teacher 親愛的會員您好，
代轉發資訊如附檔，提供大家踴躍參加，謝謝。
活動宗旨：
為擴大推廣「私有建築物耐震弱層補強」，本中心預計培訓弱層補強推動人員，為民眾提供法令解說、協助案件住戶整合意願，特辦理此教育訓練。
本中心將依推動人員各階段輔導進度，核予推動費，藉以激勵推動人員提供弱層補強宣導、諮詢、輔導、整合等服務，加速老舊建築物之補強，以達減少地震災害之目的。
活動資訊：
(一)活動時間：112年4月10日（星期一）13時30分
(二)活動地點：國家地震工程研究中心-台南實驗室101會議室（台南市歸仁區中正南路一段2001號）
(三)報名網址：https://conf.ncree.org.tw/index.aspx?n=A11204100 (報名人數上限80人)
*此活動無法申請建築師、技師及公務人員積分(點)，敬請注意
報名相關問題聯繫窗口：02-6630-5185高先生
敬祝 安好
中華民國結構工程學會 敬啟－03/29/2023
附件：
| 標題 | 與柱外緣切齊之鋼筋混凝土梁鋼筋細部對耐震性能影響之探討 |
|---|---|
| 作者 | 李台光、陳正誠 |
| 關鍵字 | 鋼筋混凝土梁、柱、外緣切齊、耐震性能 |
| 摘要 | 國內建築普遍存在RC梁與RC柱外緣切齊偏心接合的情況，偏心接合RC梁之韌性尚未獲得實驗驗證。本研究針對此類國內工程實務問題，規劃4座大型鋼筋混凝土梁試體，並於內政部建築研究所材料實驗中心進行實驗驗證。本研究共製作R試體(梁主筋不偏移，箍筋為正常尺寸)、RH/SU試體(梁主筋向內偏移且箍筋為縮小尺寸，側面配置U形橫向輔助筋並以2根縱向輔助筋固定)、RH試體(梁主筋向內偏移且箍筋為縮小尺寸)，以及SH/SLB試體(梁主筋向內偏移且箍筋為正常尺寸，箍筋增加2根縱向輔助筋固定)等4座試體，其中R試體做為本研究耐震性能之比較基準。研究結果發現：(1)所有試體(含3種外緣切齊的梁鋼筋配置細部)均可發展4%弧度的梁端轉角，皆能達到規範耐震要求。負方向韌性表現，以RH/SU試體為最優，RH試體次之，R試體再次之，SH/SLB試體為最差。SH/SLB試體上、下層梁主筋，皆僅有1根被箍筋角落束制，梁箍筋之角落配置縱向輔助筋，對於RC梁之耐震性能可能會造成不利的影響，因此建議RC梁箍筋之角落應配置梁主筋；(2)RH試體之一側混凝土保護層達85 mm，惟RH試體之側向強度及耐震性能皆不亞於R試體，因此實務上採用RH試體之型式，應為可接受的作法；(3)RH/SU及RH等2座試體之耐震性能皆優於R試體，惟RH試體裂縫寬度較寬且混凝土保護層剝落程度明顯較RH/SU試體嚴重，因此實務上建議採用RH/SU試體之型式。 |
| Title | Discussion on Influence of Reinforced Concrete Beam Reinforcement Detail with Flush Outside Faces of Beams and Columns on Seismic Performance |
|---|---|
| Author | Tai-Kuang Lee, Cheng-Cheng Chen |
| Keywords | reinforced concrete beams, columns, flush outside faces of beams and columns, seismic performance |
| Abstract | In domestic buildings, the eccentric beam-column connections with the flush outside faces of beams and columns are common, and the ductility of RC beams with eccentric connections has not been experimentally verified. In view of such domestic engineering practice problems, this study plans four large-scale reinforced concrete beam specimens, and conducts experimental verification at the Materials Experiment Center of the Architecture and Building Research Institute, Ministry of the Interior, Taiwan. In this study, the R specimen (the main reinforcement of the beam is not offset, and the stirrups are of normal size), the RH/SU specimen (the main reinforcement of the beam is offset inward and the stirrups are reduced in size, the side is equipped with U-shaped transverse auxiliary reinforcement and two longitudinal auxiliary reinforcements are fixed), the RH specimen (beam main reinforcement is shifted inward and stirrups are reduced in size), and the SH/SLB specimen (beam main reinforcement is shifted inward and the stirrups are normal size, and two longitudinal auxiliary reinforcements are arranged at the corners of the stirrup) are fabricated. The research results show that: (1) All specimens (including 3 beam reinforcement details with flush outside faces) can develop beam end rotation angle of 4% radian, which can meet the seismic requirement of the specification. For the seismic performance in the negative direction, the RH/SU specimen is the best, the RH specimen is the second, the R specimen is the third, and the SH/SLB specimen is the worst. There is only one top and bottom main reinforcement of the SH/SLB specimen, which is surrounded by the corner of the stirrup. The longitudinal auxiliary reinforcement is arranged at the corner of the stirrup of the beam, which may have an adverse effect on the seismic performance of the RC beam. Therefore, it is recommended that the corners of the stirrups should be equipped with the main beam reinforcement. (2) The concrete cover on one side of the RH specimen is 85 mm, but the results of this study show that the lateral strength and seismic performance of the RH specimen are not inferior to the R specimen. It is acceptable to use the type of the RH specimen in practice. (3) The seismic performance of the RH/SU and RH specimens is better than that of the R specimen, but the RH specimen has wider crack width and the degree of spalling of the concrete cover is obviously more serious than that of the RH/SU specimen. Therefore, it is recommended to use the type of the RH/SU specimen in practice. |
