基于SG算法的燃煤微分热重曲线平滑去噪

Coal DTG curve denoise based on Savitzky-Golay method

对燃煤微分热重曲线进行平滑去噪处理是获得煤的燃烧特性和热解反应动力学的参数的重要工作之一.运用SG(Savitzky Golay)算法进行燃煤DTG数据平滑,指出影响平滑效果的主要因素是数据窗口大小、拟合阶数和平滑次数.选用了标准差、信噪比、中位数、极差、四分位间距、李氏指数和时移等作为评价指标,运用最大熵方法对平滑去噪前后的信号进行了频谱分析,对影响平滑效果的因素进行了评判后指出,数据窗口取15左右,拟合阶数取3左右,平滑次数取2左右,可得到较好效果.

Denoise of coal derivative thermogravimetry (DTG) signal was studied with SG (Savitzky-Golay) method. The data window, order of polynomial and time of smoothing had an immediate effect on the result. Six parameters (standard deviation, range, inter quartile range, signal/noise, Lipschitz exponent and time shifting) were regarded as evaluating indictors to estimate the effect of denoise. Max entropy method was used in signal power spectrum estimation. The result shows that the preferable result could be obtained when the date window, order of polynomial and time of smoothing equals 15,3 and 2 or so respectively.