基于DBN的不均衡样本驱动民航发动机故障诊断

在结合深度置信网络(DBN)、采样与集成技术的基础上,提出了基于不均衡样本驱动的民航发动机故障诊断模型。该模型通过分析民航发动机历史飞行数据,利用DBN提取性能参数中的内部特征,利用采样技术将不均衡样本均衡化,采用集成技术进行故障分类。将该模型应用到CFM56-7B系列发动机历史飞行数据,实验结果表明:与常用故障诊断方法相比,该模型的准确率高达0.996,AUC值高达0.948,可以有效处理民航发动机样本高维、不均衡问题。      

数据缺失场合下k/N(G)系统可靠性指标的经验Bayes估计

在定数截尾缺失数据样本下,研究了不可修k/N(G)系统的可靠性指标的估计问题。将极大似然估计法和Bayes方法相结合得到了部件的平均寿命、系统可靠度及平均寿命等可靠性指标的经验Bayes估计,最后利用随机模拟例子说明了本文方法的正确性和可行性。     

指数分布基于多重定数截尾样本中异常数据的检验

讨论多重定数截尾指数型寿命数据,对同时存在异常大数据和异常小数据的情形给出了检验方法,得到了检验异常数据的判别标准,最后以一实例说明其应用.     

Gray bootstrap method for estimating frequency-varying random vibration signals with small samples

During environment testing, the estimation of random vibration signals （RVS） is an important technique for the airborne platform safety and reliability. However, the available meth- ods including extreme value envelope method （EVEM）, statistical tolerances method （STM） and improved statistical tolerance method （ISTM） require large samples and typical probability distri- bution. Moreover, the frequency-varying characteristic of RVS is usually not taken into account. Gray bootstrap method （GBM） is proposed to solve the problem of estimating frequency-varying RVS with small samples. Firstly, the estimated indexes are obtained including the estimated inter- val, the estimated uncertainty, the estimated value, the estimated error and estimated reliability. In addition, GBM is applied to estimating the single flight testing of certain aircraft. At last, in order to evaluate the estimated performance, GBM is compared with bootstrap method （BM） and gray method （GM） in testing analysis. The result shows that GBM has superiority for estimating dynamic signals with small samples and estimated reliability is proved to be 100% at the given confidence level.     

基于任意两点试验信息的寿命分散系数法

在试验数据缺失无法获得中位寿命时,为避免偶然因素导致个别试验结果偏低而造成技术寿命偏低,浪费寿命的情况,提出了基于任意两点试验信息的寿命分散系数法.推导了基于对数正态分布小子样任意两点试验信息的寿命分散系数计算公式,计算了给定置信度下相应的寿命分散系数.算例表明:利用提出的方法及相应寿命分散系数,得到的技术寿命与基于中位寿命获得的技术寿命结果接近,方法合理.最后给出了该方法的适用范围.      

基于大变形蠕变分析的持久寿命预测方法

提出一种基于真应力真应变弹塑性蠕变本构模型和大变形有限元分析的高温构件持久寿命预测方法.该方法利用以真应力-真应变表示的材料高温拉伸应力-应变曲线建立材料的弹塑性模型,基于蠕变曲线建立蠕变本构模型,并采用大变形有限元方法计算高温构件在给定载荷下的变形响应曲线,根据其响应曲线的变化趋势来确定构件持久寿命.通过TC11钛合金缺口试件500℃下的持久试验对上述方法进行验证,并与三种基于小变形分析的持久寿命预测方法进行对比.结果表明:本工作提出的方法可以较准确地预测TC11缺口试件的高温蠕变响应和持久寿命,其预测精度优于基于关键点断裂应变、缺口净截面平均有效应力以及骨点应力的小变形有限元分析的寿命预测方法.     

A fault diagnosis model based on weighted extension neural network for turbo-generator sets on small samples with noise

In data-driven fault diagnosis for turbo-generator sets, the fault samples are usually expensive to obtain, and inevitably with noise, which will both lead to an unsatisfying identification performance of diagnosis models. To address these issues, this paper proposes a fault diagnosis model for turbo-generator sets based on Weighted Extension Neural Network (W-ENN). W-ENN is a novel neural network which has three types of connection weights and an improved correlation function. The performance of the proposed model is validated against Extension Neural Network (ENN), Support Vector Machine (SVM), Relevance Vector Machine (RVM) and Extreme Learning Machine (ELM) based models. The results indicate that, on noisy small sample sets, the proposed model is superior to the other models in terms of higher identification accuracy with fewer samples and strong noise-tolerant ability. The findings of this study may serve as a powerful fault diagnosis model for turbo-generator sets on noisy small sample sets.     

基于小样本集推理的雷达信号多维分选技术

目前,直方图法、PRI变换法以及平面变换法等一维、二维分选方法无法有效利用雷达信号脉冲描述字进行分选,而现有多参数聚类的多维分选方法,又存在最佳分类个数和误差范围的选择问题。针对这些问题,提出了基于小样本集推理的雷达信号多维分选技术。计算机仿真结果表明,该方法在提高分选正确率的同时可以有效降低计算复杂度。利用卫星接收到的雷达信号数据进一步验证了采用该技术的分类器具有良好的推广性。     

样本量为2的极小样本相容性检验方法

在航空航天领域由于成本、时间周期等原因进行疲劳寿命及可靠性评估时样本量通常极少（m=1或2）,利用相容性检验方法可对样本量进行扩充。常规的Wilcoxon秩和检验和K-S（Kolmogorov-Smirnov）检验适用于小样本情形,而极小样本相容性检验方面研究较少,且缺乏对方法合理性的详细说明和对不同方法检验功效优劣的比较。航空航天产品疲劳寿命多服从正态分布,因此本文主要以正态分布作为研究对象。利用Monte Carlo仿真发现从某一正态分布N（μ,σ²）中随机抽取两个样本x₁、x₂计算均值μ₁和标准差σ₁后构建新正态分布N（μ₁,σ₁²）,其±σ₁、±2σ₁和±3σ₁范围内的点落在原正态分布N（μ,σ²）±3σ范围内的概率依次为99.80%、98.13%和97.37%。在此基础上针对现场试验数据样本量为2的情况,本文提出利用3σ原则对先验信息数据进行相容性检验从而扩充样本量的方法。将该方法与两种文献方法对比后发现其误差率明显更低并呈现出检验性能随先验数据增加而不断提高的优势。