【作者】 罗臻；

【导师】 葛升民；

【作者基本信息】 哈尔滨工业大学 ， 控制科学与工程， 2009， 硕士

【摘要】 作为惯性测量组合的基础测量单元,光纤陀螺的精度将决定惯性系统的精度,因而提高光纤陀螺的精度具有重要意义。光纤陀螺输出信号微弱,背景噪声强,由于噪声的干扰,光纤陀螺输出信号中包含着多种随机误差。本文的研究目的就是运用小波分析去噪的方法来减小陀螺信号中的随机误差。小波分析作为一种最新的时-频分析工具,在时域和频域都有表征信号局部特征的能力,具有多分辨率分析的特点,很适合处理非平稳信号。利用信号和噪声在各尺度下的小波变换系数特性不同的特点,对它们进行分离,从而可以达到去除噪声恢复信号的目的。由于可用于去噪的小波母函数是一个集合,在小波去噪的实际应用中采用哪一种小波函数才能得到最好的去噪效果,是本文的一个研究内容。小波去噪的另一个重要问题就是阈值的选取和量化问题,在去噪的过程中如何更有效地进行阈值选取和量化,使得在噪声被去除的同时尽可能的避免有用信号的丢失,是本文研究的另一个内容。同时,由于本文开放性研究的目的,还将研究原始信号长度与采样频率的变化,及小波消失距的变化对滤波效果的影响。本文首先介绍了小波分析的基本理论和小波去噪的原理与方法。重点研究了小波阈值去噪方法,对不同小波基、不同阈值选取规则、不同阈值量化方式及不同信号长度与采样频率下小波的去噪效果进行了比较和分析,研究了小波的消失矩阶数对去噪效果的影响。根据仿真实验结果,得到了最优小波滤波方法,并研究了提升方案(Lifting Scheme)对于计算速度及滤波效果的影响。最后,根据将选出的最优小波方法应用到了目标光纤陀螺上。更多还原

【Abstract】 As the basic measuring unit of inertial measurement units, the accuracy of Fiber Optic Gyroscope (FOG) will decide the inertia system’s precision. As a result, it is significant to improve the accuracy of FOG. Though, the signal of FOG is weak and the background noise is strong, so the output signal of FOG, which disturbed by the noises, contains many stochastic errors. The main purpose of this thesis is to reduce the stochastic errors in output signal of FOG by using wavelet analysis.As a new tool for time-frequency analysis, wavelet analysis has excellent localizing quality in time domain and frequency domain simultaneously, and the characteristic of multi-resolution. For this reason, wavelet analysis is suitable for processing the non-stationary signals. By the differences between the wavelet transform coefficients, noise can be filtered from the initial signals and useful signals can be restored.Because the wavelets are congregation, which one is better for denoising in the practical application is a research subject of this thesis. Another important problem of denoising by wavelet is the threshold selection and the threshold quantization method. How to select the proper threshold and the quantization method is another research subject of this thesis. In the meantime, as a result of the open ended purpose of this paper, the filtering effects with different length, sampling frequency in initial signals and different vanishing moments of wavelets will be researched.Firstly, the basic theory of wavelet analysis, the principles and methods of wavelet denoising are introduced in this thesis. Wavelet threshold denoising method is the key research point. The denoising effect by using different wavelets, thresholding rules, quantization methods, length and sampling frequency of signal are studied respectively. The number of vanishing moments of the wavelets has an effect on the denoising results, which is researched in this thesis too. By the experiment results, the optimal wavelet method is obtained. Then, Lifting Scheme has been used to compare its effect on computing speed and denoising. At last, this optimal wavelet method is realized in the FOG.更多还原