一种基于DTW算法的磁场相似性度量方法  

Similarity Magnetic Field Measurement Method Based on the DynamicTime Warping Algorithm

作　　者：胡家文 刘忠乐[1] 文无敌 张志强[1] HU Jiawen;LIU Zhongle;WEN Wudi;ZHANG Zhiqiang(School of Weaponary Engineering,Naval University of Engineering,Wuhan 430033,China)

机构地区：[1]海军工程大学兵器工程学院,湖北武汉430033

出　　处：《水下无人系统学报》2023年第3期430-435,共6页Journal of Unmanned Undersea Systems

基　　金：国家自然科学基金项目资助(42122025)。

摘　　要：在舰船磁场模拟及磁探测系统的目标识别中,需要对模拟或探测磁场与真实目标磁场分布的相似程度进行评估。文中针对以往评估方法存在的问题,提出利用动态时间规整(DTW)算法计算磁场的相似度,解决了目标速度不同和起止点不一致情况下的磁场曲线相似度评估问题;为降低磁场曲线局部扰动带来的影响,利用最长公共子串对DTW算法进行了优化,提高了相似度计算精度;最后以模型试验验证了算法的正确性。与传统评估方法相比,该方法无需人工预先设置参数,能直接给出相似度评价结果,可应用于舰船磁场模拟效果评估及磁探测系统的目标识别。Simulating the magnetic field of a ship and identifying targets using magnetic detection requires evaluating thesimilarity of the magnetic field distribution obtained from the simulation and detection with the magnetic field distribution ofthe real target.In view of the problems existing in previous evaluation methods,the use of dynamic time warping algorithm isproposed to calculate the similarity of the magnetic field and solve the problem of evaluating the similarity of the magneticfield curve under different target speeds and inconsistent starting and ending points.To reduce the influence of localdisturbances of the magnetic field curve,the longest common substring is used to optimize the dynamic time warpingalgorithm,thereby improving the calculation accuracy of similarity.Finally,the accuracy of the algorithm is verified by modeltesting.Compared with the traditional evaluation methods,no parameters need to be manually set in advance and the similarityevaluation result can be given directly,enabling the evaluation of the simulation effect of the ship’s magnetic field andidentification of targets from magnetic detection.

关 键 词：舰船 磁探测 动态时间规整 相似性度量 磁场曲线 最长公共子串 

分 类 号：TJ61[兵器科学与技术—武器系统与运用工程]