Optimization Algorithm for Kalman Filter Exploiting the Numerical Characteristics of SINS/GPS Integrated Navigation...

Area of Science:

• Navigation Systems
• Computational Optimization
• Signal Processing

Background:

• Traditional Kalman filters for SINS/GPS integration face high computational demands.
• Existing methods often involve approximations or complex matrix theories, limiting efficiency and applicability.

Purpose of the Study:

• To develop a computationally efficient Kalman filter optimization algorithm for SINS/GPS.
• To reduce CPU time and memory usage while maintaining filter accuracy.
• To create an easily transplantable optimization method for other filters.

Main Methods:

• Exploiting matrix sparseness and symmetry for computational simplification via offline derivation and block matrix techniques.
• Implementing a parallel computational mechanism through subdivision and restructuring of calculations based on "useful" data.
• Utilizing numerical approaches that avoid precise-loss transformations or approximations of system modules.

Main Results:

• Achieved approximately 90% savings in CPU processing time compared to classical Kalman filters.
• Reduced memory usage by approximately 66%.
• Maintained high accuracy, with minimal loss compared to the unoptimized filter.

Conclusions:

• The proposed algorithm offers significant computational efficiency for SINS/GPS Kalman filtering.
• The method is accurate, requires no complex matrix theories, and is easily adaptable to other filters.
• This optimization strategy enhances practical implementation of integrated navigation systems.