| 陈学文,王以波,柯昌燕,何恩,邢楷.液冷系统多模态健康监测与动态评估方法[J].装备环境工程,2025,22(5):147-153. CHEN Xuewen,WANG Yibo,KE Changyan,HE En,XING Kai.Multi-modal Health Monitoring and Dynamic Evaluation Method for Liquid Cooling Systems[J].Equipment Environmental Engineering,2025,22(5):147-153. |
| 液冷系统多模态健康监测与动态评估方法 |
| Multi-modal Health Monitoring and Dynamic Evaluation Method for Liquid Cooling Systems |
| 投稿时间：2025-03-04 修订日期：2025-03-26 |
| DOI：10.7643/issn.1672-9242.2025.05.019 |
| 中文关键词: 液冷系统 健康管理 多模态传感 凝露检测 熵权法 动态评估中图分类号:TP277 文献标志码:A 文章编号:1672-9242(2025)05-0147-07 |
| 英文关键词:liquid cooling system health management multi-modal sensing condensation detection entropy weight method dynamic evaluation |
| 基金项目: |
|
|
| Author | Institution |
| CHEN Xuewen | 29th Research Institute of CETC, Chengdu 610036, China |
| WANG Yibo | 29th Research Institute of CETC, Chengdu 610036, China |
| KE Changyan | 29th Research Institute of CETC, Chengdu 610036, China |
| HE En | 29th Research Institute of CETC, Chengdu 610036, China |
| XING Kai | 29th Research Institute of CETC, Chengdu 610036, China |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 针对液冷电子设备长期运行中因凝露隐蔽性强、检测灵敏度低、安全风险高等问题，现有方法存在动态监测能力不足、多参数协同分析缺失等局限，提出一种液冷系统多模态健康监测与动态评估方法。方法 通过集成离子溅射镀膜液体传感器、压力传感器、压差传感器及PT1000铂电阻温度探头，构建分布式多物理场监测网络，实时追踪凝露蓄积量、流道压力变化、流道堵塞率及热交换效能。创新性建立模糊综合评价模型(HIS-LC)，采用熵权法动态分配特征权重ω1~ω4，实现系统0~100标度的健康状态量化评分。结果 凝露检测限达0.05 mL(置信度95%)，1 mL凝露检测灵敏度达98%，压力突变响应时间为2 s，流道压差大于1.5 MPa时，报警率为100%。经48 h连续运行验证，散热系统健康评分稳定维持≥98(标准差σ=0.8)，散热效率衰减率<0.8%/24 h。结论 本研究为高可靠性液冷系统从“故障后维修”向“预测性维护”转型提供了技术支撑。 |
| 英文摘要: |
| The work aims to propose a multi-modal health monitoring and dynamic evaluation method for liquid cooling systems to address the challenges of concealed condensation, low detection sensitivity, and high safety risks in liquid-cooled electronic equipment during long-term operation, where existing methods suffer from insufficient dynamic monitoring and lack of multi-parameter collaborative analysis. A distributed multi-physics monitoring network was constructed by integrating ion-sputtered liquid sensors, pressure sensors, differential pressure sensors, and PT1000 platinum resistance temperature probes, enabling real-time tracking of condensation accumulation, flow channel pressure changes, flow channel blockage rate, and heat exchange efficiency. A fuzzy comprehensive evaluation model (HIS-LC) was innovatively established, utilizing the entropy weight method to dynamically allocate feature weights (ω1–ω4) and quantify system health status on a 0-100 scale. Experimental results demonstrated that the threshold of condensation detection was 0.05 mL (95% confidence), the detection sensitivity of 1 mL condensation was 98%, the pressure transient response time was 2 s, and the alarm rate was 100% when the flow channel pressure difference was greater than 1.5 MPa. Continuous 48-hour operation tests confirmed the cooling system stable health scores≥98 (standard deviation σ=0.8) and a heat dissipation efficiency decay rate <0.8%/24 h. This research provides technical support for transitioning high-reliability liquid cooling systems from “post-failure maintenance” to “predictive maintenance”. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
