SIA OpenIR  > 海洋机器人卓越创新中心
海流环境下水下滑翔机路径规划方法研究
其他题名Underwater Glider path planning methods research in ocean current environment
周耀鉴1,2
导师王晓辉
分类号TP242
关键词水下滑翔机 深平均流 路径规划
索取号TP242/Z78/2017
页数111页
学位专业模式识别与智能系统
学位名称博士
2017-11-30
学位授予单位中国科学院沈阳自动化研究所
学位授予地点沈阳
作者部门海洋机器人卓越创新中心
摘要本文旨在根据水下滑翔机运动特点,结合实际观测应用需求,解决水下滑翔机在海洋观测应用中的环境深平均流估计与流环境下水下滑翔机观测路径规划等理论与技术层面的问题。结合国家自然科学基金重点项目“水下机器人海洋环境自主观测理论与技术(61233013)”,对海流环境中的水下滑翔机路径规划问题进行了研究,具体包含以下内容: (1)水下滑翔机驱动浮力影响因素分析。首先,对水下滑翔机重力浮力配平过程进行了分析。然后考虑舱体压缩率和海水密度差异,结合水下滑翔机的海试数据,采用两种不同的计算方案进行比对,分析了水下滑翔机的净浮力与实际配平密度。(2)水下滑翔机深平均流研究,其包括深平均流估计和预测这两个部分。对于深平均流估计,一般认为其精准度取决于水下滑翔机水平方向速度的精准度。常规的滑翔机水平方向速度计算方案中,不但需要实时TCM和CTD传感器数据值,而且计算较为耗时,在实际应用时会存在不便。在舱体净浮力分析基础上,考虑滑翔机行驶过程中的非稳定区间,结合水下滑翔机运动模型建立了水平方向速度快速计算模型,基于该水平速度技术模型估计了水下滑翔机在海试中的深平均流。结合水下滑翔机海试数据,对求得的水平速度和深平均流估计的正确性进行了验证。在深平均流预测部分,考虑深平均流数据机制复杂和样本量少这两个特点,将深平均流数据看作时间序列。分别采用BP神经网络,最小二乘支持向量机,含经验模态分解的BP神经网络以及含经验模态分解的最小二乘支持向量机这四种时间序列预测方法对深平均流速度进行预测。基于模拟深平均流数据和真实深平均流数据,考虑一周期预测和多周期预测,分别对这些预测方法进行了验证。(3)水下滑翔机在海流中的路径规划研究。其包括局部路径规划研究和全局路径规划研究这两部分。首先研究水下滑翔机的局部路径规划。将深平流预测技术与客观分析技术相结合,便可构建出局部流场。考虑滑翔机应用中的路径跟踪和常规路径规划问题,以构建流场为基础,对这两类局部路径规划问题进行了研究。全局路径规划是水下滑翔机另一个研究热点以及发展方向,为此对全局路径规划进行了研究。全局路径规划问题主要考虑水下滑翔机速度可调这一被常规文献忽略的滑翔机特性,在已有海流场路径规划算法的基础上,提出了一种迭代方案,融入了精确的滑翔机能耗模型,在确保路径存在的前提下,通过不断迭代来求得更合适的速度使得每个滑翔周期的能耗最少,最终实现全局路径规划耗能的近一步优化。
其他摘要According to glider’s feature and combining with the needing of ocean observation, this dissertation seeks to solve the theoretical and technical issues in ocean observation using underwater gliders. In combination with the National Natural Science Foundation of China Projects “Autonomous observation technology of ocean environment with underwater vehicles(61233013)”,this dissertation conducts a deep research on underwater glider path planning methods in ocean current environment.The main contents of the dissertation are as follows: (1)Analysis on influencing factors of underwater glider's driven buoyancy. Firstly, the gravity and buoyancy balanced process are discussed. Then considering the pressure hull compression deformation and the seawater density difference and combining the data of underwater glier from sea trials, we adopt two different calculation schemes for comparison, and calculate the net buoyancy and real balanced density of the underwater glier. (2) Research on depth-averaged currents(DACs) of underwater glider, which contain two parts: DACs estimate and forcast. For DACs estimate,it is generally believed that the estimated accuracy depends on the accuracy of the horizontal speed of underwater glider. In conventional scheme to calculate the horizontal speed of underwater glider, not only real-time data from TCM and CTD sensors is needed, but also the time consuming is too long, which may cause inconveniences in practice.Based on the driven buoyancy analysis, considering the unstable intervals of the gliding profiles, in combination with the stable motion model of underwater glider, we build a rapid-calculation model for glider horizontal speed, then on the basis of the speed model, we estmate the DACs from sea trials. And we verified the correctness of the caculated speeds and DACs combining the sea trials data. In the part of DACs prediction, the two features are considered, which corresponding to the mechanism of DACs is sophisticated and the sample size of DACs isquite limited, respectively. We treat the DACs as a time series and then we use four time series prediction methods to predict them. Specifically, the four methods are backpropagation neural networks (BPNNs), least squares support vector machines (LSSVMs),empirical mode decomposition backpropagation neural networks(EMD-BPNNs) and empirical mode decomposition least squares support vector machines (EMD-LSSVMs).We use the data derived from a real-time simulation environment and real sea trials to test and verify the performance of these methods, and we obtain the next one profile prediction and next N profiles prediction, respectively. (3) Research on glider's path planning in flow fields, which contain two parts: path planning in local flow field and path planning in global flow field. Since ocean currents forecasts are restricted by their low resolution and a amount of errors, hence we research glider's path planning in loacl flow field firstly. Combining the DACs prediction technique and objective analysis technique, the local flow field can be built. Considering the path tracking and conventional path planning problems in practical application of glider, based on the constructed flow field,the two types path planning problems are be researched. Because the path planning in global flow field is the hotspot and development direction yet,we also research it. In this problem, the main point that we put into consideration is that the speed of underwater glider can be adjusted, which are neglected usually in current documentations. Based on the existing efficient path planning algorithm in flow fields, in combination with a accurate model of energy consumption of glider, a iterative scheme is proposed, which firstly guarantee the path exists, and then get the most appropriate speed corresponding the minimum energy consumption of each diving cycle through repeated iteration, and the consumption of energy can be saved in a further way.
语种中文
产权排序1
文献类型学位论文
条目标识符http://ir.sia.cn/handle/173321/21276
专题海洋机器人卓越创新中心
作者单位1.中国科学院沈阳自动化研究所
2.中国科学院大学
推荐引用方式
GB/T 7714
周耀鉴. 海流环境下水下滑翔机路径规划方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2017.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
海流环境下水下滑翔机路径规划方法研究 .(4457KB)学位论文 开放获取CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[周耀鉴]的文章
百度学术
百度学术中相似的文章
[周耀鉴]的文章
必应学术
必应学术中相似的文章
[周耀鉴]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。