Investigation of Algorithms for Identification of Wave Parameters in Solid-State Wave Gyroscopes without Adjusting Calculations to the Frequency of Signals

Abstract

Listen

Translate article

To improve the efficiency of production operations of diagnostics, adjustment and control in the manufacture of integrating solid-state wave gyroscopes, algorithms for improving the accuracy of identification of wave parameters from measuring signals without their additional adjustment to the frequency of signals are considered. In the first algorithm for processing measurement results, a virtual transition to the moving axes of standing waves was introduced without taking into account the influence of a quadrature standing wave. It is characterized by the lowest potential accuracy, but it is convenient for the formation of initial approximations in the problems of clarifying optimization of identification functions in other algorithms. The second algorithm is based on direct five-parameter numerical minimization of the error feature by the conjugate gradient method. To avoid local extremes, it is recommended to use it as a clarifier. In this case, the initial search points are calculated by other algorithms and therefore selected near extremum. This algorithm allows identification at the shortest time intervals, which may be required when measuring high angular velocities. In the third algorithm, computational conditionality is improved at short identification intervals for increased level of measurement noise. In the fourth algorithm, the measurement results are processed using numerical procedures of digital demodulation under noise conditions. Comparison of the identification algorithms accuracy was carried out by methods of simulation modeling for theoretically set initial signals. This made possible direct comparison of initial and identified characteristics of wave processes: amplitudes of the main and quadrature standing waves, the angle of the main standing wave and its frequency. For the four algorithms considered, the results of simulation in the absence and presence of noise in the measuring signals are given. In the first case, diagrams of identification errors are analyzed depending on the length of the interval with a fixed and variable proportion of digital sampling frequencies and resonator oscillations. In the second case, there are standard identification errors depending on the noise level in the measuring signals (up to 20 percent of the basic amplitude of oscillations). Comparison of the results obtained showed the advantages and disadvantages of the corresponding algorithms, which is important for their practical adjustment when working with real samples of gyroscopes.