摘要
物流秤在动态测量中,压力传感器因秤体振动与货物冲击的干扰易产生相位缺失、频率跳动等信号缺失问题,经皮尔逊相关性检测发现压力与振动信号相关系数为0.94呈现高度相关。提出一种使用振动信号对缺失压力信号聚类后拟合的重构模型,首先通过GMM模型在样本集中寻找待重构样本的相关样本簇,基于相似样本簇建立GA-SVM对缺失值进行拟合重构。经测试该模型当压力信号缺失比率小于80%时,补偿误差可控低于5%;在各缺失比例下较未经聚类的GA-SVM、GA-BP重构模型精度提升明显。
In the dynamic measurement of logistics scale,due to the interference between the vibration of the scale and the impact of the goods,the signal missing problems such as phase loss and frequency jump of the pressure sensor are easy to occur.The Pearson correlation detection shows that the correlation coefficient of pressure and vibration signal is 0.94,showing a high correlation.In this paper,a reconstruction model based on vibration signal clustering is proposed.Firstly,GMM model is used to find the relevant sample clusters of the samples to be reconstructed in the sample set,and then GA-SVM is established to fit and reconstruct the missing values based on similar sample clusters.After testing,when the pressure signal missing ratio is less than 80%,the compensation error can be controlled less than 5%.Under each missing ratio,the accuracy of the reconstructed model is significantly improved than that of GA-SVM and GA-BP without clustering.
作者
史柏迪
庄曙东
蔡鸣
江志伟
SHI Baidi;ZHUANG Shudong;CAI Ming;JIANG Zhiwei(School of Mechanical Engineering,Hohai University,Changzhou 213022;Jiangsu Key Laboratory of Precision Instruments,Nanjing University of Aeronautics and Astronautics,Nanjing 213009;Mettler Toledo Measurement Technology Co.,Ltd.,Changzhou 213022)
出处
《计算机与数字工程》
2023年第2期286-291,313,共7页
Computer & Digital Engineering
基金
江苏省高校实验室研究会立项资助研究课题(编号:GS2019YB18)
江苏省精密与微细制造技术重点实验室数学建模课题组(编号:CZ520007812)
中央高校基本科研业务费(编号:2018B44614)资助。
作者简介
史柏迪,男,硕士研究生,研究方向:寿命预测;庄曙东,男,博士,副教授,硕士生导师,高级工程师,研究方向:智能制造;蔡鸣,女,工程师,研究方向:动态秤精度;江志伟,男,工程师,研究方向:动态秤。
