水下滑翔机依靠净浮力和姿态调节即可实现空间运动, 是一种新兴的海洋探测机器人。该文面向水下定点探测任务, 开展滑翔机控制参数优化方法的研究。首先, 以典型水下滑翔机为研究对象, 建立了整机动力学模型, 同时建立了滑翔机下潜运动的能耗模型。在此基础上, 利用动力学仿真获取样本点, 采用四阶多项式建立了以控制参数为输入, 以滑翔机到达目标深度的能耗、运动时间和水平位移为输出的代理模型。之后, 将控制参数作为优化设计变量, 以最小化能耗和运动时间为优化目标, 利用水平位移构造约束条件, 建立优化数学模型。采用代理模型参与优化迭代计算, 确保优化计算效率。最后, 利用第二代非劣排序遗传算法(NSGA-Ⅱ)求解上述优化问题, 得到控制参数的Pareto最优解集。数值算例证明了所提出方法的正确性, 可用于指导实际探测任务的控制参数配置。

As a novel ocean exploration robot, an underwater glider can achieve space motion by adjusting its net buoyancy and attitude. In some exploration missions, the underwater glider must reach a specific location for virtual mooring and perform a fixed-point exploration, including the health monitoring of underwater equipment. The typical research aim is for the glider to reach the target exploration area at the earliest by consuming the minimum possible energy. To achieve this goal, the optimal control parameter configuration of the underwater glider must be determined. Therefore, this paper proposes the control parameter optimization method of underwater gliders for fixed-point exploration missions based on the dynamic theory, surrogate model technology, and multi-objective optimization algorithm. First, considering a typical underwater glider as the research object, this paper establishes the whole glider dynamic model using the Newton-Euler method. This dynamic model contains eight degrees of freedom and considers the effects of seawater density variation and hull deformation on the glider's net buoyancy. Considering the energy consumption of buoyancy adjustment, attitude adjustment, control, and measurement systems, the energy consumption model of the glider diving motion is established. On this basis, the sample points are obtained using an optimal Latin hypercube experimental design and dynamic simulation, and subsequently, the surrogate models are established using a quartic polynomial to fit the obtained sample points. Here, the input parameters of the quartic polynomial are the amounts of glider net buoyancy adjustment and movable internal mass block translation, and the output parameters are the energy consumption, diving motion time, and horizontal displacement of the glider to reach the target depth. Next, a mathematical optimization model is proposed. Specifically, the glider control parameters are selected as the optimization design variables; the optimization objective is to minimize the glider energy consumption and the diving motion time, simultaneously, and the horizontal displacement is used to construct the constraint. The surrogate models are employed to participate in the optimization calculation, which can improve the calculation efficiency. Finally, the non-dominated sorting genetic algorithm Ⅱ is used to solve the abovementioned optimization problem. A numerical example is provided to validate the proposed optimization method. After optimization calculation, the Pareto optimal set is obtained, consisting of 74 sets of non-dominated solutions of control parameter values. The analysis results illustrate that once the target depth has been determined, the glider horizontal displacement shows an obvious difference under different control parameter values, implying that the glider can employ different control parameter configurations to perform underwater fixed-point exploration missions. Under a specific target depth, the quartic polynomial can accurately describe the mapping relationship among the net buoyancy adjustment amount, movable internal mass block translation amount, glider energy consumption, diving motion time, and horizontal displacement. Besides, the functional relationship between the glider control and performance evaluation parameters shows obvious nonlinearity and nonmonotonicity. Optimization results of the control parameters demonstrate a contrasting relationship between the energy consumption and the diving motion time of the glider. For practical engineering missions, the selection rule of the optimal solution is listed, and the optimization results are verified via dynamic simulation. On the basis of the dynamic theory, surrogate model technology, and multiobjective optimization algorithm, the proposed optimization method exhibits high calculation efficiency and can be used for guiding the glider control parameter configuration in actual fixed-point exploration missions. Besides, this optimization method is versatile and can be used in various types of underwater gliders.