Problems and Methods of Estimation of Parameters of Mathematical Models

Abstract

The technology of step-by-step modeling of the functioning of quasi-stationary systems is described. One of the methodological problems in the implementation of this technology is the choice of a method for estimating the parameters of the model at different stages of its formation based on the sometimes contradictory, inaccurate, multi-time data available. To estimate the parameters of dependencies, various methods can be used, including least modules, least squares, minimizing the maximum deviation of the discrepancies between the estimates of the parameters used in the dependency model and the available source data. In relation to the problem of estimating the parameters of linearized dependencies, the results of a comparative analysis of these and other methods of parameter estimation are considered. An example of estimating the mortality parameters of a large and small golomyanka is considered on the basis of approximation by a truncated exponential law of the distribution of the age of fish from catches.