共同影响因素下基于对数正态分布的项目工期仿真算法  

Project Scheduling Simulation Algorithm Based on Lognormal Distribution Under Common Influence Factors

作　　者：齐德伟 QI De-wei(School of Economics and Administration,Xi'an University of Technology,Xi'an 710054,China)

机构地区：[1]西安理工大学经济与管理学院,陕西西安710054

出　　处：《数学的实践与认识》2023年第9期163-174,共12页Mathematics in Practice and Theory

摘　　要：建立了一种基于对数正态分布的项目工期仿真算法.以往对于项目工期的仿真都假设项目中活动时间相互独立,将活动时间的影响因素分为共同影响因素和独立影响因素,独立因素影响下的活动时间服从对数正态分布,共同影响因素为另一对数正态分布变量,活动的实际时间等于两变量的乘积,由于共同影响因素的引入使得活动之间不再独立.同时给出了从历史数据中估计共同影响因素分布函数参数的方法,该方法考虑了数据中存在的“帕肯森”效应.根据现实中50个项目的实际数据运用不同的分布函数进行了仿真,结果发现算法得到的项目工期分布明显好于活动时间独立下的正态分布、指数分布、贝塔分布,且对活动时间的方差不敏感.This paper establishes a new project scheduling simulation method based on lognormal distribution.In previous researches,activity duration is presumed to be mutually independent,however this is not true in reality.In our research,we classify the activity duration influence factors into two groups,one is activity related factors,the other is common factors.The task time fits lognormal distribution under the influence of exclusive factors,while the common factors are also presumed to be a lognormal random variable.Activity times equals to the product of the two independent lognormal random variables under both types of factors.We also provide a method to estimate the parameters of common factors from history data,this method considers the Parkinson effect within the data.Experiments have been conducted with 50 real-life projects data using our method,results show that project schedule distribution got from our algorithm is much more practical than normal,exponential and beta distribution under the condition of independence,and the project time distribution is not sensitive to the activity duration's coefficient of variation.

关 键 词：项目工期 仿真 对数正态分布 “帕肯森”效应 

分 类 号：TU722[建筑科学—建筑技术科学]