[{"data":1,"prerenderedAt":210},["ShallowReactive",2],{"content-query-QUX0zuHrpD":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":204,"_id":205,"_source":206,"_file":207,"_stem":208,"_extension":209},"/news/zh/3053","zh",false,"","业界SOTA！基于昇思MindSpore的大地电磁智能反演模型首开源，助力地球物理勘探加速智能化","近日，华为AI4S Lab与清华大学李懋坤教授团队、华为先进计算与存储实验室合作，基于昇腾AI处理器与昇思MindSpore AI框架打造了大地电磁智能反演模型","2024-04-08","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/fa3776e0202d4f69a648eb758f3cd689.png","news",{"type":14,"children":15,"toc":201},"root",[16,24,30,50,58,65,72,79,84,91,96,122,129,134,139,144],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"业界sota基于昇思mindspore的大地电磁智能反演模型首开源助力地球物理勘探加速智能化",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":23,"value":29},"近日，华为AI4S Lab与清华大学李懋坤教授团队、华为先进计算与存储实验室合作，基于昇腾AI处理器与昇思MindSpore AI框架打造了大地电磁智能反演模型。该模型通过变分自编码器（VAE）灵活嵌入了多物理先验知识，达到了业界SOTA。该成果已被国际顶级勘探地球物理期刊《Geophysics》收录，相关代码已在昇思MindSpore Elec电磁仿真套件代码仓中开源，同时，该成果也在昇思人工智能框架峰会2024上发布亮相。",{"type":17,"tag":25,"props":31,"children":32},{},[33,35,41,43,48],{"type":23,"value":34},"大地电磁（Magnetotelluric）反演是一种地球物理勘探技术。它利用地球表面上的电磁场与地下岩石、矿床等物质的电导率、磁导率等物理属性之间的关系，推断地下物质的电磁参数分布从而达到探测地下物质的目的。该技术被广泛应用于矿产勘探、油气勘探、地质灾害预测等领域。大地电磁反演相比地震波反演具有",{"type":17,"tag":36,"props":37,"children":38},"strong",{},[39],{"type":23,"value":40},"探测深度大、测量复杂度低、操作较为简便",{"type":23,"value":42},"等优势，但传统的大地电磁反演方法（如像素反演方法等）仍面临",{"type":17,"tag":36,"props":44,"children":45},{},[46],{"type":23,"value":47},"数据分辨率低、病态性和多解性",{"type":23,"value":49},"等问题。基于昇思MindSpore的大地电磁智能反演模型通过融入多物理先验知识，有效地解决了上述问题。",{"type":17,"tag":25,"props":51,"children":52},{},[53],{"type":17,"tag":54,"props":55,"children":57},"img",{"alt":7,"src":56},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/3e683e0e18054c9f9dc413090a8877c8.png",[],{"type":17,"tag":25,"props":59,"children":60},{},[61],{"type":17,"tag":54,"props":62,"children":64},{"alt":7,"src":63},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/907f9e6c4d20490cb38ca2519f44d12c.png",[],{"type":17,"tag":25,"props":66,"children":67},{},[68],{"type":17,"tag":54,"props":69,"children":71},{"alt":7,"src":70},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/2065b8cc70c043fb987f42afbd18c614.png",[],{"type":17,"tag":25,"props":73,"children":74},{},[75],{"type":17,"tag":54,"props":76,"children":78},{"alt":7,"src":77},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/ee0ab5eb3c7b4e96ae7bce8c3ca460df.png",[],{"type":17,"tag":25,"props":80,"children":81},{},[82],{"type":23,"value":83},"图3 大地电磁反演精度对比",{"type":17,"tag":25,"props":85,"children":86},{},[87],{"type":17,"tag":54,"props":88,"children":90},{"alt":7,"src":89},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/6a701551e318478db1e2d83e814d24d1.png",[],{"type":17,"tag":25,"props":92,"children":93},{},[94],{"type":23,"value":95},"图4 大地电磁反演收敛速度对比（Pixel-based：传统反演；Feature-based：我们的工作）",{"type":17,"tag":25,"props":97,"children":98},{},[99,101,106,108,113,115,120],{"type":23,"value":100},"此外，大地电磁智能反演模型还在",{"type":17,"tag":36,"props":102,"children":103},{},[104],{"type":23,"value":105},"南部非洲开源数据集",{"type":23,"value":107},"（SAMTEX）上做了验证。