遥控水下机器人模型辨识及预测控制算法研究

Research on Adaptive Model Predictive Control Algorithm for Remotely Operated Vehicle

针对遥控水下机器人(remotely operatedvehicle,ROV)因非对称结构布局、重心和浮心位置变化等因素造成的运动状态耦合和模型非线性问题,文章提出了一种考虑ROV典型非线性特性的在线模型辨识方法,研究并推导了基于所辨识模型的多维预测控制(model predictive control,MPC)算法及其基于算子分割二次规划的快速优化求解方法,搭建了包含ROV非线性数学模型、模型参数辨识算法模块和MPC算法模块的算法仿真系统。通过在辨识模型中分别考虑非线性特性和不考虑非线性特性,改变ROV数学仿真模型的ROV浮心位置参数,对比仿真验证了本文模型辨识算法的有效性和先进性;在上述辨识模型的基础上,对比仿真分析了4种情况下ROV的姿态控制效果,即MPC算法依托的辨识模型是否考虑ROV典型非线性特性,以及基于直接求解MPC算法和本文所提出的基于快速优化近似求解的MPC算法,验证了本文所提控制算法的有效性。

Aiming at the motion state coupling and model non-linearity of remotely operated vehicle(ROV)caused by the asymmetric structural layout of ROV and changes of center of gravity and center of buoyancy,this paper proposes a model identification algorithm considering the typical nonlinear characteristics of ROV.A model predictive control(MPC)method and its fast optimization solution method based on operator partition quadratic programming were studied and deduced.A mathematical simulation system including the nonlinear mathematical model of ROV,the parameter identification algorithm module and the MPC algorithm module were developed.The simulation was carried out under the scenarios of considering and ignoring nonlinear characteristics,and changing the position of center of buoyancy in the ROV nonlinear mathematical model.The effectiveness and advancement of the parameter identification algorithm proposed in this paper were verified.On the basis of the identification model,the simulation compared and analyzed the effect of underwater ROV attitude control under 4 scenarios,including the MPC algorithm-based identification model considering and ignoring the typical nonlinear characteristics of underwater ROV,MPC algorithm based on direct solution and MPC algorithm based on fast optimization approximate solution proposed in this paper.The effectiveness of the control algorithm proposed in this paper is verified.