一种考虑数据截尾的非参数软件可靠性模型

通过分析数据截尾对软件可靠性模型的影响，建立了一种考虑数据截尾的非参数软件可靠性模型，为的是克服一般参数模型假设约束过紧的问题，并能在估计失效率的同时估计残存缺陷数．同时，提出了基于非参数统计的软件失效率单调下降性检验方法，在加权核函数估计中引入数据截尾率作为加权系数和优化参数，并用失效率单调下降约束下的加权核函数方法来估计失效率和残存缺陷数．模拟试验和实例分析表明，所提模型可以较好地处理截尾数据，给出残存缺陷数的合理估计，其失效率估计精度较高，不低于较好的传统模型．

Nonparametric Software Reliability Model Concerned with Censored Data

After analyzing the negative influences of censored data on the software reliability model, a novel nonparametric software reliability model concerned with censored data was presented in order to solve the problem that the assumption constraints added to general parametric models are too strong and to calculate the number of residual defects while estimating failure rate. Meanwhile, based on nonparametric statistic a monotonic decreasing check method of software failure rate was proposed, the censored rate was introduced to the weighted kernel function estimation as a weighted and optimization parameter, and the weighted kernel function estimation method with constraints of monotonic decreasing failure rate was used to estimate failure rate and the number of residual defects. The simulation experiment and example analysis show that the proposed model can deal with censored data well, and can give reasonable estimation of the number of residual defects. The estimation precision of failure rate is higher than conventional models.