能力都提出了新的要求。通过前期研究，兴隆观测基地的测光辅助望远镜和30cm望远镜已经初步具备了远程观测能力。本项目拟引入人工智能解决方案，在远程控制的基础上，综合观测需求、观测计划、设备状态、环境参数等大量信息，针对进行巡天及机会源观测的小口径望远镜开展智能决策、智能化观测策略制定与调整技术研究，逐步实现观测过程的全自动化。同时，通过机器学习实现设备状态的在线监测。本项目将通过测光辅助望远镜和30cm望远镜完成相关实验及技术验证，最终实现统一接口标准和控制方式下的基于人工智能的自动观测系统，提高观测效率。

本项目针对由批量生产的天文设备组成的小口径望远镜系统的快速组建、快速投入实际观测这一需求展开研究。按照设备自动运行的要求，实现了设备通过网络接口的模块化控制。将基于机器学习的自主判断应用于小口径望远镜的观测过程中，辅助小口径望远镜的自主运行。重点研究了夜间云量的识别算法。为实现全天相机的天文校准，提出了粒子群优化算法和机器学习算法相结合的方法，提高了校准的速度和精度。同时采用机器学习算法完成了夜间云量的识别。实现了基于数据库的机会源触发机制，提高了在巡天任务中对机会源观测的响应速度。建立了多种状态监测机制，实现设备状态的在线监测。本项目的研究对同类系统的建设具有重要参考意义。