敏感性分析概率发生模型的改进

发布时间：2018-07-12 10:07

本文选题：响应分布 + 序贯试验　；参考：《湘潭大学》2015年硕士论文

【摘要】：敏感性分析的试验应用于众多工程领域中,并在经济学、生物学等其他学科的某些领域,也经常需要这一类试验.不同产品,对于外界刺激,都会有反应.但是如何判断刺激量是否对结果有影响,就是敏感性分析的任务.对应刺激量是否发生的概率,需要做一定的研究,其概率分布被称为响应分布.过去的敏感性分析试验方法中,仅仅关注这个变量对指标是否影响巨大,影响大就考虑,而对于是否发生的概率研究较少。在实际应用中,人们关注点不只是寻找敏感性因素,更关注产品在其能量输出、响应的延迟时间等多种因素下影响发生的概率,进而指导接下来的工作.只有这些都达标了,才会被认为是成功作用.工程设计中,理想水平即不低于概率P作用成功的试验水平.本文针对估计理想水平的实际问题,在对比了以往敏感性试验方法的基础上,通过序贯试验来改进估计概率P的值,同时考虑响应后的输出、延迟时间等综合效应,改进原有模型的基础上,来考虑敏感性分析的概率问题,以达到需要的效果.文中第三章详细证明并讲解了这种方法,第四章通过模拟验证了新方法的估计效果.
[Abstract]:The experiment of sensitivity analysis is used in many engineering fields, and it is often needed in some fields of economics, biology and other disciplines. Different products, for external stimulus, will have a response. But it is the task of sensitivity analysis to determine whether the stimulus has an effect on the results. It is necessary to do some research on the probability of whether stimulus occurs or not, and its probability distribution is called response distribution. In the past sensitivity analysis test methods, only attention is paid to whether this variable has a great influence on the index, but the probability of occurrence is less studied. In practical application, people pay more attention not only to find the sensitive factors, but also to focus on the probability of the product under a variety of factors, such as energy output, response delay time and so on, and then guide the next work. Only when these are met can they be regarded as successful. In engineering design, the ideal level is not lower than the successful test level of probabilistic P action. In this paper, aiming at the practical problem of estimating ideal level, on the basis of comparing the previous sensitivity test methods, we improve the estimation value of probability P by sequential test, and consider the comprehensive effects such as output after response, delay time and so on. On the basis of improving the original model, the probability problem of sensitivity analysis is considered to achieve the desired effect. In the third chapter, this method is proved and explained in detail. In the fourth chapter, the estimation effect of the new method is verified by simulation.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：O212

【相似文献】