观探测AUV行为动力学建模及行为优化控制方法研究

Autonomous behavior ability is the kernel to improve the intelligent level and autonomy of AUV. This project focuses on the autonomous behavior ability problems of ocean survey AUV and its typical ocean survey operations. Three key scientific problems on behavior system modeling, behavior combinatorial optimization solving, and behavior on-line rolling optimizing are researched closely surrounding the research objectives of building a stable and commensurable behavior system and realizing on-line behavior optimizing. First, improving conventional behavior modeling method, dynamics modeling method with differential constraints is proposed to model the AUV behavior system. Second, based on the behavior integration strategy, a new continuous space discrete particle swarm optimization (CDPSO) method is designed to solve AUV behavior combinatorial optimization problem. Third, extending the fixed time domain rolling mechanism of conventional receding horizon control (RHC), hybrid rolling mechanism based on behavior time domain and discrete event is proposed to realize on-line optimizing control of AUV behavior. Finally, simulations and experiments in reservoir using a small AUV system will be verified. Development of the subject will lay the technical foundation of improving the adaptive ability of AUV ocean survey to environment and task, and will provide theory and method reference for deepening and perfecting the research of AUV autonomous behavior ability.

自主行为能力是提高AUV智能水平和自主能力的核心。本项目重点针对海洋观探测AUV及其典型作业的自主行为能力问题，紧密围绕构建稳定可公度的行为系统和实现行为在线优化的目标，开展AUV行为系统建模、行为组合优化求解、行为在线滚动优化三个关键科学问题研究。第一，改进传统行为动力学建模方法，提出带运动学和动力学微分约束的建模方法，建立AUV行为系统模型。第二，基于行为融合协调策略，设计用于求解行为组合优化问题的新型连续空间离散粒子群（CDPSO）方法，进行AUV行为的优化解算。第三，扩展传统滚动优化控制（RHC）的固定时间域滚动机制，提出基于行为时间域和离散事件混合的滚动机制，实现AUV行为的在线滚动优化控制。最后，进行仿真和微小型AUV系统的外场水库试验验证。通过研究，为提高AUV观探测作业对环境和任务的自适应能力奠定技术基础，为深化和完善AUV的自主行为能力研究提供理论和方法参考。

自主水下航行器（AUV）是高新技术的集成装备，是目前世界上公认的最具有应用前景的海洋开发工具，可以执行多样性的海洋作业使命。本研究着眼于AUV的一个热点应用领域：海洋观探测使命。重点针对海洋观探测AUV及其典型作业的自主行为能力问题，开展AUV行为系统建模、行为组合优化求解、行为在线滚动优化三方面的研究。完成了观探测AUV典型作业行为分析及行为系统框架构建，建立了基于微分约束行为动力学的AUV观探测行为模型，分析形成了模型参数作用机理及对AUV行为系统稳定性的影响规律。采用基于融合机制的AUV行为协调方法，构建了AUV观探测行为的多目标组合优化求解模型，设计了基于连续空间离散PSO算法的AUV行为组合优化求解方法。提出了基于行为时间域和离散事件混合的AUV观探测行为滚动优化机制，形成了离散PSO求解和滚动优化结合的在线行为优化算法。最后对所提出的方法和算法的可行性和有效性基于仿真系统、实际航行器系统及无人车模拟系统进行了验证。