机器人柔顺装配的BP神经网络系统

BP Neural Network System for Robotic Compliance Assembly Process

设计了用于机器人装配的主被动复合柔顺手腕,简化了手腕结构,缩短了搜索时间,提出了一种称为“一维搜索”的机器人装配作业主被动复合柔顺策略,将主动搜索自由度降低到一维,同时最大限度地利用被动柔顺。设计了以装配力信号为输入的BP神经网络,建立了机器人复合柔顺装配作业的BP神经网络系统,收到良好的装配效果。

To simply the wrist structure, lessen then calculation of control system and shorten the search time, the wrist of active-passive compound compliance is developed. A new strategy named one dimension search for active-passlve compound compliance is presented. The core of this strategy is that the degrees of freedom of active search are reduced to one dimension, meanwhile the passive compliance is employed as much as possible. BP neutral network controller is designed, which receives the assembly force signals. BP neutral network system of active-passive compound compliance in the robotic assembly process is setup, and good result is obtained.