基于神经网络的多维力传感器静态解耦的研究

提出了基于人工神经网络进行多维力传感器静态解耦的方法。维间耦合是制约多维力传感器测量精度的主要因素，为克服传统线性解耦方法的局限性，利用BP神经网络的强非线性逼近能力研究了多维力传感器的非线性静态解耦。以研制的微型5维指尖力／力矩传感器为对象进行了解耦实验，结果表明，与基于最小二乘的线性解耦方法相比，提高了解耦精度。

Study on Static Decoupling of Multi-Axes Force Sensor Based on Neural Network

A static decoupling method of multi axes force sensor based on ANN (Artificial Neural Network) is proposed in this paper. The coupling problem is an important factor limiting measuring accuracy of multi axes force sensor. In order to overcome the limitation of traditional linear decoupling, a nonlinear static decoupling method is designed based on BP neural network which possesses strong nonlinear approaching ability. The decoupling experiment is implemented by using a miniature five axes fingertip force sensor and the experimental results show that the method greatly improves the decoupling accuracy in comparison with traditional linear decoupling based on LS (Least Square).