{"id":276405,"date":"2023-03-24T22:26:51","date_gmt":"2023-03-24T22:26:51","guid":{"rendered":"\/forum\/forums\/topic\/ansys-insight-yongtapuyouhualaijinxingyfenshuqidenixiangsheji-inverse-design\/"},"modified":"2023-04-05T21:26:59","modified_gmt":"2023-04-05T21:26:59","slug":"ansys-insight-yongtapuyouhualaijinxingyfenshuqidenixiangsheji-inverse-design","status":"closed","type":"topic","link":"https:\/\/innovationspace.ansys.com\/forum\/forums\/topic\/ansys-insight-yongtapuyouhualaijinxingyfenshuqidenixiangsheji-inverse-design\/","title":{"rendered":"Ansys Insight: \u7528\u62d3\u6251\u4f18\u5316\u6765\u8fdb\u884cY\u5206\u675f\u5668\u7684\u9006\u5411\u8bbe\u8ba1 Inverse design"},"content":{"rendered":"

 2019b R2 \u4ee5\u53ca\u66f4\u65b0\u7248\u672c\u652f\u6301\u62d3\u6251\u4f18\u5316\u3002 \u76f8\u5173\u4f8b\u5b50\u53c2\u89c1<\/p>\n

Inverse Design of a Splitter Using Topology Optimization<\/a><\/h3>\n

Topology Optimization of a 4-channel wavelength demultiplexer 2D-TE<\/a><\/strong><\/p>\n

Topology optimization of a O-band\/C-band wavelength de-multiplexer with 100nm feature size constraint<\/a><\/strong><\/p>\n

Inverse design of waveguide crossing<\/a><\/p>\n

Optimizable Geometry – Python API<\/a> <\/p>\n

\u7b80\u4ecb <\/strong><\/p>\n

\u4f7f\u7528\u62d3\u6251\u4f18\u5316\u8fdb\u884c\u9006\u5411\u8bbe\u8ba1\u53ef\u4ee5\u4f7f\u7528\u7528\u6237\u81ea\u5b9a\u4e49\u8bc4\u4ef7\u51fd\u6570FOM\u751f\u6210\u4f18\u5316\u7684\u5668\u4ef6\u3002\u5728\u672c\u4f8b\u4e2d\u6211\u4eec\u4f18\u5316\u7684\u662fT_up \u6700\u5927\u5316\uff0c\u5373\u4e0a\u81c2\u7684\uff08\u5e73\u5747\uff09 \u900f\u5c04\u7387\uff0c\u6240\u4ee5\u5728\u4eff\u771f\u65f6\u4f7f\u7528\u4e86\u5bf9\u79f0\u6027\u8fb9\u754c\u6761\u4ef6\uff08\u6a21\u5f0f\u5149\u6e90\u4e5f\u5177\u6709\u6b64\u5bf9\u79f0\u6027\uff09\u3002 \u4e0a\u81c2\u7684\u6700\u5927\u900f\u5c04\u7387\u662f0.5\uff0c \u6240\u4ee5\u6211\u4eec\u5c06\u4f18\u5316\u51fd\u6570\u5b9a\u4e49\u4e3a FOM=0.5 – T_up\uff0c\u76f8\u5f53\u4e8e\u63d2\u5165\u635f\u8017\u3002<\/p>\n

\u6b64\u79cd\u4f18\u5316\u7684\u7279\u70b9\u662f\u7528\u6237\u53ea\u9700\u8981\u6307\u5b9a\u4f18\u5316\u7684\u533a\u57df\uff0c\u800c\u4e0d\u9700\u8981\u53c2\u6570\u5316\u7ed3\u6784\u548c\u4f18\u5316\u7ed3\u6784\u5f62\u72b6\uff08\u53c2\u89c1\u4e4b\u524d\u7684API\u4f18\u5316\uff09\u3002\u867d\u7136\u7ed3\u679c\u770b\u8d77\u6765\u5e76\u4e0d\u662f\u5f88\u76f4\u89c2\uff08\u4f8b\u5982\u4e3a\u4ec0\u4e48\u662f\u8fd9\u4e2a\u6837\u5b50\uff09\uff0c\u4f46\u662f\u900f\u5c04\u6548\u7387\u5f88\u9ad8\u3002<\/p>\n

