针对在系统辨识过程中面临的人体头部过载模型阶次辨识问题,介绍了常用的模型阶次辨识方法及利用残差平方和即损失函数J估计模型阶次的原理,并基于20组试验数据开展人体头部过载模型阶次辨识研究。通过系统辨识获得了人体头部过载模型后,采用Gram阵对Hankel奇异值分解方法实施模型简化和降阶处理。通过比较降阶前后模型,表明模型降阶方法是有效和正确的。%Aiming at the human head overload model order identification problem existing in the system identification pro-cess,the common model order identification methods and principle using the sum of residual squares which is also called loss function J to estimate model order are introduced. Based on 20 sets of test data,the research on human head overload model or-der identification wasperformed. After the human head overload model was obtained by system identification,the Gram matrix was used to simplify and reduce the model of Hankel singular value decomposition. By comparing the models before and after re-duction order,the effectiveness and correction of model order reduction method were verified.
展开▼