航空发动机故障诊断算法性能分析及参数优化

摘要

基于航空发动机故障诊断测试平台,通过蒙特卡洛仿真得到涡扇发动机部件、传感器及执行机构典型故障数据,经过参数换算、趋势监测、异常检测等过程进行故障诊断,采用加权最小二乘法实现故障隔离.利用故障诊断测试平台中的评估指标准则对诊断算法性能做出定量评估,采用遗传算法对虚警率、漏报率和Kappa系数等关键参数进行优化.%Based on Aero-engine Fault Diagnosis Test Rig,fault data of turbofan engine components,sensors and actuators were obtained by Monte Carlo simulation,and the fault diagnosis process included parameter conversion,trend monitoring and anomaly detection.A weighted least squares method was used to realize the fault isolation.The performance of the algorithm was evaluated quantitatively using the evaluation criterion of the fault diagnosis test platform.The key parameters such as false alarm rate,false negative rate and Kappa coefficient were optimized utilizing genetic algorithm.