\u5f53\u7136\uff0c\u6b64\u79cd\u4f18\u5316\u7684\u7f3a\u70b9\u662f\u53ef\u5236\u9020\u6027\u3002\u4e3a\u4e86\u63a7\u5236\u6700\u5c0f\u5c3a\u5bf8\uff0c\u53ef\u4ee5\u5b9a\u4e49\u534a\u5f84\u51fd\u6570\u6ee4\u6ce2\uff0c\u8fdb\u800c\u4e00\u5904\u592a\u5c0f\u7684\u51e0\u4f55\u5f62\u8c8c\u5e76\u5c06\u5c16\u89d2\u5149\u6ed1\u8bdd\uff0c\u8fdb\u800c\u6539\u5584\u53ef\u5236\u9020\u6027\u3002<\/p>\n

\u4e0b\u9762\u662f\u4f18\u5316\u6b65\u9aa4\u3002<\/p>\n

Step 1: Define base simulation project<\/h2>\n

\u9996\u5148\u9700\u8981\u5b9a\u4e49\u57fa\u7840\u4eff\u771f\u6587\u4ef6\uff0c\u5b83\u542b\u6709\u5668\u4ef6\u7684\u666e\u901a\u7ed3\u6784\uff0c\u542b\u4eff\u771f\u533a\uff0c\u8f93\u5165\u8f93\u51fa\u6ce2\u5bfc\uff0c\u6a21\u5f0f\u5149\u6e90\u548c\u76d1\u89c6\u5668\u3002\u672c\u4f8b50:50 splitter\u7684\u5de5\u4f5c\u6ce2\u957f\u662f1450nm \u5230 1650nm\u3002\u53ef\u4ee5\u4f7f\u7528\u811a\u672c\u6587\u4ef6\u751f\u6210splitter_base_2D_TE_topology.lsf <\/p>\n

\u6253\u5f00FDTD\u5e76\u8fd0\u884c\u811a\u672c\u6587\u4ef6\u540e\u53ef\u4ee5\u770b\u5230<\/p>\n

<\/a><\/p>\n

\u7ed3\u6784\u7ec4‘initial_guess’\u751f\u6210\u4f18\u5316\u7684\u5668\u4ef6\u7684\u8d77\u70b9\uff0c\u4e0d\u8fc7\u8fd9\u4e2a\u5e76\u4e0d\u662f\u5fc5\u8981\u7684\uff0c\u62d3\u6251\u4f18\u5316\u4e5f\u53ef\u4ee5\u4ece\u96f6\u7ed3\u6784\u5f00\u59cb\u4f18\u5316\u3002 \u5f53\u7136\uff0c\u5148\u6709\u521d\u6b65\u7ed3\u6784\u53ef\u80fd\u4f1a\u51cf\u5c11\u8fed\u4ee3\u6b21\u6570\u3002<\/p>\n

Step 2: Define optimization region<\/h2>\n

\u9996\u5148\u9700\u8981\u4f60\u5b9a\u4e49\u51e0\u4e2a\u8981\u4f18\u5316\u7684\u53c2\u6570\uff1a<\/p>\n

\n\n\n\n\n\n\n\n\n\n\n\n
Name<\/th>\nValue in the example<\/th>\nDescription<\/th>\n<\/tr>\n<\/thead>\n
size_x<\/td>\n3000 [nm]<\/td>\nLength of the device<\/td>\n<\/tr>\n
size_y<\/td>\n1800 [nm]<\/td>\nSince we use symmetry along the y-axis, the value is half the height of the device<\/td>\n<\/tr>\n
delta_x<\/td>\n20 [nm]<\/td>\nSize of a pixel along x-axis<\/td>\n<\/tr>\n
delta_y<\/td>\n20 [nm]<\/td>\nSize of a pixel along y-axis<\/td>\n<\/tr>\n
filter_R<\/td>\n500 [nm]<\/td>\nRadius of the smoothing filter which removes small features and sharp corners<\/td>\n<\/tr>\n
eps_max<\/td>\n2.8^2=7.84<\/td>\nEffective permittivity for a Silicon waveguide with a thickness of 220nm<\/td>\n<\/tr>\n
eps_min<\/td>\n1.44^2=2.0736<\/td>\nPermittivity of the SiO2 cladding<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

