A.Ya.Kalyuzhny An algebraization of the acoustic tomography problem using the principal informative components |
Acoustic bulletin, Vol. 1 ¹ 4, (1998) |
| It is shown that representing of the field of medium's parameters to be reconstructed as a special finite-dimension basis may increase the efficiency of solving of the acoustic tomography problems. The extremal properties of the Fisher information operator eigenfunctions form the background of the proposed approach. The basis for medium characteristics representation constructed by these functions provides the minimum of their reconstruction error. It is shown that the optimal dimension of the basis exists that provides the maximal accuracy of measurements in the preset conditions of the topographic experiment. The criterion for selection of the optimal basis functions with the allowance of both the fluctuation and systematic components of the resulting error is formulated. The projection approach of basis construction is offered that combines the advantages of purely physical description (evidence, economy) and the statistical-informational approach (error minimization). The structure of the information operators is studied for the typical models of measuring signal fields. The efficiency of the offered method is illustrated with the examples from the ocean acoustic tomography. |
|
KEY WORDS: ocean acoustic tomography, Fisher's information operator, error minimization |
| TEXT LANGUAGE: Russian |