传感器动态补偿的神经网络逆系统方法  

A Novel Dynamic Compensating Method Based on ANN Inverse System for Sensors

作　　者：戴先中[1] 殷铭[1] 王勤[1] 

机构地区：[1]东南大学自动控制系,南京210096

出　　处：《仪器仪表学报》2004年第5期593-596,共4页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金资助项目

摘　　要：给出一种基于神经网络逆系统的传感器动态补偿策略。无需传感器具体模型和参数 ,即可实现传感器系统的近似单位线性化 ,达到动态补偿的目的。仿真实验和动态标定试验结果表明 ,应用这种新型的易于工程实现的动态补偿方法可显著地提高传感器的动态特性 。A dynamic compensating method based ANN (artificial neural networks) inverse system is proposed for linear or nonlinear sensor systems according to the theory of inverse methods and using excellent approach abilities of ANN. ANN inverse dynamic model of sensor is achieved through offline training the ANN that combines the static ANN and dynamic factors. The dynamic performance of sensor is improved effectively by cascading ANN inverse dynamic model with the original sensor. Simulation and experiment results indicate that the dynamic performance of robot six-axis wrist force sensor is remarkably improved and enhanced using the proposed method.

关 键 词：传感器系统 动态补偿 神经网络 逆系统方法 仿真实验 动态品质 线性化 试验结果 改善 

分 类 号：TH715[机械工程—测试计量技术及仪器] TP212[机械工程—仪器科学与技术]