燃气轮机气路故障多学科影响机理及平行诊断方法研究

The effects of gas path faults in gas turbines are complex and highly hazardous. It is a key issue throughout the full life cycle of the equipment and has significant impacts on its reliability, economy and safety. Gas path fault diagnosis consists of the establishment of the mapping relationship between fault modes and monitoring parameters, and the fault diagnosis based on the monitoring parameters. As the recent studies focused more on the latter one, and the mapping relationship between fault modes and monitoring parameters has been still in use since the 1970s and 1980s. It cannot demonstrate the effects of gas path faults accurately, leading to the diagnostic results to be simple. Besides, the new achievements of fault effects cannot be brought into this architecture, and the increasing multi-source monitoring parameters cannot be adopted in the fault diagnosis. Therefore, the purpose of this research is to investigate multi-disciplinary effect mechanism of gas path fault, explore multi-disciplinary effect rules deeply, and establish the novel fault knowledge base and mathematical description of fault effects.The parallel systems theory, rising in the complex system analysis field in recent years, will be applied in gas path fault diagnosis. The artificial system will be established to describe multi-disciplinary effects and simulate spatial-temporal evolution rule coupling between thermodynamics and dynamics. Finally, the parallel diagnostic platform for gas turbines will be established. It reveals the effect mechanism of gas path faults and explores the approach to diagnose faults of complex equipment based on the parallel system theory, so as to provide theoretical and technical support for gas turbine in performance testing methods, control strategies, operation and maintenance rules and so on.

燃气轮机的气路故障影响复杂且危害性高，是贯穿设备全寿命周期的关键问题，对其可靠性、经济性、安全性影响重大。气路故障诊断包括建立故障与监测参数的映射关系，通过监测参数估计故障两项任务。近年的研究多聚焦于后者，故障与监测参数的映射关系从上世纪七八十年代沿用至今，不能准确地表达气路故障的影响，导致诊断结果过于简单，新的故障影响成果不能被纳入该体系，日益增加的多尺度监测数据无法被用于故障诊断。据此，本课题拟探究气路故障的多学科影响机理，深入挖掘多学科影响规律，建立全新的故障知识库与故障影响的数学描述；将近年来兴起于复杂系统分析领域的平行系统理论应用于气路故障诊断，建立可描述多学科影响及模拟热/动力耦合时空演变规律的人工系统，搭建燃机平行诊断平台。通过该研究进一步揭示气路故障的影响机理，探索利用平行系统诊断复杂设备故障的途径，从而为燃气轮机在性能测试方法、控制策略、运维规则等方面提供理论和技术支撑。

由于不可避免与危害性高的特点，气路故障影响机理与诊断成为贯穿燃气轮机设计、制造、控制、运行、维护等多领域的关键问题。本项目基于平行系统理论，通过理论推导、数值分析和实验研究，揭示燃气轮机气路故障的多学科影响机理，提出一种故障定位准确、故障模式识别度高、诊断精度高、诊断速度快的平行诊断方法，为动力机械的控制与运行维护决策提供了全新的状态观测方法，同时也为基于热力学的诊断方法的发展提供了理论与实验基础。在本项目研究中，针对上百台各型燃气轮机（进口/国产、工业驱动/发电）7年的故障数据开展数据挖掘与特征提取研究，形成故障知识库；建立热力学问题数学描述和人工智能问题数学描述之间的纽带，打破气路故障诊断领域按测量参数特征分专业建立数学模型的局限，建立统一的故障模拟模型，更直观地研究气路故障的多学科影响；在实验室通过3D扫描与三维模型重构复现气路故障叶片的几何特征，研究各类气路故障引起的流场、压力分布、振动特性变化特征，采用几何特性代替中间健康参数，提高了故障观测的准确性与观测结果的实用性。本项目研究成果指导了我国天然气长输管道关键装备国产化研制及应用，项目成果集成开发的燃气轮机气路故障平行诊断系统作为衢州压气站1#和2#国产燃气轮机关键技术国产化研制与应用性试验过程的重要保障平台，配合完成了机组24小时机械运转测试、喘振确认测试、72小时性能测试、机组切换和负荷分配测试等测试项目，保障衢州压气站1#机组和2#机组累计安全平稳运行11456小时和6309小时。本项目为推进我国长输天然气管网核心设备国产化，保障国家能源安全，降低输送能耗做出了贡献。