江虹 周上清 劉鵬輝 鄭曉丹



摘 要: 為提高超弱光纖光柵反射信號(hào)光電檢測電路的測量精度,提出一種新的變步長LMS算法去除光電探測器信號(hào)中的高斯噪聲。針對基于對數(shù)函數(shù)的變步長LMS(Least Mean Square)算法其步長因子易受噪聲干擾,導(dǎo)致其步長因子取值在最優(yōu)解附近波動(dòng)的問題。根據(jù)高斯白噪聲相關(guān)性差的特性,使用相關(guān)特性改進(jìn)步長因子的更新公式,降低不相關(guān)噪聲對穩(wěn)態(tài)誤差的影響。通過理論分析和MATLAB軟件仿真結(jié)果對比分析改進(jìn)前后收斂速度和穩(wěn)態(tài)誤差,本算法在低信噪比條件下穩(wěn)態(tài)誤差減小三分之一,收斂速度更快,具有更好的抗噪、濾波性能,有效抑制了超弱光纖光柵反射信號(hào)檢測的噪聲。
關(guān)鍵詞: LMS算法;變步長;相關(guān)特性;超弱光纖光柵
中圖分類號(hào):TN? 文獻(xiàn)標(biāo)識(shí)碼:A? 國家標(biāo)準(zhǔn)學(xué)科分類代碼:510.1050
Study on An improved step size factor of variable LMS algorithm
Jiang Hong ZhouShangqing Liu Penghui Zheng Xiaodan
Changchun university of technology JilinChangchun 130000
Abstract: In order to improve the measurement accuracy of the photoelectric detection circuit of the ultra-weak FBG reflection signal.A new variable step LMS algorithm is proposed to remove Gaussian noise from photodetector signal.Aim at the problem that the step size factor of variable LMS algorithm based on logarithmic function is easy to be disturbed by noise,lead to the step size factor waveing among the optimal solution.According to the bad correlation of white Gaussian noise,using correlation characteristic to improve the update formula of step size to reduce the impact of noise.The theory analyze and simulation comparative analysis of convergence rate and steady-state error before and after improvement show that this algorithms steady-state error reduced by one third and has better ability of resist noise and smoothing in low SNR environment.The noise of the reflection signal detection of the ultra weak fiber grating is effectively suppressed.
Key words: LMS algorithmstep;size factor of variable;correlation characteristics;ultra-weak FBG
隨著軍事、大型工程、工業(yè)等領(lǐng)域?qū)τ跍y量規(guī)模的要求日益增加,傳統(tǒng)的光纖布拉格光柵復(fù)用數(shù)量少已成為首先需要解決的關(guān)鍵問題[1-2]。超弱光纖光柵傳感器作為一種新型的光纖光柵傳感器,由于其反射率低的特點(diǎn)使其復(fù)用率高,可以實(shí)現(xiàn)大規(guī)模復(fù)用,但也因此其反射信號(hào)非常微弱,如何防止噪聲干擾提取是目前研究的熱門方向。目前在實(shí)際的測量中主要使用光電轉(zhuǎn)換電路[3]來對微弱反射信號(hào)進(jìn)行采集和放大,其中影響測量準(zhǔn)確度的主要為來自器件和環(huán)境的高斯白噪聲。
LMS算法是高精度測量中削減高斯白噪聲的重要方法,該算法……