基于条件对抗域自适应的燃气轮机跨工况气路故障诊断技术

近年来,基于模式识别的方法已大量应用于燃气轮机气路故障诊断,并取得一定效果。然而其故障识别准确率常局限于单一工况。为解决基于模式识别的气路故障诊断方案在多种工况下的故障识别准确率低下的问题,该文建立基于深度学习和条件对抗域自适应的模型,用于提取跨工况不变特征,进而提升模型对工况差异干扰的抗性,并保障模型在多种工况下的气路故障识别准确率。此外,该文设计多种跨工况故障诊断任务来验证所提出方案的有效性。

In recent years,the method based on pattern recognition has been widely used in gas turbine gas path fault diagnosis,and achieved certain results.However,the accuracy of fault identification is often limited to a single working condition.In order to solve the problem of low fault recognition accuracy of the gas path fault diagnosis scheme based on pattern recognition under various working conditions,this paper establishes a model based on deep learning and condition adaptive resistance domain,which is used to extract invariant features across working conditions,improve the resistance of the model to the interference of working conditions,and ensure the gas fault recognition accuracy of the model under various operating conditions.In addition,a variety of cross-mode fault diagnosis tasks are designed to verify the effectiveness of the proposed scheme.