Observability numerical analysis method for navigation system

Abstract

The invention provides an observability numerical analysis method for a navigation system. According to the method, an observability matrix of a system is constructed firstly, state variable decoupling is achieved by conducting row transformation on the observability matrix, relational expressions of state variables after decoupling are obtained, the coupling relation between the state variables can be visually seen through the relational expressions, and therefore observability analysis is conducted. The method does not require a system to be completely observable; key state variables are purposefully selected for decoupling by utilizing priori knowledge of a navigation system; when the state variable is decoupled, the specific form of the decoupled state variable relational expression isnot solved, software such as Matlab is used as a mathematical auxiliary tool to substitute known parameter values of the system into an observability matrix; and then row transformation is performedon the observability matrix in the numerical form to obtain a numerical form of a decoupled state variable relational expression, and finally observability analysis is performed on the system by utilizing the state variable relational expression in the numerical form.

Description

technical field [0001] The invention relates to a numerical analysis method for the observability of a navigation system, which is suitable for the observability analysis of a multi-dimensional complex navigation system. Background technique [0002] When using the filtering algorithm to estimate the state of the navigation system, whether each state variable can be accurately estimated, as well as the speed and accuracy of the estimation are determined by its observability. If the system is fully observable, the determining factors affecting the accuracy of its state estimation are only system noise and measurement noise; State variables can get good estimation effect. Therefore, before performing state estimation, it is necessary to analyze the observability of the system to determine whether the system is fully observable, and for an incompletely observable system, determine its observable state variables and unobservable state variables. [0003] Existing observability...

AI Technical Summary

Problems solved by technology

However, in a multidimensional complex system, a linearly independent row of the observability matrix usually corresponds to a combination of multiple state variables, and the singular value-based observability analysis method simply assumes that the singular value obtained by singular value decomposition has the largest corresponding coefficient state variables, this method of stripping combined state variables is not comprehensive

Therefore, in actual use, the analysis results may not be completely consistent with the filtering estimation results.

Navigation systems are usually multi-dimensional complex systems, and the above two observability analysis methods may be difficult or inapplicable in application

Images