Application of Wavelet Analysis in Testing of Dynamic Characteristics of Fiber Optic Gyroscope

摘要

This paper introduces testing methods of dynamic characteristics of fiber optic gyroscope (FOG). We test dynamic characteristics of FOG by using turntable, which can simulate a ship under different swaying amplitudes and swaying frequencies. By controlling turntable under sway motion, we test output signals of FOG and compare them with output signals of turntable. Then we can figure out the dynamic error of FOG that gives the quantitative determination to the dynamic characteristics of FOG. The turntable satisfies precision requirement of the experiment. However, the turntable itself is a kind of mechanical instrument, and it inevitably produces some jitter noises, which will generate output signals together with the output of FOG and influence dynamic test results. Thus it is necessary for us to use reasonable methods to filter out these noises so that we can extract accurately nonlinear errors within the output of FOG and analyze its dynamic characteristics. In this paper we use some filtering methods, which are weighted average method, finite impulse response (FIR) filter, infinite impulse response (IIR) filter and wavelet analysis method. They are used to filter the output signal of the turntable which can also be deemed as theoretical output of FOG and the output signal of FOG respectively in order to show the jitter noises in these two signals. It is clear to see that jitter noises exist in both signals. To highlight nonlinear error in the dynamic error, we apply these methods in filtering the theoretical and experimental output signals of FOG at the same time and make a comparison between them. From experiment results we can see the superiority of the wavelet analysis method. The dynamic signal in the test are nonstationary and their frequencies change with time. These changes can be divided into two parts that are slow-varying part and fast-varying part. The multi-resolution analysis in wavelet analysis can be used to show the slow-varying part and the fast-varying part with different resolutions. We apply the filter based on wavelet analysis to the signal through choosing an appropriate wavelet base and corresponding decomposition level. The experimental results show that wavelet analysis can effectively filter out the high-frequency jitter noises of the dynamic output signal, and retain the component which we need. It provides an efficient filtering method to accurate analysis of dynamic error of FOG.