Ordinary gray relational ( OGR) analysis usually stresses comprehensive performance of time serial data. When used in valida-tion of simulation models, its comprehensive performance ignores time series’ details, which is usually part of model error. To reduce the risk in application of OGR, a new gray relational analysis with dog watch mechanism was proposed, which is in view of the details of time series. When deviation excesses threshold value, dog watches and numbers are recorded. The numbers can provide technical support for model modifying. Finally, a case study was given to show reasonability and validity of the improved model.%灰色关联结果注重数据序列的整体性能评价,从算法设计上忽略数据序列细节。将灰色关联分析应用于仿真模型验证时,存在部分阶段不准确模型被大量其余阶段准确模型淹没的风险,针对这一问题,提出一种带看门狗机制的灰色关联模型验证方法,看门狗识别、吠叫和记录大差距序列数据,记录的数据结果可为后续模型修正提供细节基础,提高了模型验证的可靠性和细节可辨识度。最后,通过实例分析验证了改进模型的合理性和有效性。
展开▼