针对受非平稳高斯随机激励下的线性时不变系统,基于连续时间AR模型,提出了一种时域模态识别的新方法,这种方法仅从响应数据就能够识别系统的物理参数.为了对参数进行估计,把结构动力学方程看作一个连续时间AR模型,并给出了它的状态空间形式,其中的状态方程就是一个随机微分方程;接着利用在非常短的时间段内均匀调制函数接近于一个常数矩阵的事实,得到均匀调制函数的估计;然后再利用Girsanov定理,得到物理参数的精确极大似然估计;最后进行特征分析,从而实现线性系统模态参数的识别 Aiming at the linear time-invariant system under non-stationary Gaussian random excitation, a new method of time-domain modal identification is proposed based on the continuous-time AR model. This method can identify the physical parameters of the system only from the response data The parameters are estimated, and the structural dynamic equation is treated as a continuous-time AR model, and its state-space form is given, in which the state equation is a stochastic differential equation; then the uniform modulation function is used in a very short period of time Close to a constant matrix to obtain the uniform modulation function estimate; then use the Girsanov theorem to get the exact maximum likelihood estimation of the physical parameters; and finally carry out the characteristic analysis to achieve linear system modal parameter identification