大量程柔性铰六维力传感器静态解耦的研究

为提高大量程六维力传感器的测量精度,提出了一种新型的六维力传感器非线性静态解耦方法,该方法结合混合递阶遗传算法和小波神经网络的优点,采用递阶遗传算法与最小二乘法分别对小波神经网络隐层结构参数以及输出层权值进行优化,再将优化后的小波神经网络模型用于六维力传感器非线性解耦.建立了基于混合递阶遗传算法和优化小波神经网络的六维力传感器非线性解耦模型,设计了基于混合递阶遗传算法的小波神经网络结构及参数优化算法,给出了六维力传感器非线性解耦的具体实现流程.以最新研制的6-UPUR大量程柔性铰六维力传感器为对象进行实验,结果表明,采用该方法六维力传感器的Ⅰ类误差和Ⅱ类误差分别为1.25％和2.59％,比采用BP和RBF神经网络方法的测量精度高.

Research on static decoupling of large range flexible joint six-axis force sensor

A new six-axis force sensor nonlinear static decoupling method that has the advantages of hybrid hierarchy genetic algorithm and wavelet neural network is proposed to improve the measurement accuracy of large range six-axis force sensor.Hierarchical genetic algorithm and least-squares method are respectively used to optimize the hidden layer structure parameters and output layer weights of wavelet neural network,and the optimized wavelet neural network model is applied to six-axis force sensor nonlinear decoupling.A six-axis force sensor nonlinear decoupling model based on the optimized wavelet neural network model is established,an algorithm based on hybrid hierarchy genetic algorithm is designed to optimize the structure and parameters of wavelet neural network.The nonlinear decoupling realization process of the six-axis force sensor is given.Experiment was carried out on a 6-UPUR large range flexible joint six-axis force sensor recently designed,and the experimental results show that the class I errors and class II errors,measured using the method proposed in this paper,are 1.25% and 2.59%,respectively;the measurement accuracy is better than those using the methods of BP and RBF neural networks.