基于声反射相位特征的UUV避障目标识别技术研究

In recent years, with the rapid development and wide application of underwater vehicles, the path planning and the autonomous underwater navigation of UUV have become the focus of the field. And the identification of obstacles needed to be avoided has become another core issue of the autonomous underwater vehicles. In this subject, the recognition and classification of some typical objectives, such as sea bottom, sea surface, reefs, fish and bubble wake, are studied based on phase characteristics of the target echo and combined with classic detection and identification method on energy, using the forward looking sonar in UUV. It includes the characteristics analysis of the target echo phase, the extraction methods of the echo phase characteristics and the classification methods of the echo phase. This field is still blank in China, and it is also a frontier problem abroad. In this research, the phase models of some typical underwater target echoes need to be built firstly. Then, combined with the present situation of the UUV sonar in China, the transmitted waveform design method is given for the target echo phase extraction. And some post-processing algorithms of the target classification are proposed under the different SNRs and the different moving conditions. In this research, the identification method of the above true or false obstacles needed to be avoided is obtained for the path planning of UUV.

近些年随着水下机器人的快速发展和广泛应用，UUV路径规划或水下自主航行也成为是该领域所关注的重点，而其中避障目标的识别问题更成为自主水下机器人领域中的核心问题。本课题针对UUV水下航行时所遇到的海底、海面、礁石、鱼群和气泡尾流等典型真假目标，结合常规能量检测和识别方法，基于目标反射回波中相位信息，利用UUV前视声纳研究上述目标的识别分类方法，主要包括目标反射回波的相位特征分析、回波相位特征的提取方法、回波相位的分类识别算法等，该领域尚属国内空白，国际上也属于前沿问题。其中，需要构建水下典型目标反射回波相位特征模型，并结合目前国内UUV前视声纳的情况，提出目标回波相位特征提取的发射波形设计和后置处理算法，以此来完成不同信噪比、不同运动条件下的目标分类问题。通过上述研究，来解决UUV路径规划中真假避障目标的声学识别问题。

近些年随着水下机器人的快速发展和广泛应用，UUV路径规划或水下自主航行也成为是该领域所关注的重点，本项目针对避障目标的识别问题开展了深入的研究。针对刚性、弹性球、单体鱼等目标，完成了目标反射回波相位建模的理论分析和计算机仿真。结合多种主动声纳发射波形，研究了双频发射信号的相位特征提取处理方法，并分析多普勒频移对不同信号相位测量的影响，提出LFM-BPSK的水下目标抖动引起多普勒效应的抑制方法。为了提高水下多途条件下的时延估计精度，在匹配滤波的基础上提出了二次相关时延估计方法。对不同发射信号的混响进行统计分析，完成K分布混响模型的建模和仿真，提出相位测量中杂波抑制的有效方法。针对三种单体鱼，采用小波包和双谱分析方法进行了目标分类的研究。此外项目还搭建了一套完整的主动声呐试验平台，进行了大量试验，验证了算法的正确有效性，完成了项目研究各项预期目标，并积累了宝贵的数据，为今后的研究提供了有力的保障。