天文望远镜智能控制技术研究

Small and medium aperture ground-based optical telescopes, particularly 1-meter class telescopes,still play an important role in many fields of astronomical observations, especially in the study of time domain astronomy. The observations of time-domain astronomy raises new requests to the geographical distribution, sky coverage, response speed, and duration of observational time of instruments, as well as the ability of coordination among different instruments. Through our previous research, we have preliminarily achieved the ability to observe with centralized-control system and unattended operation. This project proposes to introduce the artificial intelligence to astronomical observations. On the basis of the centralized control of multiple telescopes, combining the information of observational requests, observational plans, logs of instruments, environmental parameters, etc., we will carry out the research on the intelligentized customization and the real-time adjustment of the observational strategy of both single telescope and united/cooperated multiple telescopes in this project, aiming to progressively achieve intelligent autonomous observations. Moreover, based on our experience of years of operation and maintenance of telescopes and the huge amount of historical data, we will collate and establish a fault tree of astronomical observational system, and break through the key technique of intelligent fault diagnosis through machine learning. Finally, relying on seven telescopes, this project will complete the intelligentized control system under the unified interface standard and operation interface, improving the observational efficiency and quality, as well as the capability and response speed of the time-domain astronomical research.

中小口径的地基光学望远镜，特别是1米级望远镜仍然在天文观测的很多领域,尤其是时域天文学研究中发挥着重要作用。时域天文学对设备的地理分布、天区覆盖、响应速度、连续观测时间和设备间的协同能力都提出了新的要求。通过前期研究，兴隆基地的部分望远镜已经初步具备了集中控制、无人值守的观测能力。本项目拟引入人工智能解决方案，在多台望远镜集中控制的基础上，综合观测需求、观测计划、设备日志、环境参数等大量信息，开展单台望远镜观测策略和多台望远镜联合/协同观测策略的智能化制定和实时调整技术研究，逐步实现智能自主观测。同时，基于多年的望远镜运行维护经验和海量历史数据，梳理并建立天文观测系统的故障树，通过机器学习突破故障智能诊断的关键技术。最终，本项目将依托7台望远镜完成统一接口标准和操作界面下的智能控制系统，提高望远镜的观测效率和观测质量，提高时域天文学研究的观测效能和响应速度。

时域天文观测需求的蓬勃发展对望远镜的观测效率、响应速度以及协同能力提出了更高的要求。本项目通过引入人工智能解决方案，在多台望远镜集中控制技术的基础上，结合综合观测需求、观测计划、设备日志、环境参数等信息，基于兴隆基地50cm、60cm、80cm、85cm、1m、1.26m和2.16m七台望远镜开展了以下研究：.（1）开展了单台望远镜观测策略和多台望远镜联合/协同观测策略的智能化制定和实时调整技术的研究，初步实现了智能化自主观测；.（2）将智能诊断技术应用于整个天文观测过程，实现了望远镜的健康状态管理以及无人值守情况下的故障预警与快速定位；.（3）根据实际观测对探测器的性能需求建立了一套天文光学探测器性能自动检测系统，完成了机下、机上测试和实测验证；.（4）开发了基于ZeroMQ和Protobuf的新型望远镜控制框架RACS2；建立了基于RACS2框架的望远镜通用控制平台，在兴隆观测基地60cm和80cm望远镜上通过实验并完成了在兴隆基地50cm、60cm、80cm、85cm、1m、1.26m和2.16m七台望远镜上的布控；.（5）基于不同科学目标的观测需求提出了智能天文观测策略制定与望远镜协同观测控制的方法。开展了结合观测实际执行情况与环境参数，智能化动态实时调整观测队列与观测参数设置的技术研究，并完成了跨越兴隆观测站和南山观测站的多台望远镜的协同观测与观测调度策略的测试；.（6）开展了实时天文定位与光度测量技术的研究，完成了一套具备实时天文定位以及光度测量功能的软件。