Spectroscopy measurements of chemical compounds yield information correlated to the elements forming the compound's composition. A valid representation of this measured spectral response is described in terms of summed gaussian models. To interpret this spectral data, using this model, a recursive algorithm is established to quantize the nonlinear parameters associated with each gaussian model and separate each element's contribution from the data. The parameter identification is accomplished using a Marquardt procedure over a limited portion of the normalized data set. A nonlinear representation of a Kalman filter is subsequently used to strip away each gaussian model from the summed gaussian data set. The recursion between parameter estimation and filtering results in the separation of each elemental contribution from the measured data set, while reducing computer time and memory requirements normally encountered.
|Original language||English (US)|
|Number of pages||8|
|Journal||Proceedings - Micro-Delcon|
|State||Published - 1981|
ASJC Scopus subject areas