基于遗传小波神经网络的机器人腕力传感器动态建模研究

• 作 者: 俞阿龙
• 机 构: 淮阴师范学院电子与电气工程系 淮安223300
• 出 处: 物理学报
• 发表日期: 2008年06期
• 关键词: 腕力传感器 动态建模 小波神经网络 遗传算法

提出一种基于改进遗传算法进化小波神经网络用于机器人腕力传感器动态建模的新方法,介绍了该算法原理.该方法利用腕力传感器的动态标定数据,用改进的遗传算法来优化小波神经网络结构和参数,建立腕力传感器的动态模型.结果表明,采用遗传小波神经网络进行腕力传感器动态建模,能克服误差反向传播算法存在易陷入局部极小点的缺点,网络的复杂度、收敛性和泛化能力得到了好的综合,建模的速度和精度得到提高.

This paper presents a new method based on the improved genetic algorithm and its application to the dynamic modeling of the robot wrist force sensor.The principle of the algorithm is introduced.The dynamic calibration data of the wrist force sensor is used to optimize the wavelet neural network structure with the improved genetic algorithm The dynamic model of wrist force sensor is established.The results show that using WNN to make dynamic modeling of wrist force sensor can overcome the shortcoming that error back propagation algorithm is easy to fall into local minimum point, the complexity of network, convergence and Generalization ability has been well integrated, modeling speed and accuracy are improved.