针对基于协方差驱动随机子空间辨识法虚假模态影响识别结果的问题,提出了一种基于模态能量的虚假模态剔除方法.利用输出矩阵、状态矩阵的特征值与特征向量以及状态-输出协方差矩阵计算出识别结果中各阶模态分量的模态能量,对各假设模型阶数下计算出来的能量进行排序,保留能量最大的前j个模态用于绘制出稳定图,剩下的模态视作为虚假模态予以剔除.通过对3自由度的线性时不变系统和重庆朝天门长江大桥模型进行辨识,验证了该方法的有效性. Aiming at the problem of false modal influence recognition based on covariance driven stochastic subspace identification, a modal energy-based false modal elimination method is proposed. By using the output matrix, the eigenvalues ​​and eigenvectors of the state matrix and the state-output The covariance matrix calculates the modal energy of the modal components of each order in the recognition result, and sorts the energies calculated under the order of the assumed models, preserves the first j modals with the largest energy for drawing a stable graph, and leaves the rest As a false modal.It verifies the effectiveness of the proposed method by identifying the three-degree-of-freedom linear time-invariant system and the Chongqing Chaotianmen Yangtze River Bridge model.