\u6b64\u5916\u8fd8\u9700\u8981\u5b9a\u4e49 FOM \u548c\u4e00\u4e9b\u5176\u5b83\u7684\u4f18\u5316\u8bbe\u7f6e\u3002 \u6240\u7528\u811a\u672c\u53c2\u89c1 pt_2D_TE_topology.py
\u8fd9\u91cc\u4e5f\u5b9a\u4e49\u4e86\u521d\u59cb\u53c2\u6570\u3002 \u524d\u9762\u8bb2\u8fc7\u62d3\u6251\u4f18\u5316\u7684\u4f18\u70b9\u662f\u4e0d\u9700\u8981\u8bbe\u7f6e\u590d\u6742\u7684\u521d\u59cb\u6761\u4ef6\uff0c\u800c\u662f\u7b80\u5355\u5730\u5728\u4f18\u5316\u533a\u8bbe\u7f6e\u4e3a\u5747\u5300\u7684 eps_max \u6216 eps_min.<\/p>\n

Step 3: Run topology optimization<\/h2>\n

\u8fd0\u884cPython \u811a\u672c\u6587\u4ef6\u5c31\u53ef\u4ee5\u5f00\u59cb\u4f18\u5316\u4e86\u3002\u5728\u8fd9\u4e2a\u4f8b\u5b50\u4e2d\uff0c\u62d3\u6251\u4f18\u5316\u6709\u4e24\u4e2a\u9636\u6bb5\uff1a \u7b2c\u4e00\uff0c\u7070\u8272\u9636\u6bb5 ‘grayscale phase’\uff0c\u4f18\u5316\u5668\u53ef\u4ee5\u81ea\u7531\u5730\u4f7f\u7528 [eps_low,eps_high ] \u4e2d\u7684\u4efb\u4f55\u6570\u503c\u3002\u5f53\u4f18\u5316\u5668\u8fbe\u5230\u4e00\u4e2a\u4f18\u5316\u540e\uff0c \u5b83\u5c06\u5f00\u59cb\u4e8c\u503c\u5316\u9636\u6bb5 ‘binarization phase’\u3002 \u6b64\u65f6\u4f18\u5316\u5668\u5c06\u8fde\u7eed\u7684\u4ecb\u7535\u5e38\u6570\u5206\u5e03\u53d8\u6210\u4e8c\u503c\u7ed3\u6784\uff0c\u5373 eps_low \u6216\u8005 eps_high\u4f46\u662f\u6ca1\u6709\u5176\u5b83\u4e2d\u95f4\u503c\uff0c\u6b64\u65f6\u7528\u6765\u9009\u62e9\u5b9e\u9645\u6750\u6599\u3002\u5927\u7ea6\u7ecf\u8fc7500\u6b21\u8fed\u4ee3\u540e\u6709\u7ed3\u679c\u3002<\/p>\n

\u6211\u4eec\u770b\u5230FOM\u66fe\u6682\u65f6\u53d8\u5dee\uff0c\u6bcf\u4e00\u4e2a\u6781\u503c\u9884\u793a\u7740\u4e00\u4e2a\u4e8c\u503c\u5316\u56e0\u5b50\u7684\u589e\u52a0\uff0c\u8fdb\u800c\u8fdb\u4e00\u6b65\u6539\u5584FOM\u3002\u4f18\u5316\u7684\u6700\u540e\u6211\u4eec\u770b\u5230FOM\u7a0d\u6709\u53d8\u5dee\uff0c\u8fd9\u662f\u56e0\u4e3a\u4e8c\u503c\u5316\u7ed3\u6784\u8868\u73b0\u5f97\u4e0d\u5982\u8fde\u7eed\u7ed3\u6784\u3002\u4f46\u662f\u5b9e\u9645\u4e0aFOM\u662f\u6309\u534a\u5bf9\u6570\u753b\u51fa\u7684\u7ed3\u679c\uff0c\u6240\u4ee5\u7edd\u5bf9\u7684\u53d8\u5316\u662f\u5f88\u5c0f\u7684\u3002<\/p>\n

