军事知识服务平台
基于MATLAB并行计算的国产全海深多波束测深数据高效鲁棒性解析方法研究
Development of a High-Efficiency and Robust Parsing Method for Localized Full-Ocean-Depth Multibeam Bathymetry Data via MATLAB Parallel Computing
摘要 《1°×1°全海深多波束测深系统采集协议 V6.5》(以下简称V6.5协议)是中国科学院声学研究所自主研发的国产全海深多波束测深系统专属协议,为我国深海勘测数据标准化采集与处理提供了自主可控的技术支撑.针对该协议下海量二进制测深数据解析效率低下、传统并行解析方案存在线程不安全、容错能力薄弱及协议适配性不足等关键问题,本文提出"内存映射分块预读+动态并行调度+结构化解析"三位一体的高效鲁棒性解析方法.该方法依托 MATLAB 并行计算工具箱构建分层解耦架构,深度适配 V6.5 协议的动态数据特征与国产化装备性能需求,通过内存映射、分块重叠魔数检索、算法级效率提升等关键技术,同时融入全链路合法性校验与脏数据智能过滤设计,实现 6 类核心数据包的精准、高效解析.实验验证表明:该方法较传统串行解析效率提升 4.97~32.64 倍,较无优化并行解析方案内存占用降低 63.1%,解析正确率达 99.996%,在脏数据干扰、动态波束数等复杂作业场景下运行稳定可靠,具备重要的工程应用价值和广阔的国产化技术推广前景.
Abstract:
1°×1° Full-Ocean-Depth Multibeam Bathymetric System Acquisition Protocol V6.5(hereinafter referred to as Protocol V6.5)is a proprietary protocol independently developed by the Institute of Acoustics,Chinese Academy of Sciences for domestic full-ocean-depth multibeam bathymetric systems.It provides independent and controllable technical support for the standardized acquisition and process-ing of deep-sea survey data in China.Aiming at the critical problems including low parsing efficiency of massive binary bathymetric data under this protocol,as well as thread unsafety,insufficient fault tolerance and poor protocol adaptability in traditional parallel parsing schemes,this paper proposes a highly efficient and robust parsing method featuring the trinity of"memory-mapped block prefetching+dy-namic parallel scheduling+structured parsing".Based on the MATLAB Parallel Computing Toolbox,this method constructs a hierarchical decoupled architecture,which is deeply adapted to the dynamic data characteristics of Protocol V6.5 and the performance requirements of domestic equipment.By adopting key technologies such as memory mapping,block-overlapped magic number retrieval and algorithm-lev-el efficiency optimization,and integrating the design of full-link legitimacy verification and intelligent dirty data filtering,it achieves ac-curate and efficient parsing of six types of core data packets.Experimental verification shows that the proposed method improves the pars-ing efficiency by 4.97~32.64 times compared with the traditional serial parsing,reduces memory usage by 63.1%compared with the un-optimized parallel parsing scheme,and reaches a parsing accuracy of 99.996%.It runs stably and reliably in complex operational scenar-ios such as dirty data interference and dynamic beam number,and has important engineering application value and broad promotion prospects for domestic technologies.
刊 名 海洋技术学报 ISTIC
年,卷(期): 2026, 45(4)
分类号: P743.3
机标分类号: P229 TP391 P736.21
在线出版日期: 2026年9月8日
基金项目: E5LZZ10801:国家发改委项目,