摘要
类空化效应是由于水下螺旋桨推进器近水面高速旋转时,在螺旋桨与水面之间形成漩涡,从而将空气吸入桨内导致推进器效率极大降低,引起推力损失和噪声的现象。螺旋桨推进器作为大多数水下机器人的唯一动力源,类空化效应的产生将极大影响机器人运动控制的稳定性。提出了一种基于高斯过程的水下螺旋桨推进器推力预测方法,可以实现类空化效应下的高精度推力预测。介绍了类空化效应并揭示类空化效应的产生机理。建立了基于贝叶斯估计的推进器推力模型,对未出现空化效应时的推进器推力进行准确预测。在此基础上,提出了基于高斯过程的推进器推力预测模型,利用基于贝叶斯估计的推力预测模型与基于高斯过程的类空化误差补偿,实现对类空化效应下的推力预测。通过试验验证了基于高斯过程的类空化预测模型的精确性与有效性,为水下机器人近水面的高精度运动控制奠定基础。
Quasi-cavitation is caused by the vortex formed between the propeller and water surface when the propeller rotates at high speed near the water surface,which leads to a great reduction of thruster’s efficiency,thrust loss and noise.As the only power source of most underwater vehicles,quasi-cavitation on the thruster will significantly decrease the stability of motion control.A novel approach for thrust prediction of underwater blade-propeller-type thrusters under quasi-cavitation is proposed,which can realize thrust prediction with high accuracy.The mechanism of quasi-cavitation is introduced and revealed.A Bayesian estimation based thrust model(BETM)is established to perform accurate thrust prediction without quasi-cavitation.On this basis,a quasi-cavitation thrust model based on Gaussian process(QCTM-GP)is proposed,which utilizes BETM and error compensation based on Gaussian process to complete the prediction of quasi-cavitation.The accuracy and validity of the proposed prediction model are verified via experiments.QCTM-GP lays a foundation for the high accurate motion control of underwater vehicles near surface.
作者
罗阳
李战东
陶建国
邓立平
邓宗全
LUO Yang;LI Zhandong;TAO Jianguo;DENG Liping;DENG Zongquan(State Key Laboratory of Robotics and System,Harbin Institute of Technology,Harbin 150000;Civil Aviation Institute,Shenyang Aerospace University,Shenyang 110136)
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2020年第17期1-11,共11页
Journal of Mechanical Engineering
基金
国家自然科学基金(61673138)
机器人技术与系统国家重点实验室(哈尔滨工业大学)自主研究课题(SKLRS201804B)
国家重点基础研究发展计划(973计划,2013CB035502)资助项目。
关键词
水下螺旋桨推进器
类空化效应
推力预测
贝叶斯估计
高斯过程
underwater blade-propeller-type thruster
quasi-cavitation
thrust prediction
Bayesian estimation
Gaussian process