Abstract: In the problem of optimal estimation of model parameters using risk criteria we propose an approach of separation of linear parameters from the nonlinear ones. In the problem of finding a global minimum of risk criteria our approach leads to a decrease of the dimension of the space of free variables up to the dimension of the space of the nonlinear parameters. This allows one to obtain a simpler minimization problem, which can be solved more efficiently via Monte Carlo methods. Such an improvement is very significant in estimation of the models of the object “aging” while investigation of geophysical objects, models of which typically have high dimensionality. We illustrate the proposed method with processing and analysis of data obtained during the field observations in the regime of monitoring.

Keywords: math model, active monitoring, risk criteria, linear and nonlinear parameters.

ACM Classification Keywords: G. 1. 6. Mathematics of Computing, Numerical Analysis, Optimization.

Link:

ABOUT CRITERIA FOR AN ESTIMATION OF NONLINEAR PARAMETERS IN MODELS OF MONITORING

Sergii Mostovyi, Vasilii Mostovyi

http://www.foibg.com/ijita/vol19/ijita19-2-p02.pdf