二维多关节机器鱼远场流速和攻角预测研究

Study on Prediction of Far-Field Velocity and Attack Angle of Two-Dimensional Multi-Joint Robotic Fish

远场流速和攻角预测是多关节机器鱼实现水下自主巡航与控制的前提,本文通过理论分析、流体力学仿真及估计的方法对此问题进行了研究。首先,采用势流理论和伯努利方程分析了摆动工况下Joukowski对称翼型的表面压力,发现翼型表面压力与远场流速、攻角和摆动频率有关。其次,以Joukowski翼型为基础建立二维多关节机器鱼几何模型,并给出了鱼身波动方程。再次,基于鱼体均布的多个压力监测点,定义了无量纲化压力系数,并根据各监测点组合形式提出了四种测量模型。通过计算流体力学(Computational Fluid Dynamics,CFD)技术分析了多关节机器鱼摆动频率固定情况下各测量模型输出结果与远场流速和攻角之间的关系,采用最小二乘法确定了四种测量模型输出的回归方程,并作为观测方程。最后,给出了由远场流速和鱼体转动角度构成的状态方程,进行数值计算,结果表明各测量模型均可以有效追踪实际流场状态,其中多测量模型组合方式对流场预测效果要好于单个测量模型。

The prediction of far-field velocity and attack angle is the prerequisite for a multi-joint robotic fish to achieve underwater autonomous control and cruise,and this problem is studied by theoretical analysis,fluid dynamic simulation and estimation methods in this paper.First,the potential flow theory and Bernoulli equation are used to analyze the surface pressure of the Joukowski symmetric airfoil under swing conditions,and it is found that the surface pressure of the airfoil is related to the far-filed velocity,attack angle and swing frequency.Second,the geometric model of a two-dimensional multi-joint fish robot is established on the basis of the Joukowski airfoil,and the fish body wave equation is given.Third,based on multiple pressure monitoring points,a dimensionless pressure coefficient is defined,and four measurement models are proposed according to the combination of each monitoring point.Through computational fluid dynamics(CFD),the relationship between the output results of each measurement model and the far-field velocity and attack angle under the condition of fixed swing frequency of the robotic fish is analyzed.The regression equations of the output of the four measurement models are established by the least-square method and used as the observation equations.Finally,an equation that takes the far-field velocity and the rotation angle of the fish body as the state is given,and the numerical calculations are carried out.The results show that each measurement model can effectively track the actual flow field state,and the combination of multiple measurement models has a better effect on the flow field prediction than that of the single measurement model.