Error compensation of photoelectric encoder based on improved BP neural network

Author(s):  
Xiao-gang Wang ◽  
Tao Cai ◽  
Fang Deng ◽  
Li-shuang Xu
2015 ◽  
Vol 740 ◽  
pp. 120-126
Author(s):  
Zhi Peng Zhang ◽  
Kang Liu ◽  
Feng Guo

In order to improve the process precision of the machine tool, further development of SVMR was achieved by QT Creator. Support vector machine was applied to the ARM11 development board, SVMR model was online trained and real-time predicted the values of machine tool thermal error. Compared with the widely used BP neural network, this method has the characteristics of high compensation precision and strong generalization ability. Experiment research has proved that the stronger effectiveness and higher accuracy using this method.


Optik ◽  
2016 ◽  
Vol 127 (8) ◽  
pp. 4083-4088 ◽  
Author(s):  
Bing Wu ◽  
Shaojun Han ◽  
Jin Xiao ◽  
Xiaoguang Hu ◽  
Jianxin Fan

2009 ◽  
Vol 407-408 ◽  
pp. 140-145
Author(s):  
Xu Ming Pei ◽  
Jie Liu ◽  
Chao Zhang

It were researched that the modeling methods of machine accuracy and the control techniques of the error compensation based on BP neural network(BPNN) for parallel machine tool(PMT)with five degrees of freedom(DOF). The samples are obtained to train the BP neural network which has good capacity for non- liner mapping, learning and generalization. The machine accuracy mathematics model is established for the error compensation, in order to study the nonlinear input and output problem of the parallel machine which difficultly modeling described. The trained neural network was applied to error compensation of PMT to realize modifying errors real-timely. Finally, simulation analysis was performed through the MATLAB software. The results expressed that the control strategies for error compensation were simple, efficient and practicable. Machine accuracy can be increased greatly after compensation.


2018 ◽  
Vol 2018 ◽  
pp. 1-10
Author(s):  
Wei Shao ◽  
Peng Peng ◽  
Awei Zhou ◽  
Quanquan Zhu ◽  
Di Zhao

In view of the high precision requirement for mechanical structure of aeronautical blade measuring system, this paper proposes a laser interferometer to measure the error of the spatial nodes of the measuring system based on a comprehensive analysis of domestic and foreign error compensation methods for the measuring system. The optimized algorithm backpropagation (BP) neural network (OA-BPNN) compensation method is utilized to adaptively compensate for the systematic error of the mechanical system. Compared with the traditional polynomial fitting and genetic algorithm BP neural network (GA-BPNN) algorithm, the results show that the OA-BPNN algorithm is characterized by the best adaptability, precision, and efficiency for the adaptive error compensation. The spatial errors in the XYZ directions are reduced from 10.9, 60.1, and 84.2 μm to 1.3, 4.0, and 2.4 μm, respectively. The method is of great theoretical significance and practical value.


Sign in / Sign up

Export Citation Format

Share Document