该反演区域位于南部非洲西海岸附近，长度约为750km，深度选定为80km。该测区显著特征为在",{"type":17,"tag":36,"props":109,"children":110},{},[111],{"type":23,"value":112},"水平方向100km至400km之间，深度20km以浅的区域存在的高导结构",{"type":23,"value":114},"。由于低频电磁波在导体结构中的衰减，MT方法对高导结构下部区域的敏感度很低, 因此无先验知识约束的传统MT反演难以准确重建高导地层的下边界位置。大地电磁智能反演对",{"type":17,"tag":36,"props":116,"children":117},{},[118],{"type":23,"value":119},"高导地层的下边界重建较为清晰准确",{"type":23,"value":121},"，较好地将地层厚度的先验知识融入了反演。",{"type":17,"tag":25,"props":123,"children":124},{},[125],{"type":17,"tag":54,"props":126,"children":128},{"alt":7,"src":127},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/10057dfb65ec41c29cb52530bdac753e.png",[],{"type":17,"tag":25,"props":130,"children":131},{},[132],{"type":23,"value":133},"图5 南部非洲MT反演示例图（上图：传统反演；下图：我们的工作）",{"type":17,"tag":25,"props":135,"children":136},{},[137],{"type":23,"value":138},"大地电磁智能反演模型基于高效易用的昇思MindSpore融合框架，实现了SOTA的结果。未来，我们将进一步拓展该模型，推动应用落地到更多领域。同时，希望有更多的企业、科研院所能够参与共建、共享，依托昇思MindSpore共同打造更多智能电磁的创新应用。",{"type":17,"tag":25,"props":140,"children":141},{},[142],{"type":23,"value":143},"相关工作：",{"type":17,"tag":145,"props":146,"children":147},"ul",{},[148,166,177,189],{"type":17,"tag":149,"props":150,"children":151},"li",{},[152,154,158],{"type":23,"value":153},"了解更多可查看论文：",{"type":17,"tag":155,"props":156,"children":157},"br",{},[],{"type":17,"tag":159,"props":160,"children":164},"a",{"href":161,"rel":162},"https://library.seg.org/doi/10.1190/geo2022-0774.1",[163],"nofollow",[165],{"type":23,"value":161},{"type":17,"tag":149,"props":167,"children":168},{},[169,171],{"type":23,"value":170},"昇思MindSpore Science开源路径",{"type":17,"tag":159,"props":172,"children":175},{"href":173,"rel":174},"https://gitee.com/mindspore/mindscience",[163],[176],{"type":23,"value":173},{"type":17,"tag":149,"props":178,"children":179},{},[180,182],{"type":23,"value":181},"昇思大地电磁智能反演模型开源路径",{"type":17,"tag":159,"props":183,"children":186},{"href":184,"rel":185},"https://gitee.com/mindspore/mindscience/tree/master/MindElec/examples/hybrid%5C_driven/feature%5C_based%5C_MT%5C_inversion",[163],[187],{"type":23,"value":188},"https://gitee.com/mindspore/mindscience/tree/master/MindElec/examples/hybrid\\_driven/feature\\_based\\_MT\\_inversion",{"type":17,"tag":149,"props":190,"children":191},{},[192,194,197],{"type":23,"value":193},"添加小猫子@MindSpore助手加入社群",{"type":17,"tag":155,"props":195,"children":196},{},[],{"type":17,"tag":54,"props":198,"children":200},{"alt":7,"src":199},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/04/12/eb5ba71bc2fe43239de0f81ee03bcf19.png",[],{"title":7,"searchDepth":202,"depth":202,"links":203},4,[],"markdown","content:news:zh:3053.md","content","news/zh/3053.md","news/zh/3053","md",1776506078865]