On error evaluation of the quasilinear filtering method

The paper is concerned with errors of the linear filtering method which statisticall y linearizes the nonlinear functions in the right hand sides of differential equations which describe the useful signals. The numerical results of the studies demonstrate that application of Kalman filters to a statistically linearized nonlinear system can result in biased estimates and significant deflection of the model signal filtering error variance from that of the actual useful signal.