基于贝叶斯法和蒙特卡洛仿真的威布尔型装备器材需求预测

摘要

The demand of new equipment and materials cannot be mastered well because of less historical demand data and undefined demand. To address this problem, a demand forecasting method based on Weibull distribution is proposed for equipment and materials in the case of small failure samples. The parameters of equipment and material life distribution are estimated by Bayes estimation and MCMC simula-tion for K-S goodness-of-fit test, including scale and shape parameters. A Monte Carlo simulation-based forecasting method for the annual demand of equipment and materials is presented, in which repairing maintenance, preventive maintenance and service time of equipment and materials are considered. The analysis of examples shows that the life distribution model derived from Bayes estimation has higher degree of fitting in the case of few samples, and the Monte Carlo simulation-based forecasting method is simple and effective.%为了解决装备 器材历史需求数据少、需求规律不明确的问题,提出一种基于贝叶斯法和蒙特卡洛仿真的威布尔型装备器材需求预测方法.针对威布尔分布尺度参数未知以及形状和尺度参数均未知两种情况,分别基于贝叶斯方法通过解析求解和数值模拟的方式进行了参数估计,并引入柯尔莫哥洛夫-斯米尔诺夫检验法对寿命分布模型进行拟合优度检验;综合考虑修复性维修、预防性维修和装备器材的已使用时间,提出了基于蒙特卡洛仿真的部队装备器材年度需求预测方法.算例分析表明:小样本下通过贝叶斯估计得到的寿命分布模型拟合度高,基于仿真的需求预测方法简单、有效.