ID 原文 译文
48206 为解决系统输出为非平稳快变数据的仿真模型验证问题,提出了基于经验模态分解和灰色关联分析的验证方法。 In order to solve the system output for non-stationary fast simulation model validation problem of variable data, is proposed based on empirical mode decomposition and validation of the grey correlation analysis method.
48207 首先采用经验模态分解方法将仿真输出和参考输出均分解为趋势项和平稳项两部分, First using empirical mode decomposition method to simulation output and the reference output are decomposed into trend item and smooth two parts,
48208 然后从位置差异和外形差异两方面刻画趋势项之间的差异,用谱密度差异刻画平稳项之间的差异,再者采用熵权法确定各类差异的权重,最后基于灰色关联度分析综合三类差异得到验证结果。 and then from two aspects: the position and shape differences describe the difference between a trend, using spectral density difference depict the differences between a steady, moreover USES the entropy weight method is applied in determining weights of all kinds of differences, based on the grey correlation analysis of comprehensive three differences between validation results are obtained.
48209 通过实例应用,验证了方法的有效性。 Through the application instance, verify the validity of the method.
48210 为了提高软件可靠性预计精度和稳定性,本文提出基于组合模型的软件可靠性预计方法。 In order to improve the software reliability prediction accuracy and stability, this paper put forward based on the combination model of software reliability prediction method.
48211 构建组合模型的过程包括选取基模型和确定各基模型的权值。 Build a portfolio model selection process including the base model and to determine the value of the base model.
48212 选取基模型时依据的准则有生命周期阶段、模型用户需求、故障数据类型、故障趋势匹配、模型假设吻合和模型预计偏好,并在此基础上阐述了基模型选取算法。 Selecting base model based on the principles of life cycle phase, model user requirements, the types of failure data, fault trend matching, model assumption and model is expected to preference, and on this basis, this paper expounds the model selection algorithm.
48213 基模型权值的确定采用序列似然比方法。 The base model using sequential likelihood ratio method of determining value.
48214 最后,引用一组软件故障数据,用于检验此组合模型的预计效果。 Finally, reference to a set of software failure data, used for testing the expected effect of this combination model.
48215 通过与单个模型的对比,显示组合模型预计结果更为可信。 Through comparing with a single model, which shows a portfolio model is expected to result more credible.