On State Estimation of Dynamic Systems by Applying Scalar Estimation Algorithms

The scalar estimation algorithms are low-sensitive to input noise statistics due to adaptive adjustment of the gain coefficient depending on current estimation errors. Scalar approaches to state vector estimation differ from others by its capability to form estimation equation independently for each observable component of the state vector. In order to increase the accuracy of scalar estimation algorithms, the quantitative criteria of observability was proposed. By applying error-models of inertial navigation systems, the formulae of observability degree of misalignment angle and drift rate were deduced. For the purpose of analyzing the capacity of suggested approaches, laboratory tests based on actual inertial navigation systems were applied. The analyzed results indicate that the growth of sampling time within a certain range generates the increase of the degree of observability.