摘要
在进行载荷谱时域外推时,为更好地估计超阈值载荷数据所服从的广义帕累托分布(Generalized Pareto Distribution,GPD)参数,分别对比极大似然估计(Maximum Likelihood Estimation Moments,ML)、矩估计(Method of Moments,MoM)及概率加权估计(Probability Weighted Moments,PWM)对GPD分布参数选取的影响,并选取最优参数估计方法。首先搭建了挖掘机压力测试系统,并采集挖掘机在典型工况工作下载荷数据;其次以最小均方根误差所对应值为最优阈值,分别计算3种参数估计方法所对应最优阈值及GPD拟合参数;最后分别绘制3种参数估计方法下超阈值载荷数据和GPD拟合CDF图和Q-Q图。结果表明:3种参数估计方法的拟合结果和超阈值样本之间的相关系数均在0.99以上,都可以较好地拟合超阈值载荷数据的GPD分布,但PWM参数估计方法能够包含较多有效载荷数据且拟合效果更好。
To investigate estimate the Generalized Pareto Distribution(GPD)parameters of the super threshold load data when the load spectrum is extrapolated in time domain,the GPD parameters of the super threshold load data are obtained.Compare the influences of Maximum Likelihood estimation moments(ML),the Method of Moments estimation(MoM)and the Probability Weighted Moments estimation(PWM)on the parameter selection of GPD distribution respectively and the optimal parameter estimation method is selected.Firstly,we built the excavator pressure test system and collect the load data of the excavator under typical working conditions.Secondly,the value corresponding to the minimum root mean square error is taken as the optimal threshold,and the optimal threshold and GPD fitting parameters corresponding to the three parameter estimation methods are calculated respectively.Finally,the above threshold load data and GPD fitting CDF and Q-Q plots are respectively drawn under the three parameter estimation methods.The results show the correlation coefficients between the fitting results of the three parameter estimation methods and the super-threshold samples are above 0.99,which can fit the GPD distribution of the super-threshold load data well,but the PWM parameter estimation method can contain more payload data and has a better fitting effect.
作者
王普长
孙辉
陈晋市
张淼淼
何春晖
WANG Pu-chang;SUN Hui;CHEN Jin-shi;ZHANG Miao-miao;HE Chun-hui(Jiangsu XCMG Construction Machinery Research Institution Ltd. , Xuzhou, Jiangsu 221000;School of Mechanical and Automotive Simulation and Control, Jilin University, Changchun, Jilin 130025)
出处
《液压与气动》
北大核心
2021年第9期38-43,共6页
Chinese Hydraulics & Pneumatics
基金
国家重点研发计划(2018YFB2000900)。
关键词
载荷谱
广义帕累托分布
拟合优度检验
参数估计方法
load spectrum
Generalized Pareto Distribution
goodness of fit
parameter estimation method
作者简介
王普长(1952—),男,河南新乡人,工程师,硕士,主要研究方向为工程机械液压系统及元件的测试与可靠性。
