An upper hessenberg sparse matrix algorithm for modal identification on minicomputers
✍ Scribed by S.R. Ibrahim
- Publisher
- Elsevier Science
- Year
- 1987
- Tongue
- English
- Weight
- 694 KB
- Volume
- 113
- Category
- Article
- ISSN
- 0022-460X
No coin nor oath required. For personal study only.
✦ Synopsis
The time domain modal identification problem is here reduced to an eigenvalue problem of a sparse upper Hessenberg matrix. Such a matrix has only a number of elements equal to its order (one column), subdiagonal elements of unity and all the other remaining elements are zeros. The one vector elements of this matrix are obtained by using Cholesky decomposition to solve a positive definite system of equations which is constructed by using a structure's free decay or impulse time histories. Any number of measurements may be used, from one or more tests with a single or multiple shaker. The direct use of an upper Hessenberg matrix eliminates the need of transforming a full matrix to such a form and consequently greatly reduces the computational requirements when the QR algorithm is used for the eigensolution. In addition to reducing computer time and storage, the proposed technique has more flexibility than other comparable time domain approaches. It also produces higher identification accuracy especially in the identified damping factors. Sample results are presented with comparisons to those of other time domain algorithms.