基于模糊熵加权融合的单轨运输机器人动态倾角辨识研究  

Research on Dynamic Inclination Angle Identification of Monorail Transportation Robot Based on Fuzzy Entropy Weighted Fusion

作　　者：刘泽朝 李敬兆[1] 郑昌陆 王国锋 LIU Zechao;LI Jingzhao;ZHENG Changlu;WANG Guofeng(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan,Anhui 232001,China;Shanghai Shenchuan Electric Co.,Ltd.,Shanghai 201800,China;Huaihe Energy Holding Group Co.,Ltd.,Huainan,Anhui 232001,China)

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001 [2]上海申传电气股份有限公司,上海201800 [3]淮河能源控股集团有限责任公司,安徽淮南232001

出　　处：《计量学报》2024年第7期941-951,共11页Acta Metrologica Sinica

基　　金：国家重点研发计划(2020YFB1314100);国家自然科学基金(52374154);安徽理工大学博士研究生创新基金(2022CX1008)。

摘　　要：针对现有传感器检测单轨运输机器人动态倾角时存在辨识精度低的问题,提出了一种基于模糊熵加权融合的单轨运输机器人动态倾角精准辨识方法。首先,基于构建的轨道曲率与倾角变化双模型,通过改进的遗忘递归最小二乘(IFFRLS)算法分别计算出轨道曲率值和倾角动态变化率;其次,再以轨道曲率值和倾角动态变化率作为输入值,通过无迹卡尔曼滤波(UKF)和扩展卡尔曼滤波(EKF)算法不断迭代更新分别计算出倾角动态角度;最终,采用全局模糊熵加权融合(GFEWF)将各角度值深度融合,提高动态倾角的检测精度。实验表明,基于双模型的全局模糊熵加权融合(GFEWF)与基于单一模型的UKF或EKF算法相比,其辨识的动态倾角精度在轨道段1和轨道段2分别提升了10.38%和25.60%。For the problem of low identification accuracy in detecting the dynamic inclination angle of monorail robots,a precise identification method for the dynamic inclination angle of monorail transport robot based on fuzzy entropy weighted fusion is proposed.Firstly,based on the constructed dual model of orbit curvature and inclination angle change,the improved forgetting recursive least squares(IFFRLS)algorithm is used to calculate the dynamic change rate of orbit curvature and inclination angle respectively.Secondly,taking the orbit curvature value and the dynamic change rate of inclination angle as input values,the extended Kalman filter(EKF)and unscented Kalman filter(UKF)algorithms are used to iteratively update and calculate the dynamic angle of inclination angle respectively.Finally,the global fuzzy entropy weighted fusion(GFEWF)is used to deeply fuse the angle values to improve the detection accuracy of dynamic inclination angle.The experiments show that the global fuzzy entropy weighted fusion(GFEWF)algorithm based on double model improves the identified dynamic inclination accuracies in rail segment 1 and rail segment 2 by 10.38%and 25.60%on average,respectively,compared with the single model-based UKF or EKF algorithms.

关 键 词：几何量计量 单轨运输机器人 动态倾角 递归最小二乘算法 无迹卡尔曼滤波 全局模糊熵 

分 类 号：TB922[一般工业技术—计量学]