【作者】 王玉甲；

【导师】 张铭钧；

【作者基本信息】 哈尔滨工程大学 ， 机械设计及理论， 2006， 博士

【摘要】 自主式无人无缆水下机器人(AUV)既是海洋资源开发和水下环境作业的重要载体。智能化是AUV的重要标志，也是AUV研究和应用的基础和关键技术。作为在复杂海洋环境下工作的载体，自主性及安全性是AUV的基本要求，智能控制是实现其自主控制和完成任务使命的核心技术。随着AUV功能的不断完善和拓展，系统的复杂程度也在不断增加，其安全性越来越受到广泛的高度重视，而状态监控技术为AUV的安全性提供了技术保障。正确认识水下机器人的本质特征，建立完善的状态智能监测与控制系统，完善相应的理论，对提高水下机器人安全性和智能化水平具有重要的理论研究意义和实际应用价值。水下机器人状态监测技术研究前提是对监测对象的详细了解，掌握被监测对象的类型、参数、性能等，其研究的关键在于监测对象的建立，信息处理与特征提取方法的选取，以及故障检测方法、监测策略的制定以及故障程度及运行状态的分析。本文综述了国内外机器人系统及其故障诊断、状态监测方面的研究现状，全面地总结了水下机器人系统状态监测研究的关键技术、主要研究结果和存在的问题。针对水下机器人控制过程所包含的基本环节，本文从传感器系统及推进器系统的故障机理出发，建立了基于控制器的智能状态监测系统，主要包括了推进器性能监测子系统，水下机器人运行状态监测子系统，残差融合与状态评价子系统，实现了基于多源残差信息的融合故障检测与状态评价，实现了对水下机器人推进器系统与传感器系统的状态监测。针对复杂的海洋环境，本文提出了水下机器人系统设计的评价标准，提出一种适用于水下机器人系统的体系结构，给出了硬件系统设计方法及中央控制器、推进器，传感器，能源及通讯系统设计方法。提出了基于改进的动态递归网络的水下机器人状态模型及基于径向基神经网络的水下机器人推进器性能模型，研制了水下机器人试验平台并进行了水池实验，基于水池实验数据，获取了建立AUV运行状态模型和推进器性能模型所需的学习样本，建立了推进器动力学模型。进行了推进器故障模拟实验研究，进行了AUV推进器、传感器故障状态监测水池实验，实验结果验证了本文所提出的水下机器人智能状态监测系统及方法的有效性和可行性。更多还原

【Abstract】 Autonomous underwater vehicles (AUV) are the important carrying agents in the ocean resource development and underwater environment operation. Intelligence is one of the prominent features of AUV, as well as the basis and key technology for research and application of AUV.As a facility operating in the complicated sea environment, the capabilities in autonomy and security are fundamental requirements for AUV, and intelligent control is the core technology to realize its autonomous control and implement the required tasks. With the increasing development and enhancement of AUV function, the complexity of AUV system is also increasing with them, thus its security has been more and more paid attention to. Condition monitoring technology can guarantee the AUV safety. Therefore, it can be provided with theory research significance and application value to model and design intelligent condition monitoring and control system, as well as improve on relevant theoretic.The AUV condition monitoring technology is based on comprehensively identifying the monitored objects, and mastering their types, parameters and performance, etc. The key technologies of research are to model the monitored objects, deal with the collected information, select the appropriate method of feature extraction, establish the fault detection method and monitoring strategy, and diagnose degree of malfunction as well as analyze the running state.Art of the state of AUV system and fault diagnosis is overviewed in this thesis. While the key technology, mostly research results and existing issues of AUV condition monitoring are summarized.According to the basic components of AUV control processes, the intelligent condition monitoring system based on controller has been established on the basis of the fault mechanism of sensor system and thruster system, which constitutes the thruster fault detection subsystem, the AUV motion state detection subsystem, the residual fusion and condition assess subsystem. As a result, it is realized that the fused fault detection and condition evaluation based on multi-residual information, as well as the condition monitoring for the thruster system and the sensor system of AUV. For the intricate sea environment, the thesis presented the assess standard of AUV system design, a system structure that can be applicable to the AUV system, and the design method of hardware system, the center controller, thruster, sensor, power source and communication system.The condition model of AUV based on the improved dynamic recursion network and the thruster performance model based on the RBF network are developed, and the test platform of AUV is implemented and carry out the experiments in pond. From the experiment data, the learn sample is obtained that is requisite for setting up the AUV running state model and thruster performance model. Then, the thruster dynamic model is developed. Furthermore, the research has been made for simulating the thrusters and sensors fault, and the experiment in the pool of AUV thruster and sensors fault monitoring was carried out. The results prove that the models and methods developed in this thesis have been evaluated to be valid and feasible for AUV intelligent condition monitoring system.更多还原