利用方差分析法进行模型验证

Application of Variance Analysis in Model Validation

计算机仿真在许多领域得到了越来越广泛的应用,这给模型验证方法的研究提出了更高的要求。众所周知,方差分析法是一种常用的统计方法。它通过比较因素在不同水平下进行试验所得到的试验指标,来检验不同水平下试验结果的差别是否显著。如果将仿真模型与实际系统分别看作试验因素的水平,在其他条件相同时进行试验,获得两组试验指标。对所获得的数据利用方差分析法进行分析,则可检验其差别是否显著。若差别不显著,则表明仿真模型与实际系统在给定的显著性水平下是一致的。该文最后结合具体数据给出了该方法的具体应用,所得结论与利用其它验证方法得到的结论相一致,从而说明利用方差分析法进行模型验证是可行的。

Computer simulation is used in more and more fields, which has brought more demands on the research of model validation methods. As is known to all, variance analysis is a common used statistical method. It tells the difference between different states of factors via comparing the result of experimentation. Take the simulation model and the real system as two states of a factor and get two groups of data in the same condition. Analyse variance of the data ,then tell whether the difference between two groups is obvious or not . The simulation model accords with the real system if the outcome is not obviously different from each other at the given threshold value. In the end, the paper gives an example of the method with some data. The conclusion is in accordance with those made by other methods which proves that variance analysis can be used for model validation.