摘要
无失效数据的出现,使分布参数的估计成为可靠性评定的又一关键问题。在对指数分布进行参数估计时,引入模糊加权系数,利用模糊加权最小二乘估计对数据进行回归拟合,推导出模糊加权失效率的估计式。由于无失效数据失效概率的确定对参数的估计结果有较大的影响,以减函数法确定失效概率的Bayes估计,并进行模糊加权线性回归。回归结果与常规线性回归结果相比,更接近工程实际。
Parameter estimation is very important for reliability estimation of zero failure data. For the question of poor stabilization in conventional Linear Regression equation, it is usually improved by a weighted coefficient. The fuzzy weight is put forward and for exponent distribution, estimated equation of failure rate is given. Due to the fact that failure probability has large effect for the estimated value, the Bayes estimation is given which bases on decrease function. The fuzzy weighted linear regression model can get low departure. The regression result shows that the new method is more approach to the engineering practice.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2005年第6期1373-1375,共3页
Journal of System Simulation
关键词
线性回归
加权系数
隶属度
无失效数据
linear regression
weighted coefficient
subordination
zero-failure data
