基于知识图谱的航空发动机涡轮叶片故障自动检测系统设计

摘要:

涡轮叶片的故障特征涉及振动、温度、应力等多个方面,这些特征之间存在复杂的关联。现有检测系统难以深入分析这些特征之间的内在联系,从而影响故障诊断的准确性。为此,设计并开发航空发动机涡轮叶片故障自动检测系统。系统硬件部分包括传感模块、数据处理模块、主控模块以及数据传输模块等多个部分的设计。在硬件系统支持下,首先利用传感模块获取涡轮叶片的实时工作数据,并提取实际工作特征。然后构建的航空发动机涡轮叶片故障知识图谱。并将知识图谱中的标准故障特征与工作特征进行匹配计算,得出故障匹配度。最后,根据涡轮叶片的运行工况设定故障匹配度阈值,以此实现涡轮叶片故障的自动检测功能。实验结论表明:在正常、高温高压和高速旋转工况下,与传统检测系统相比,优化设计系统的故障类型误检率明显降低,振动幅值和表面温度故障参数检测误差分别减小0.4um和3.6℃。

Abstract:

The fault characteristics of turbine blades involve multiple aspects such as vibration, temperature, stress, etc., and there are complex correlations between these characteristics. The existing detection system is difficult to deeply analyze the inherent connections between these features, which affects the accuracy of fault diagnosis. To this end, design and develop an automatic detection system for turbine blade faults in aircraft engines. The hardware part of the system includes the design of multiple components such as sensing module, data processing module, main control module, and data transmission module. With the support of the hardware system, the sensing module is first used to obtain real-time working data of the turbine blades and extract actual working features. Then construct a knowledge graph of turbine blade faults in aircraft engines. And match the standard fault features in the knowledge graph with the working features to calculate the fault matching degree. Finally, a fault matching threshold is set based on the operating conditions of the turbine blades to achieve automatic detection of turbine blade faults. The experimental conclusion shows that under normal, high temperature, high pressure, and high-speed rotation conditions, compared with traditional detection systems, the optimized design system significantly reduces the false detection rate of fault types, and reduces the detection errors of vibration amplitude and surface temperature fault parameters by 0.4um and 3.6 ℃, respectively.