首页> 外文会议>International Modal Analysis Conference >AN ALTERNATIVE TIME - DOMAIN SYSTEM IDENTIFICATION ALGORITHM
【24h】

AN ALTERNATIVE TIME - DOMAIN SYSTEM IDENTIFICATION ALGORITHM

机译:替代时域系统识别算法

获取原文

摘要

A time-domain system-identification method, developed for damped sinusoids in additive noise, is presented. The method is based on a modification of the Pisasrenko Harmonic Decomposition analysis of pure sinusoids. Singular value decomposition is used to determine the eigenvectors of a non-Toeplitz covariance matrix. System parameters are obtained from the eigenvectors corresponding to the minimum eigenvalues. These vectors span a space orthogonal to the deterministic space of the damped sinusoids. Theory and implementation of the method is presented, along with a model order estimation procedure based on analysis of the singular values. The method is not subject to the bias of autoregressive least squares methods, and thus no over specification of model order is required. Performance of the method is evaluated based on simulated data. The method is compared to autoregressive least squares and correlation fit methods in the test using simulated data. Results indicate the new method provides improved frequency and damping estimates over the other methods.
机译:提出了一种用于增压噪声的阻尼正弦曲线的时域系统识别方法。该方法基于纯正弦曲线的Pisasrenko谐波分解分析的修改。奇异值分解用于确定非卷积协方差矩阵的特征向量。从对应于最小特征值的特征向量获得系统参数。这些矢量跨越与阻尼正弦曲线的确定性空间正交的空间。提出了方法的理论和实现,以及基于分析奇异值的模型顺序估计过程。该方法不受自回归最小二乘法的偏差,因此不需要过度规格。根据模拟数据评估该方法的性能。将该方法与自动增加最小二乘和使用模拟数据进行测试的相关拟合方法进行比较。结果表明,新方法提供了其他方法的改进频率和阻尼估计。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号