奇异值分解技术对从噪声背景下提取振动或其它信号的本质特征是一非常有效的工具，时序信号在噪声情况下的嵌入（相空间中）会产生错误的结果，本文应用本征值分解技术对动力系统实测数据嵌入空间矩阵的本征值进行了计算，提出了具体计算嵌入空间矩阵的本征值及其本征向量的计算方法，以及嵌入空间矩阵基的具体选取方法，从而完成了对动力系统实测混沌数据的相空间重构的最基础的工作 The singular value decomposition technique is a very effective tool for extracting the essential features of vibration or other signals from noisy background. The embedding of time-series signals in phase space will produce erroneous results. In this paper, we use eigenvalue decomposition The eigenvalues ​​of dynamical system embedded data matrix are calculated. The eigenvalues ​​of eigenvalues ​​and their eigenvectors are calculated, and the method of choosing eigenvector of embedded matrix is ​​proposed. The most basic work of reconstructing the phase space of chaos data in dynamical system