METHODOLOGICAL BASES AND PRACTICAL ASPECTS OF OPTIMIZATION TASKS OF THE BEARING STRUCTURES OF THE STRAPDOWN INERTIAL NAVIGATION SYSTEMS

Abstract

This article describes approaches to solving problems of optimization of bearing structure
of strapdown inertial navigation systems (SINS). A typical optimization problem in this case is
multiobjective parametric optimization of the bearing structure of the SINS accelerometer triad in
order to minimize the mass of the bearing structure and minimize deviation angles of the accelerometer
axes under the action of external loads. The ANSYS Mechanical and ANSYS
DesignXplorer modules are used as a tool for numerical modeling and optimization, respectively.
Practical issues related to parameterization of SINS bearing structure 3D-models, calculation of
accelerometer axes deviation angles, possible variants of numerical experiment plans, estimation
of response sensitivity to input parameters, generation and refinement of the response surface, and
multiobjective optimization are considered. For the rational parametrization of geometry, the
SINS device assembly was decomposed, as a result of which the parts and structural elements that have the greatest influence on the considered objective functions were identified. To calculate the
deviation angles of the sensitive elements axes, special two-node finite elements and relations for
the Bryant angles were used, which describe the relative position in space of two coordinate systems.
When planning a numerical experiment, at the first stage of optimization, a central composition
plan was used, and at subsequent stages, the parameter space was filled using the Latin hypercube
method with the option of relations between parameters, which made it possible to avoid
degenerate design options. The response surface was built using the genetic aggregation method
and subsequently refined based on a set of optimal solutions. Optimization for conflicting goals of
mass minimization and stiffness maximization was carried out using a multiobjective genetic algorithm.
The described set of approaches to solving optimization problems as a result of an exemplary
series of calculations made it possible to reduce the mass of a serial SINS bearing structure
part by 23% with fixed stiffness.