The estimation of sinewave parameters has many practical applications in test and data processing systems. Measuring the effective bits of an analog-to-digital converter and linear circuit identification are some typical examples. If a sinew ave's frequency is known, there is an established linear method to estimate the other parameters. But when none of the parameters are known (which is usually the case in practical situations), the estimation problem becomes more difficult. Traditional approaches to this task applied an iterative, sinewave curve-fit algorithm. Two problems with this technique are that convergence is often slow and not always guaranteed and the results of different trials may be inconsistent due to trapping at a local minimum. Recently, a non-iterative algorithm has been developed which computes all four sine wave parameters directly. The algorithm combines a nonlinear technique and windowing to compute the estimates. Although this method is faster and more consistent than the curve-fit approach, one disadvantage is that the accuracy of some estimates tends to deteriorate rapidly if the sinusoid is corrupted by a high level of noise distortion. This study presents an improved algorithm to extract the four parameters of an unknown sinusoid from a sampled data record even though the samples may be distorted by a high level of noise. Given this record, the proposed method first computes the FFT (Fast Fourier Transform) of the data. Analysis of the resulting frequency spectrum provides a rough estimate of the sinewave's fundamental frequency. Next, a bandpass filter designed around this frequency is used to eliminate much of the noise from the samples. Applying the existing four-parameter estimation algorithm to the filtered data, yields a more accurate frequency estimate. Finally, this new value, together with the original (noisy) data record is input to the three-parameter estimation algorithm to determine the remaining sinewave parameters. Simulation results indicate this proposed (new) algorithm not only shows substantial improvement in the accuracy of parameter estimates, but also produces consistent results for higher levels of noise distortion than previous methods have achieved.
Identifer | oai:union.ndltd.org:pdx.edu/oai:pdxscholar.library.pdx.edu:open_access_etds-5756 |
Date | 30 April 1993 |
Creators | Rydell, Kendall Ann |
Publisher | PDXScholar |
Source Sets | Portland State University |
Detected Language | English |
Type | text |
Format | application/pdf |
Source | Dissertations and Theses |
Page generated in 0.001 seconds