仿生机器鱼高效率推进的预测/调节双闭环控制方法

Although the propulsion model and bionic mechanism of robotic fish have been extensively studied, the low efficiency which causes the short life time still remains the key problem in the development of robotic fish. Therefore, a prediction/regulation double closed-loop bionic control method is proposed for high-efficient propulsion. The control laws of the forecast planning and the feedback adjustment are deduced in the view of negative work. The feedback links of the robotic fish are designed to mimic the lateral line and the muscle spindle, so the double closed-loop control structure could be set up as the bionic prototype of the neuromuscular system to improve the efficiency of the robotic fish. Based on the high-efficient propulsion mechanism in the phenomenon of negative work, we will develop a variable-stiffness driven algorithm for the joints combining with the spindle model to give feedback of the work state, and we will design a virtual lateral line sensing system to identify the flow information. Then, the iterative learning algorithm will be presented to generate the knowledge base for the planning and the adjustment. At last, the virtual muscle closed-loop algorithm will be achieved in the master-slave distributed controller for the platform experiment to optimize the algorithm. The main significance of this project is to explore new bionic ideas and to present new control methods. The improvement of the key technology will speed up the application process of the robotic fish, and will promote the development of the robotic fish and its related fields.

深入分析和研究了机器鱼的推进模型和仿生机理，针对当前机器鱼研制过程中遇到的推进效率低下、续航时间短的关键问题，提出一种高效率推进的预测/调节双闭环仿生控制方法：从鱼类游动时肌肉做负功的角度探讨预测规划和反馈调节的控制规律，模拟鱼类的侧线和肌梭的传感原理设计反馈环节，以肌肉神经控制为仿生原型建立双闭环控制结构，提高机器鱼的推进效率。课题以负功所揭示的高效率推进机理为依据，开发关节模块及其驱动算法实时控制刚度和角度并结合肌梭模型反馈关节做功状态，设计虚拟侧线的传感系统识别流场信息；以此为基础，迭代学习建立运动规划和刚度调节的知识库，用虚拟肌肉的思想实现双闭环的控制；针对所提出算法开发主从分布式控制器，制作机器鱼样机平台实验验证并优化算法。项目的主要意义是探索新的仿生思路和控制方法，以求在关键技术上有所突破，从而提高机器鱼的实用水平，推动相关学科和仿生机器鱼技术的研究和发展。

为进一步提高机器鱼推进效率，通过深入研究鱼类侧线系统的感知反馈，利用压力传感器建立机器鱼的虚拟侧线系统，使机器鱼可以高效便捷地感知流场，根据负功的机理设计机器鱼的闭环控制算法，实现预测/调节双闭环控制，提高机器鱼推进效率。.课题以生物鱼侧线系统的结构和功能为基础，研究基于平均值方差的机器鱼虚拟侧线传感布局优化算法，引入三种相对评价指标进行对比分析，利用虚拟侧线实验平台进行仿真工况的复现，并建立评价模型对比布局结果，验证了基于平均值方差算法对提高流场参数辨识准确度的有效性。对优化布局后的压力数据进行预处理，基于概率神经网络和广义回归神经网络各自的特点和优势，设计了基于神经网络的信息处理算法。在仿真探索神经网络处理算法的基础上，采集实验数据验证了流场辨识的准确性。应用刚度调节知识库设计了闭环环节，将得到的流场信息反馈给控制系统，依据流场信息反馈预测规划下一周期的运动规律；分析负功对效率的影响，引入关节空间控制算法，反馈关节做功状态，从而调节驱动刚度。基于负功模型设计模糊控制的隶属度函数和推理知识库，实现预测/调节双闭环控制算法。实验优化两个闭环的控制参数，提高控制精度的同时提高推进效率。最终推进效率提高10%左右。.项目的主要工作是完成机器鱼的流场感知系统，实现机器鱼对流体环境实时反馈的闭环行为，以求推进效率的进一步提高，加速机器鱼的实用化进程。此外，本课题的研究成果也为水下驱动带来一种新的模式，为相关学科的发展提供一些理论和技术的支持。