基于样本分位数的机载燃油泵故障状态特征提取及实验研究

发布时间：2018-04-23 19:07

本文选题：燃油泵 + 样本分位数　；参考：《航空学报》2016年09期

【摘要】：机载燃油泵的健康状态关系着飞行任务的完成和飞行安全,对机载燃油泵的故障状态特征提取及诊断成为亟需解决的问题。通过对机载燃油实验系统的振动与压力信号进行综合分析,提出了一种基于样本分位数的故障状态特征提取方法。首先,根据样本分位数的渐近分布定理,讨论了样本分位数的统计特性,分析了故障状态与样本分位数的对应关系,从理论上保证了该方法的可行性,在实测数据统计分析的基础上,讨论了样本容量对样本分位数稳定性的影响;其次,根据样本分位数渐近分布定理计算各故障状态的置信区间,并与Bootstrap方法得到的置信区间进行对比,结果显示,依据样本分位数渐近分布定理得到的置信区间真实可靠,为在线故障诊断提供了依据;然后,以各故障状态下提取的样本分位数为特征向量构建贝叶斯判别函数,进行故障诊断;最后依据故障诊断的正确率对传感器进行优化,结果表明,同时安装振动传感器与压力传感器可以提高故障诊断的正确率,并且只安装1个压力传感器与1个特定方向的振动传感器即可对机载燃油泵的故障状态进行完全识别。为快速准确的在线判断机载燃油泵的状态提供了理论支撑,并且可以降低工程应用中机载燃油泵监测系统的体积、功耗及复杂性。
[Abstract]:The health state of the airborne fuel pump is related to the completion of the flight mission and flight safety. It is urgent to solve the problem of extracting and diagnosing the fault state characteristics of the airborne fuel pump. A fault state feature extraction method based on sample quantiles is proposed by synthetically analyzing the vibration and pressure signals of the airborne fuel oil experimental system. Firstly, according to the asymptotic distribution theorem of sample quantiles, the statistical characteristics of sample quantiles are discussed, and the corresponding relationship between fault state and sample quantiles is analyzed. The feasibility of this method is guaranteed theoretically. Based on the statistical analysis of measured data, the influence of sample size on the stability of sample quantiles is discussed. Secondly, the confidence intervals of each fault state are calculated according to the asymptotic distribution theorem of sample quantiles. Compared with the confidence interval obtained by Bootstrap method, the results show that the confidence interval obtained by the asymptotic distribution theorem of sample quantiles is true and reliable, which provides the basis for on-line fault diagnosis. The Bayesian discriminant function is constructed with the sample quantiles extracted from each fault state as the eigenvector to carry out fault diagnosis. Finally, the sensor is optimized according to the correct rate of fault diagnosis, and the results show that, The accuracy of fault diagnosis can be improved by installing vibration sensor and pressure sensor at the same time, and only one pressure sensor and one vibration sensor in specific direction can fully identify the fault state of airborne fuel pump. It provides a theoretical support for fast and accurate on-line judging of the status of airborne fuel pump, and can reduce the volume, power consumption and complexity of the monitoring system of airborne fuel pump in engineering application.
【作者单位】：空军工程大学航空航天工程学院;鲁东大学数学与统计科学学院;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;
【分类号】：V263.6

【相似文献】