基于磁矩量法和多层自适应交叉近似法的船舶磁场大规模建模方法研究

项目名称： 基于磁矩量法和多层自适应交叉近似法的船舶磁场大规模建模方法研究

项目编号： No.51277176

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 电工技术

项目作者： 郭成豹

作者单位： 中国人民解放军海军工程大学

项目金额： 72万元

中文摘要： 随着磁探测技术的飞速发展，急需进一步提高钢铁船舶的磁场隐身能力，实现这一目标的首要前提是获取高精度的船舶磁场特征。由于船舶结构和磁化机理的复杂性，现有船舶磁场小规模建模技术无法满足这一紧迫需求，严重影响船舶航行安全。本研究基于磁矩量法、多层自适应交叉近似法和GPU并行计算技术，建立船舶磁场大规模正演和反演建模新方法，实现船舶磁场的高精度建模，重点在于采用多层自适应交叉近似法和GPU并行计算技术对磁矩量法得到的大型稠密系数矩阵进行压缩和加速。根据船舶结构和船钢磁特性，可快速精确地正演分析船舶感应磁场，评估船舶永久磁场。根据船舶磁场测量值，可快速精确地反演分析船舶的感应和永久磁性，从而修正船舶磁场正演建模参数。最终获得精度大于99%的船舶磁场特征，进而最优化设计船载消磁线圈，实现消除98%以上船舶磁场的目标。通过上述研究，为进一步提高我国船舶磁场隐身能力奠定理论和关键技术基础。

中文关键词： 舰船磁场；消磁；磁矩量法；多层自适应交叉近似法；图形处理单元

英文摘要： With the rapid developments in magnetic sensor technology, the ability of magnetic field silencing of the steel ships should be improved, in which the precondition is to achieve the precise magnetic signature of the ship. Because the ship's construction and magnetization are very complicated, the current small scale modeling of the ship magnetic signature can't satisfy this urgent need, and this affects the safety of our country's ships. This research presents a novel large scale modeling method of ship magnetic fields with the Magnetic Moment Method, Multilevel Adaptive Cross Approximation and GPU parallel computing technology, to realize the high accurate modeling of the ship magnetic fields, which keypoint is to compress and accelerate the large dense matrix from the Magnetic Moment Method by the Multilevel Adaptive Cross Approximation and GPU parallel computing technology. From the ship's construction and the magnetic character of ship steel, the ship induced fields can be predicted rapidly and accurately, and the ship remanent fields can be evaluated. From the measured values of the ship magnetic signature, the ship's induced and remanent magnetizations can be calculated inversely, and then the parameters in the forward modeling of the ship magnetic signature are corrected. Finally, the resulted modeling of

英文关键词： ship magnetic fields；degaussing；magnetic moment method；multilevel adaptive cross approximation；GPU