Abstract: In order to study the formation and evolution of the subjective probability cumulative deviation of grouped population under the extreme loss of individual death, the model of subjective probability cellular automata under different grouping modes is constructed and the Netlogo simulation platform is used for experiments.Then, On the basis of the graph analysis of the simulation results, the regression analysis and case verification analysis of the experimental data are further carried out. The research results show that the evolution of subjective probability of a group generally goes through the process from drastic change to stabilization; Group operation can improve safety redundancy, but this is not good for individual accident experience accumulation; The individual death will lead to the subjective probability underestimation and differentiation of groups, but grouping measures can reduce this effect; The reduction of objective accident probability will increase the difference of subjective probability within the group but grouping measures can reduce this effect; there is a complementary effect between accident rate and mortality and grouping measures have a two-way regulating effect on them; The increase of population size is helpful to reduce subjective probability deviation and difference.

Key words: grouped population, extreme loss, cellular automata, subjective probability, sampling bias

摘要： 为了研究个体死亡这种极端损失条件下分组群体的主观概率累积偏差的形成与演化规律,构建不同分组方式下的群体主观概率元胞自动机模型并运用Netlogo仿真平台进行实验。在对仿真结果进行图形分析的基础上,进一步对实验数据进行了回归分析和案例验证分析。结果表明:分组群体的主观概率演化一般会经历从急速变化到趋于稳定的过程;分组作业方式能够提高安全冗余,但不利于个体事故经验积累;个体死亡等极端损失会导致群体的主观概率低估和分化,但分组措施能够降低此作用;客观事故概率的减少会增大群体内部主观概率的差异,但分组措施能够降低此作用;事故率与死亡率之间存在互补效应,分组措施对其存在双向调节作用;群体人数增加有助于减少主观概率偏差和差异。

关键词: 分组群体, 极端损失, 元胞自动机, 主观概率, 采样偏差