水下机械臂运动空间分析与轨迹跟踪算法优化

针对水下机械臂动力学模型不确定和未知外界干扰问题，采用基于HJI理论的径向基函数神经网络自适应控制算法对水下机械臂进行控制。首先，以水下六自由度机械臂为例，基于D-H法则对水下机械臂的运动学进行分析，通过仿真验证该方法的正确性；接着，基于蒙特卡洛法构建水下六自由度机械臂的运动空间云图，真实反映水下机械臂的运动空间；然后，以二自由度水下机械臂为例，设计基于HJI理论的RBF神经网络自适应控制器，利用神经网络的万能逼近原理逼近不确定干扰项，考虑到神经网络逼近存在误差，将逼近误差看作外界干扰项并通过HJI理论对逼近误差在线评价，评价系统对干扰项的抑制能力，并采用自适应算法在线估计网络权值，加快系统收敛；最后，通过仿真可知，该机械臂能较好地完成轨迹跟踪。

Underwater Manipulator Motion Spatial Analysis and TrackingAlgorithm Optimization

Anadaptive control algorithm, based on Hamiltion-Jacobi Inequality (HJI) radial basis function(RBF) neural network, was introduced tocontrol the robotic armunderwaterwhere uncertain dynamic model and unknown external interference mayoccur. First, an underwater six-degree-of-freedom robotic arm was taken as an example. The kinematics of the underwater mechanical arm was analyzed based on the DH rule, and the correctness of the method was verified by simulation. Then, a six-degree-of-freedom underwater robot was constructed based on the Monte Carlo method. The cloud image of the arm’s motion space truly reflected the motion space of the underwater robotic arm. Finally, a two-degree-of-freedom robotic arm was taken as another example, and an RBF neural network adaptive controllerwas designed based on HJI theory.The universal approximation principle of the neural network was applied to uncertain interference terms. Due to the existence of errors in the neural network approximation, the approximation errors were taken as external interference terms and the approximation errors online were evaluatedusing HJI theory, the system’s ability to suppress the interference termswas also evaluated,and an adaptive algorithm was applied to estimate the network weight online to accelerate the system convergence.It concludes that the trajectory tracking can be achieve better through simulation.