An optimized BP neural network based on genetic algorithm for static decoupling of a six-axis force/torque sensor

2017 International Conference on Sensors, Materials and Manufacturing

工业技术;材料科学;机械制造

Fu, Liyue^1 ; Song, Aiguo^1

School of Instrument Science and Engineering, Southeast University, Nanjing, China^1

关键词: 6-axis force/torque sensors;    BP neural networks;    Decoupling algorithms;    Decoupling results;    GA-BP algorithms;    Measurement precision;    Pseudo inverse matrix;    Six-axis force/torque sensor;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/311/1/012002/pdf
DOI  :  10.1088/1757-899X/311/1/012002

来源: IOP

【 摘 要 】

In order to improve the measurement precision of 6-Axis force/torque sensor for robot, BP decoupling algorithm optimized by GA (GA-BP algorithm) is proposed in this paper. The weights and thresholds of a BP neural network with 6-10-6 topology are optimized by GA to develop decouple a six-Axis force/torque sensor. By comparison with other traditional decoupling algorithm, calculating the pseudo-inverse matrix of calibration and classical BP algorithm, the decoupling results validate the good decoupling performance of GA-BP algorithm and the coupling errors are reduced.