4. design for the manufacturing (DFM) phase<\/strong><\/p>\n

\u4e8c\u503c\u5316\u540e\u6709\u4e9b\u5730\u65b9\u53ef\u80fd\u6709\u5f88\u5c0f\u7684\u7ec6\u8282\/\u5b64\u5c9b\uff0c\u4e0d\u5229\u4e8e\u5b9e\u9645\u5236\u9020\u3002TopologyOptimization class\u91cc\u9762\u6709\u4e00\u4e2amin_feature_size\uff0c\u53ef\u4ee5\u6307\u5b9a\u6700\u5c0f\u534a\u5f84 filter_R\u3002<\/p>\n

geometry = TopologyOptimization2D(params=params, eps_min=eps_min, eps_max=eps_max, 
           x=x_pos, y=y_pos, z=0, filter_R=filter_R, min_feature_size=filter_R)<\/pre>\n
\u6ce8\uff1a \u6b64\u4f18\u5316\u65b9\u6cd5\u56e0\u4e3a\u9700\u8981FDTD\u5185\u90e8\u7684\u4e00\u4e9b\u53c2\u6570\uff0c\u6240\u4ee5\u5fc5\u987b\u4f7f\u75282019b R2\u53ca\u4ee5\u4e0a\u7248\u672c\u3002 <\/p>\n
\u66f4\u591a\u7ec6\u8282\u8bf7\u53c2\u8003\u4f8b\u5b50\uff0cGitHub\u548cLumApi\u3002<\/p>\n
\u867d\u7136\u4f8b\u5b50\u7528\u7684\u662f\u5355\u4e2a\u4ecb\u7535\u5e38\u6570\uff0c\u539f\u5219\u4e0a\u4e5f\u53ef\u4ee5\u7528\u8272\u6563\u6750\u6599\u3002\u4e0d\u8fc7\uff0c\u4f18\u5316\u4ec5\u5bf9\u4e00\u4e2a\u76ee\u6807\u51fd\u6570\u8fdb\u884c\uff0c\u5149\u6e90\u662f\u5355\u8272\u7684\uff0c\u5982\u679c\u9700\u8981\u5bbd\u8c31\uff0c\u5e94\u8be5\u4fee\u6539\u5149\u6e90\u548c\u76ee\u6807\u51fd\u6570\uff0c\u6bd4\u5982\u7528\u5e73\u5747\u900f\u5c04\u7387\u7b49\u3002<\/p>\n
\u6b64\u5916\uff0c\u5f3a\u70c8\u5efa\u8bae\u4f7f\u7528\u8f6f\u4ef6\u81ea\u5e26\u7684Python\uff0c\u4e0d\u8981\u81ea\u5df1\u4e0b\u8f7dPython\u548c\u5176\u7f16\u8f91\u5668\uff0c\u5426\u5219\u4e0d\u4ec5\u9700\u8981\u8bbe\u7f6e\u597d\u8def\u5f84\uff0c\u6709\u4e9b\u529f\u80fd\u8fd8\u4e0d\u652f\u6301\uff0c\u6216\u8005\u4e0d\u517c\u5bb9\u7b49\u3002<\/p>\n
<\/h2>\n
 <\/p>\n","protected":false},"template":"","class_list":["post-276405","topic","type-topic","status-closed","hentry"],"aioseo_notices":[],"acf":[],"custom_fields":[{"0":{"_bbp_subscription":["2592"],"_bbp_author_ip":["23.192.164.14"]," _bbp_last_reply_id":["0"]," _bbp_likes_count":["0"],"_btv_view_count":["1798"],"_edit_lock":["1680729986:70791"],"_bbp_topic_status":["unanswered"],"_bbp_status":["publish"],"_bbp_topic_id":["276405"],"_bbp_forum_id":["27835"],"_bbp_engagement":["2592"],"_bbp_voice_count":["1"],"_bbp_reply_count":["0"],"_bbp_last_reply_id":["0"],"_bbp_last_active_id":["276405"],"_bbp_last_active_time":["2023-03-24 22:26:51"]},"test":"gsun"}],"_links":{"self":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/topics\/276405","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/topics"}],"about":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/types\/topic"}],"version-history":[{"count":0,"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/topics\/276405\/revisions"}],"wp:attachment":[{"href":"https:\/\/innovationspace.ansys.com\/forum\/wp-json\/wp\/v2\/media?parent=276405"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}