基于小波包BP_AdaBoost算法的机载燃油泵故障诊断研究
发布时间：2019-01-09 11:55
【摘要】：针对机载燃油泵故障数据少、诊断效率低、维护成本高、缺乏有效诊断方法的问题,搭建了机载燃油泵燃油转输系统实验平台,提出利用小波包分析进行特征提取和基于BP_AdaBoost机载燃油泵故障诊断方法。首先测量燃油泵7种典型状态模式所对应的振动信号和出口压力信号;然后在分析信号时频特性和统计特性的基础上,利用小波包分解提取振动信号不同频段能量值作为故障特征参数,结合振动信号峭度以及压力信号均值构造特征向量;最后利用特征向量训练和验证BP_AdaBoost分类模型。实验结果不仅优化了传感器,而且表明BP_Adaboost算法与SVM、BP算法相比,能够有效实现对机载燃油泵的故障诊断。
[Abstract]:Aiming at the problems of low fault data, low diagnostic efficiency, high maintenance cost and lack of effective diagnostic methods, an experimental platform for fuel transfer system of airborne fuel pump is built. A method of feature extraction based on wavelet packet analysis and fault diagnosis of airborne fuel pump based on BP_AdaBoost is proposed. First, the vibration signal and outlet pressure signal corresponding to 7 typical state modes of fuel pump are measured. Then on the basis of analyzing the time-frequency and statistical characteristics of the signal, the wavelet packet decomposition is used to extract the energy values of different frequency bands of the vibration signal as the fault characteristic parameter, and the eigenvector is constructed by combining the kurtosis of the vibration signal and the mean value of the pressure signal. Finally, the feature vector is used to train and verify the BP_AdaBoost classification model. The experimental results not only optimize the sensor, but also show that the BP_Adaboost algorithm can effectively realize the fault diagnosis of the airborne fuel pump compared with the SVM,BP algorithm.
【作者单位】： 空军工程大学航空航天工程学院;中航工业金城南京机电液压工程研究中心;航空机电系统综合航空科技重点实验室;
【基金】：航空科学基金(20142896022)项目资助
【分类号】：V267
，
本文编号：2405590
[Abstract]:Aiming at the problems of low fault data, low diagnostic efficiency, high maintenance cost and lack of effective diagnostic methods, an experimental platform for fuel transfer system of airborne fuel pump is built. A method of feature extraction based on wavelet packet analysis and fault diagnosis of airborne fuel pump based on BP_AdaBoost is proposed. First, the vibration signal and outlet pressure signal corresponding to 7 typical state modes of fuel pump are measured. Then on the basis of analyzing the time-frequency and statistical characteristics of the signal, the wavelet packet decomposition is used to extract the energy values of different frequency bands of the vibration signal as the fault characteristic parameter, and the eigenvector is constructed by combining the kurtosis of the vibration signal and the mean value of the pressure signal. Finally, the feature vector is used to train and verify the BP_AdaBoost classification model. The experimental results not only optimize the sensor, but also show that the BP_Adaboost algorithm can effectively realize the fault diagnosis of the airborne fuel pump compared with the SVM,BP algorithm.
【作者单位】： 空军工程大学航空航天工程学院;中航工业金城南京机电液压工程研究中心;航空机电系统综合航空科技重点实验室;
【基金】：航空科学基金(20142896022)项目资助
【分类号】：V267
，
本文编号：2405590
本文链接：https://www.wllwen.com/kejilunwen/hangkongsky/2405590.html
