基于储备池计算的散裂中子源靶体拖动系统的健康状态监测研究

A target trailer is one of the most important systems in a spallation neutron source (SNS) which is a large-scale scientific facility. On the one hand, the trailer suffers from harsh working condition of being prone to failure, due to extreme radiation resulting from spallation reaction and intermittent operation with heavy load transmission. On the other hand, direct diagnosis with conventional in-situ sensors is limited by harsh monitoring condition of the radiation environment. Taking both harsh working and harsh monitoring conditions into consideration, this project proposes reservoir computing-based health monitoring techniques by analyzing power signals at the supply side which is outside the radiation environment of the SNS target trailer. By analyzing electromechanical failure characteristics of the motor and the planetary gearbox in the trailer, faulty models represented by power supply signals are accordingly developed. Through acquiring electrical current signals outside the radiation at constant speed and considering the faulty mechanism, reservoir computing-based signal filtering is presented for the fault diagnosis under short-term scale. Moreover, reservoir computing-based feature representation is developed for performance deterioration prognosis under long-term scale. The presented approaches are tested on the SNS target trailer. In this proposal, health monitoring techniques are proposed for electromechanical systems under both harsh working and harsh monitoring conditions. The addressed work is helpful in maintenance decision making for China SNS, the largest national key research infrastructure in China, and other systems under similar harsh conditions. The proposed research has therefore evident scientific significances and application values.

靶体拖动系统是散裂中子源大科学装置的关键组成部分，一方面散裂反应强辐射与间歇启停重载传动等苛刻运行工况使其容易发生故障，另一方面辐射环境的苛刻监测工况限制本地安装常规传感器对其直接诊断。本项目在运行与监测双重苛刻工况下，拟在辐射环境外的供电侧对能量波动进行储备池计算，以监测靶体拖动系统的健康状态。通过分析电动机驱动行星齿轮减速箱的机电传动损伤特性，建立拖动系统故障的供电侧电流信号表征模型；在辐射环境外的供电侧采集恒速电流信号并结合故障机理分析，研究短时尺度下基于储备池计算信号滤波的早期故障诊断方法，研究长期尺度下基于储备池计算特征表示的性能退化预示方法；将所提出的方法在散裂中子源靶体拖动系统上试验验证。本项目提出运行与监测双重苛刻工况下机电系统的健康监测方法，可以为我国迄今最大的国家重大科技基础设施即中国散裂中子源工程和同类苛刻工况的系统维护提供决策支持，具有显著的科学意义和应用价值。

散裂中子源是国际前沿的大型科学装置，其核心系统之一的靶体拖动系统工作在强辐射、微真空、辐射限制油润滑、间歇启停、一般数十吨重载拖动的苛刻运行工况下，容易产生故障。由于苛刻工作环境对状态监测的限制作用，损伤传感器又反过来降低了维护效率。本项目面向散裂中子源靶体拖车等在苛刻工况下的关键机电系统健康监测需求，通过健康监测提升散裂中子源重大科技基础设施关键系统的服役能力。. 为此，本项目在运行与监测双重苛刻工况下，通过能量波动的储备池计算，开展健康状态监测研究。通过分析电动机驱动行星齿轮减速箱的机电传动损伤特性，建立拖动系统故障的动态信号表征模型。通过信号采集并结合故障机理分析，研究短时尺度下基于储备池计算信号滤波的早期故障诊断方法，研究长期尺度下基于储备池计算特征表示的性能退化预示方法。将所提出的方法在散裂中子源工程上进行试验验证，所提出的运行与监测双重苛刻工况下机电系统健康监测方法，为我国迄今建成最大的国家重大科技基础设施和同类苛刻工况的系统维护提供决策支持。. 在该项目资助下，标注本项目编号资助发表期刊论文33篇，其中SCI论文30篇（包括SCI一区期刊论文17篇），2篇国内卓越期刊论文（机械工程学报、仪器仪表学报）均为EI收录；发表国际学术会议论文1篇；在科学出版社出版学术专著1部；授权发明专利5件，其中中国发明专利授权3件，美国发明专利授权2件；以第一完成人获得2020年度重庆市自然科学二等奖1项。项目负责人于2019年12月入选英国工程技术学会会士（IET Fellow）。. 通过该项目的研究，研发了散裂中子源机电装备健康监测系统样机，已经在中国散裂中子源一期工程水化系统进行安装并已稳定运行了半年。本项目是一个从重大科技基础设施工程中提炼的应用基础研究项目，反过来在大科学工程中得到工程样机应用，因而具有显著的科学意义和具体的应用价值。