大型离心压缩机叶轮流致振动机理与智能诊断

Faced with the challenges of deterioration of flow induced vibration, and fault identification for large centrifugal compressor impellers, this project is aimed to investigate physics characteristics of unsteady flow field, time-spatial evolution and characterization methods of aerodynamic loading acting on the impeller blades under different operating conditions. These will be achieved by geometrical model establishment, numerical simulations, and experimental tests. Analysis methods and a calculation software platform of high-efficiency impeller frequency random vibration response will be constructed. Transformation law for unsteady flow excitation mechanism and random vibration response with flow characteristics are investigated. Surface pressure fluctuations on high rotating blades, and in situ blade vibration measurement are investigated under the conditions of unsteady flow. These will contribute to theory analysis and simulation result verification. This project expounds the fatigue failure mechanism of impeller under unsteady flow excitation, and puts forward the method of safety checking for impeller based on unsteady excitation. Feature extraction method of weak signal characterizing blade crack are investigated. Sparse matrix representation method based on multi-Source information monitoring will be developed. In the end, an intelligent diagnosis method for impeller fault based on deep learning will be constructed. Unsteady flow characteristic, impeller dynamic analysis and design under unsteady flow excitation, the advanced measurement and experimental methods, and intelligent fault diagnosis methods for impeller can be achieved for large centrifugal compressor impellers through the study of this project. The results obtained can provide effective support for impeller design with high-performance and high reliability. It can also be applied on turbomachinery.

针对大型离心压缩机的流致振动导致叶轮疲劳失效问题，及叶轮故障准确识别的紧迫需求，通过建模、仿真和试验测试，研究不同工况下非定常流场的物理特征与作用于叶片表面的流动载荷演化规律及表征方法；建立高效叶轮频域随机振动响应分析方法及计算平台，研究非定常流对叶轮的激振机理及振动响应随流动特性的变化规律；研究非定常流状态下高速旋转的叶片表面压力脉动、叶片振动的原位测试方法，对理论分析和仿真结果进行验证。阐明非定常流动激励下的叶轮疲劳失效机理，提出基于非定常流激励下的叶轮安全校核方法。研究表征叶轮故障的微弱信号特征提取方法，发展基于多源信息融合的稀疏矩阵表征方法，建立基于深度学习的叶轮故障智能诊断方法。本项目将获得大型离心式压缩机叶轮的非定常流动特性、考虑非定常激励的叶轮动力学分析与校核方法、先进测试方法、叶轮故障的智能诊断等理论与方法，为其高安全可靠运行提供技术与理论支撑，并可在叶轮机械中推广应用。

大型离心压缩机作为天然气运输、石油化工以及航空航天等重要领域的核心动力装备，为国家能源提供自主保障，为持续推进绿色低碳发展和双碳行动的强有力支撑。随着装置的大型化以及非设计工况运行需求，叶轮的工作条件日益严苛，常处于非定常流体激振环境和高应力状态，极易诱发叶轮叶片的异常振动，进而面临结构疲劳损伤和失效破环的问题。为保证高端装备的长效运行，本项目从理论建模、仿真和试验测试出发，对叶片振动的流体激振原理、振动响应计算、流致振动破坏机理、安全校核、以及故障叶轮的状态识别与智能诊断等方面开展系统深入的研究。研究了不同干扰源下压缩机内部的非定常流动特性，涉及导叶开度及扩压器类型对压缩机内部流动的影响机理，进一步对失速工况下的流场开展全通道数值模拟，提出气动载荷POD降维降阶提取方法，明晰叶片振动的流体激振原理。针对先进测试方法开展了大量研究，提出了基于改进叶尖定时技术的叶片振动监测方法、基于叶顶间隙的振动监测系统，以及基于测点位置追踪的旋转叶片五维度振动测量方法，搭建了压缩机叶片流固耦合试验测试系统，开展实验测试研究工作。针对叶轮流致振动疲劳失效机理开展深入研究，从试验测试数据出发，分析流体激振下的同步和非同步振动特性，重点揭示了失速边界变转速运行下叶轮流固耦合特征、气动稳定极限边界叶轮流固耦合特性以及旋转失速参数表征与识别定位，进一步将旋转失速诱发的非同步振动纳入安全校核，提出了非定常激励下的叶轮安全校核新方法。针对故障叶轮与工业失谐叶轮，开展叶轮运行状态的特征参数提取与智能诊断研究，研究了基于基本失谐模型的动应力反演方法。实现少监测点下叶轮运行状态的参数表征，提出了工业失谐叶轮叶片振动测量与数值评价方法、基于叶尖定时传感器随机布局的同步共振频率识别方法、基于失谐识别的叶盘孪生模型构建方法，进一步研究含裂纹叶片的故障识别方法，提出了基于叶尖定时测振系统的叶片裂纹故障检测方法。结合大型压缩机的信号特征提取方法，以神经网络模型为基础，采用无监督特征学习与有监督状态识别相结合的方式，开发了大型压缩机叶轮故障智能诊断方法。相关研究结论为高性能离心压缩机自主设计和安全长效运行提供理论与技术支撑，具有重要的工程实际应用价值。