Method and device for performing robust state estimation on state-limited nonlinear system

Abstract

The invention discloses a method and a device for performing robust state estimation on a state-limited nonlinear system. The method comprises the following steps: performing state space implementation on the state-limited nonlinear system with a preset application background; converting the nonlinear system into a linear system by using a statistical linear regression method; processing abnormal interference and correlation introduced by linearization by using a regularization least square principle; processing the constraint condition by using a Lagrange multiplier method; simultaneously estimating a state value and abnormal interference of the system by using an alternating direction multiplier method; and taking a corresponding regularization parameter used for adjusting the abnormal interference detection intensity when the difference between the estimated probability and the empirical probability of abnormal interference occurrence is minimized as an optimal regularization parameter. In this way, under the condition that the state-limited nonlinear system is affected by abnormal interference, additional negative effects caused by correlation introduced by linearization of the system can be eliminated, and an accurate robust state estimation result is obtained.

Description

technical field [0001] The invention relates to the technical field of discrete nonlinear system filtering, in particular to a method and device for robust state estimation of a state-limited nonlinear system. Background technique [0002] In practical applications, many engineering systems have nonlinear and asymmetric system dynamic characteristics. Although the nonlinear model with Gaussian noise interference is widely used to describe the actual physical system, the Gaussian noise model is only an approximation to the real noise. In engineering practice, factors such as changes in working conditions, human error, and system failures can introduce abnormal disturbances into the system, making the actual noise non-Gaussian. When the actual noise deviates from the Gaussian assumption, traditional estimation methods will lose their proper accuracy and robustness. At the same time, due to the limitations of objective physical conditions, the constraints of equipment perform...

AI Technical Summary

Problems solved by technology

However, linearization means that both the system function and the observation function need to be expanded around an estimated state value, thus introducing additional correlations to the linearized system function and observation function

When an outlier caused by abnormal interference appears in the estimated state value, both the system equation and the observation equation will be affected by the outlier, which makes existing estimators suc

Images