IMPULSE RESPONSE IDENTIFICATION FROM TRUNCATED INPUT DATA USING FFT TECHNIQUES,
Abstract
A method is presented for the noise-free identification of the kernel function of a finite memory linear system from normal operating input-output data. The problem is formulated taking into account the error in the output data which results from the truncation of the input to a finite time interval. Iteration in the frequency domain is used to overcome the truncation error, and a simple bound involving the input data is derived which is useful in predicting convergence. The Fast Fourier Transform algorithm is employed at each stage in the iteration which results in a computationally efficient identification procedure. Computer simulation studies of the algorithm are summarized which substantiate the theory and provide additional information of a practical nature. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1970
- Accession Number
- AD0710650
Entities
People
- Allan E. Pearson
- H. F. Silverman
Organizations
- Brown University