基于模糊RBF神经网络的水下机械臂控制研究

针对水下复杂工作环境下机械臂控制性能易受影响，而传统控制方法效果不佳的问题，提出了一种基于模糊RBF（radial basis function，径向基函数）神经网络的智能控制器，用于精确、稳定地控制水下机械臂。考虑到在水扰动环境下，机械臂通常受到附加质量力、水阻力和浮力的影响，运用拉格朗日法和Morison方程，建立包含水动力项的二杆机械臂动力学模型，通过模糊RBF神经网络对水下机械臂动力学方程中的水动力不确定项进行总体识别并拟合，利用模糊系统启发式搜索和RBF神经网络推理速度较快的优点，使水下机械臂系统具有较高的控制精度和较强的自适应性。考虑到水动力项，采用Lyapunov稳定性理论验证了水下机械臂系统的稳定性。最后利用MATLAB对二杆机械臂进行轨迹跟踪控制仿真实验，并对比模糊RBF神经网络与常规RBF神经网络识别方法和传统模糊控制方法的控制效果。仿真结果表明：与常规RBF神经网络识别方法相比，模糊RBF神经网络控制下二杆机械臂关节1的响应时间缩短了91%，相对误差减小了88%，关节2的响应时间缩短了92%，相对误差降低了77%；与传统模糊控制方法相比，关节1的相对误差减小了65%，关节2的相对误差减小了10%。研究结果表明模糊RBF神经网络的控制效果优于常规RBF神经网络识别方法和传统模糊控制方法，可为水下机械臂的控制提供一种精度较高、较有效的方法。

Research on control of underwater manipulator based on fuzzy RBF neural network

Aiming at the problem that the control performance of the manipulator is easily affected in the complex underwater working environment, but the traditional control method is not effective, an intelligent controller based on the fuzzy RBF (radial basis function) neural network method is proposed to control the underwater manipulator precisely and stably. In the water disturbance environment, the manipulator was usually affected by the additional mass, water resistance and buoyancy. The Lagrange method and Morison equation were used to establish a dynamics model of the two-bar manipulator, which included those hydrodynamic terms mentioned above. By using the fuzzy RBF neural network, the hydrodynamics uncertainties in the dynamics equation of the underwater manipulator were identified overall and fitted. With the advantages of heuristic search of fuzzy system and high reasoning speed of RBF neural network, the control performance of underwater manipulator system had better precision and strong adaptability. Considering the hydrodynamic terms, the stability of the underwater manipulator system was proved by using Lyapunov stability theory. Finally, the simulation experiment of trajectory tracking control for a two-bar manipulator was carried out by using MATLAB, and the control effect of fuzzy RBF neural network, conventional RBF neural network identification method and traditional fuzzy control method was compared. The simulation results showed that compared with the conventional RBF neural network identification method, under the control of fuzzy RBF neural network, the response time and relative error of joint 1 of the two-bar manipulator was reduced by 91% and 88%, the response time and relative error of joint 2 was reduced by 92% and 77%; compared with the traditional fuzzy control method, the relative error of joint 1 and joint 2 was reduced by 65% and 10%. The research results show that the control effect of fuzzy RBF neural network is better than that of conventional RBF neural network identification method and traditional fuzzy control, which can provides a high precision and effective control method for the underwater manipulator control.