Determining simulation sample size for given precision requirements

• 作 者: SHIPeng;YANGMing;
• 机 构: Control and Simulation Centre, Harbin Institute of Technology, Harbin 150080, China
• 出 处: 2011年中国智能自动化会议
• 发表日期: 2011年期

　　Simulation credibility evaluation concerns
　　comparisons of the performance measures between
　　simulation system and its
　　reference counterpart, In fact, the required
　　precision is known as prior information, which brings
　　the problem of determining
　　simulation run length according to the estimation of
　　performance measures and the required precision. A
　　feedback method was
　　proposed from model analysis to data collection. For
　　transient performance measures, operating
　　characteristic curves were employed
　　to calculate the simulation replications to detect
　　the difference of means in parameter range. For
　　steady-state performance measures,
　　two phase method was put forward to satisfy the warm-
　　up sizes at the first phase, and a run length for the
　　relative- or
　　absolute-precision requirement at the second phase.
　　Some empirical experiments illustrate the validity of
　　the proposed method.