基于组合优化算法的ROV控制器参数优化仿真

ROV Controller Parameter Optimization Simulation Based on Combinatorial Optimization Algorithm

为解决有缆水下机器人控制器设计中多参数优化效率低、拟合响应面模型复杂的问题,本文提出了田口法与序列二次规划法结合的优化方案。应用田口正交实验在全局选定初值,将初值代入序列二次规划法进行局部精确收敛。采用田口法的组合优化方案,对有缆水下机器人控制器参数进行优化。仿真验证结果表明,优化后的有缆水下机器人跟踪螺旋路径效果良好,鲁棒性良好;相较于拉丁超立方的组合优化方案,田口法的组合优化方案迭代次数降低了70次,目标函数缩小了23.2%;相较于田口法,速度超调量减小,更具鲁棒性,该组合算法可高效地解决ROV多参数优化问题。

In order to solve the problems of the low efficiency of multi-parameter optimization and the complexity of fitting response surface model in ROV controller design,a combinatorial optimization scheme combining Taguchi method and sequential quadratic programming method is proposed in this paper.Taguchi orthogonal table experiment is used to select the initial value globally,and the initial value is substituted into the sequential quadratic programming method to achieve precise local convergence.The Taguchi combinatorial optimization scheme is applied to the parameter optimization of the ROV controller.The simulation results show that the optimized ROV tracking spiral path achieves good performance and robustness.Compared with the Latin hypercube combinatorial optimization scheme,the iterative times of Taguchi method's combinatorial optimization scheme are reduced by 54.5%and the objective function is reduced by 23.2%.Compared with traditional Taguchi method,the velocity overshoot is reduced by 60%and is also more robust.This combined algorithm can efficiently solve the ROV multi-parameter optimization problem.