基于役龄和健康指数的可加故障率模型研究

针对传统的可加故障率模型(AHM)在描述设备状态对故障率的影响时存在缺陷,提出采用健康指数作为表征设备状态的变量,建立起以役龄和健康指数作为变量的AHM模型,基于该模型的构建原理提出模型解法,共三步:首先,求解基准故障率函数时,运用基于LS-SVM的EMD改进算法获取故障率演变曲线,用遗传算法优化该曲线;然后,求解协变量函数时,运用了灰色理论的NGDM模型;最后,对基准故障率函数和协变量函数进行联合优化。通过仿真数据验证,该方法的拟合误差小于2%,预测误差小于3%。结合真实案例探讨了模型适用性的初步检验方法,最终对模型的使用方法进行了论述。     

A high-order nodal discontinuous Galerkin method to solve preconditioned multiphase Euler/Navier-Stokes equations for inviscid/viscous cavitating flows

In this study, a high-order accurate numerical method is applied and examined for the simulation of the inviscid/viscous cavitating flows by solving the preconditioned multiphase Euler/Navier-Stokes equations on triangle elements. The formulation used here is based on the homogeneous equilibrium model considering the continuity and momentum equations together with the transport equation for the vapor phase with applying appropriate mass transfer terms for calculating the evaporation/condensation of the liquid/vapor phase. The spatial derivative terms in the resulting system of equations are discretized by the nodal discontinuous Galerkin method (NDGM) and an implicit dual-time stepping method is used for the time integration. An artificial viscosity approach is implemented and assessed for capturing the steep discontinuities in the interface between the two phases. The accuracy and robustness of the proposed method in solving the preconditioned multiphase Euler/Navier-Stokes equations are examined by the simulation of different two-dimensional and axisymmetric cavitating flows. A sensitivity study is also performed to examine the effects of different numerical parameters on the accuracy and performance of the solution of the NDGM. Indications are that the solution methodology proposed and applied here is based on the NDGM with the implicit dual-time stepping method and the artificial viscosity approach is accurate and robust for the simulation of the inviscid and viscous cavitating flows.     

两类离散GM(1,1)模型及其软件缺陷预测建模

为了拓广离散GM(1,1)模型的应用范围,建立了近似非齐次指数序列的离散GM(1,1)模型,NDGM(1,1)模型和直接离散GM(1,1)模型,即DDGM(1,1)模型,证明了其可以完全拟合非齐次指数序列,最后将两类离散GM(1,1)对软件进行缺陷预测建模,结果显示,DDGM(1,1)模型具有较高预测精度,可以对后续软件开发中缺陷的存在情况做出相应预测.