摘要
介绍了模拟人类皮肤结构而设计的光纤智能皮肤触觉传感器 ,它输出信号为多路和非线性的 ,采用传统方法处理有一定的困难 .为此 ,利用了准共路光路补偿方法对信号进行预处理 ,分析表明在一定程度上可以消除光源起伏和外界光路随机微弯损耗以及温度变化所带来的干扰 .又采用人工神经网络对其进行信号处理 .为了简化网络 ,探测面被划分成 9个网格区域 ,并联合使用了Kohonen算法和BP模型 .测试的结果表明 ,对施加在智能皮肤表面上任意一点上的外加压力 ,可以分辨出外加压力的大小和位置 .
An artificial neural network was used to process.The mult1idimensional and nonlinear outputs of the sensor array. It's difficult to process the signal in traditional way.Pretreated by a quasi_syn_optical path compensating method, which eliminates the interferences resultign from fluctuating voltage and temperature. To simplify the network, skin was divided into nine sections,and both kohonen arithmetic and BP model were used . Results show that when an intelligent skin was pressed, the sensor can reliably estimate teh magnitude of pressure value and its position.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
2002年第2期82-86,共5页
Journal of Harbin Engineering University
关键词
光纤智能皮肤
人工神经网络
触觉传感器
fiber_optical intelligent skin
artificial neural network
tactile sensor
