基于灰色模型KFMC的航天发动机故障诊断算法

摘要

针对以往的故障诊断方法往往基于已经采集的数据,无法对故障诊断进行及时预测,同时基于单一传感器的测量信息难以全面准确反映航天发动机工作状态,从而造成故障诊断的不确定和不精确的问题,提出了一种基于灰色模型KFMC数据融合的航天发动机故障诊断算法.采用有标签的训练数据初始化KFMC模型,将预测的数据采用KFMC模型估计其所属的故障类别以及隶属度,然后将数据对应各诊断类别的隶属度初始化传感器的初始信度分配,将各传感器采集的数据作为证据体,采用DS数据融合方法融合各证据体,获得最终的诊断结果.通过飞机发动机故障诊断实例进行实验,结果表明文中方法能正确及时地预测故障,克服了单个传感器故障诊断具有的不确定和不精确性,是一种适于航天发动机的故障诊断方法.%The previous fault diagnosis model is based on the collected data,not able to predict the fault in time,and the traditional single measuring information based on the single sensor is not reflecting the working state of aero engine,therefore,leading to the uncertain and inaccurate problem.Aiming at the problems,an aero engine fault diagnosis method is proposed based on gray model KFMC and DS data fusion.The samples with labels is used to initialize the KFMC model and the predicted data uses the KFMC model to estimate the classification and its attributing probability for all the sensor data.Then the atributing probability is used to initialize the initial believe assign,the data collected by sensors is used as the evidence and the DS fusion method is used to fusion all the evidence to get the result.Simulation experiment is implemented to predict the fault,which also shows it solves the problems such as diagnosis uncertainty and inaccuracy and it is a diagnosis method suiting for aero engine.