基于改进型迭代神经网络的三自由度并联机构位置正解分析

Forward Position Kinematics Analysis of Three Degree-of-Freedom Parallel Manipulator Based on Improved Iterative Neural Network

针对模拟船舶在海上运动的摇摆姿态及重载的要求,根据并联机器人机构综合理论,提出运用4SPS-1S结构的并联机构实现模拟船舶在重载情况下的摇摆运动。通过对该机构的运动学分析,推导出三自由度并联机构位置逆解的解析表达式。考虑到位置正解的解析解难以求出,运用一种改进型、高效率的迭代神经网络对其位置正解模型进行求解,最后借助Matlab软件对位置正解模型进行了仿真研究。研究结果表明:该改进型迭代BP神经网络不但性能上优于普通的BP神经网络和误差补偿函数为f(ε)=ε的迭代BP神经网络,且所建立的位置正解模型可以满足实时控制要求。

For the simulation of the ship swing orientation and heavy load demand, 4 SPS-1S three degree-of-freedom parallel manipulator was used according to the theories of the structure synthesis of parallel robotic mechanisms to realize simulation of the ship＇s swing movement under heavy load. Through analysis of the kinematics to this mechanism, the inverse kinematics formula for its posi- tion was deduced. An efficient improved type of iterative neural network was used to research the forward kinematics problem, because the analytical solution of forward kinematics was difficult to be solved. The research results show： not only the improved iterative BP neural network is superior in performance than the common BP neural network and the iterative BP neural network with error compensation function as f（ε） = ε, but also the real-time control requirements can be met by the forward kinematics model.