基于LS-SVM的航空发动机喘振故障诊断研究

Study of the Surge Fault Diagnosis of an Aeroengine Based on the LS-SVM(Least Square-Supporting Vector Machine)

利用航空发动机健康状态的气路参数,建立最小二乘支持向量机(Least Squares Support Vector Machine,简称LS-SVM)回归模型,对航空发动机进行状态监控。根据模型监控低压压气机转速(N1)、压比(EPR)和燃油流量(FF)预测值与真实值的相对误差率来分析喘振故障,验证LS-SVM模型作为喘振故障诊断方法的可行性。结果表明,利用LS-SVM模型建立的航空发动机喘振故障模型,监控结果 N1、EPR和FF相对误差率分别达到9%、11%和29%,可以作为快速诊断喘振的依据。

By making use of the gas path parameters of an aeroengine in good health,established was a regressive model based on the least square supporting vector machine for monitoring the state of the aeroengine.The relative error rates between the predictive values and real ones of the rotating speed(N1),pressure ratio(EPR) and fuel oil flow rate(FF) of the low pressure compressor monitored by using the model were based to analyze the surge fault and verify the feasibility of the LS-SVM model as a method for diagnosing the surge fault.It has been found that the N1,EPR and FF relative error rates monitored by using the surge fault model for aeroengines based on the LS-SVM model can hit 9%,11% and 29% respectively,thus can be used as the basis for a quick diagnosis of a surge.