| 標題 | 國家地震工程研究中心新建與既有樓板接合分析與設計 |
|---|---|
| 作者 | 林昱成、莊明介、林冠泓、鄧彬斌 、蔡克銓、蔡青宜、吳安傑、林瑞良 |
| 關鍵字 | 國家地震工程研究中心、耐震設計、接頭補強、化學錨栓、有限元素模型分析、非線性反應歷時分析 |
| 摘要 | 國家地震工程研究中心為因應發展，於2019年10月至2020年11月，在既有六層樓鋼筋混凝土(reinforced concrete, RC)辦公大樓上，增建七層樓成為十三層RC、鋼骨鋼筋混凝土與鋼結構的複合結構。本文介紹二樓至七樓新建服務核樓板及鋼梁與既有 RC 結構接合處的分析與設計。本研究以梁腹側角鋼加勁，同時考量托架有無封板的型式，探討三種接合設計，並進行一系列的有限元素分析。經Abaqus有限元素模型推估接點元素在交界面的軸向勁度為100 tf/mm，配合PISA3D結構分析軟體建立十三層增建大樓的結構模型，進行16組地震的非線性反應歷時分析，二至七樓新舊樓板採雙質心雙剛性樓板的設定，並用6自由度接點元素來分析新舊樓板交界面接頭的受力。經PISA3D歷時分析結果可得二至七樓服務核與既有 RC 結構交界處之單一梁端最大軸力以及最大層間位移角，分別為165 tf以及 1% rad。本研究以保守方式，在不考慮樓板的貢獻，以上述力量和變形作為接頭的設計需求，再對各個接合設計進行 Abaqus 有限元素分析，接頭錨栓採用只受拉之線性彈簧，另以只受壓之線性彈簧來模擬接頭端鋼板抵靠於 RC 面之邊界條件，並在梁端反曲點 3,800 mm 處同步施以軸拉 1.5 mm 與面內向上 40 mm 之位移，來模擬設計地震下之設計需求。分析結果顯示基本設計的接頭型式，直托架接合之托架腹板產生明顯剪力降伏並可觀察到錨栓受力不均的情形。三種改良方案中，比較結果後可知新增封板能使錨栓最大受力降低 35% 以上，且封板傳力使錨栓受力分布均勻的效果顯著。此外，在封板角落可切除短邊長為200 mm之等腰直角三角形低應力區塊，以利未來錨栓受震後的安全性檢查。本研究確認以雙腹側角鋼加勁及直托架封板切角式為最佳方案，除能符合設計需求，可使梁發展出韌性較好之彎矩塑鉸，並具可供未來檢視錨栓的開孔。 |
| Title | Analyses and designs of the connections jointing the new steel structure and the existing RC structure in NCREE’s office building extension |
|---|---|
| Author | Yu-Cheng Lin, Ming‐Chieh Chuang, Guan-Hong Lin, Pin-Pin Deng, Keh-Chyuan Tsai, Ching-Yi Tsai, An-Chien Wu, Jui-Liang Lin |
| Keywords | seismic design, strengthened connection, chemical anchor, finite element analysis, nonlinear response history analysis |
| Abstract | The existing six-story office building of the National Center for Research on Earthquake Engineering (NCREE) was extended to thirteen-story from October 2019 to November 2020. This study presents the analysis and design of the connections jointing the existing RC structure and the new steel service core from the second to seventh floors. A steel bracket, which is welded to the steel beam bottom flange and connected to the RC column using chemical anchors, was considered as the basic design of the connection to transfer the gravity shear from the steel beam. This shear connection detail could avoid the congested reinforcing bars inside the RC beam-to-column joint, however, three alternative designs using strengthened schemes are investigated in this study. Nonlinear response history analyses (NRHAs) using PISA3D program for the NCREE’s 13-story composite building were conducted in order to gain insight into the possible maximum seismic force and deformational demands on the connections. The features of dual mass centroids and dual rigid diaphragms for the second to seventh floors are incorporated into the PISA3D model. The 6DOF joint elements are utilized to represent the aforementioned bracketed connections on the interface. The axial push-pull stiffness of 100 tf/mm, computed from an Abaqus finite element model (FEM) analysis for the connections, is utilized for all the 6DOF joint elements. A total of sixteen sets of ground accelerations are utilized. NRHA results indicate that the maximum axial force and maximum in-plane rotation of the connection at the steel beam end are about 165 tf and 0.01 radian, respectively. The Abaqus FEM analyses were conducted for investigation of four connection types including the basic design and three improved designs. The FEM analysis results indicate that the beam bracket of the basic design exhibits obvious shear yielding. The uneven anchor bolt forces developed are very evident. The improved load-carry capacity and a more uniform bolt forces developed are found in the strengthened designs, such as the stiffened steel angles at the beam web and bracket with cover plate. In addition, an approximately 200×200 mm triangular region on the cover plate corner with a low stress demand is observed. Based on these connections’ FEM analysis results, this study confirms that the strengthened scheme adopts double-sided stiffened steel angles at the beam web, the straight bracket with the angular cuts on the cover plates is suitable. The strengthened scheme adopted in the project not only meets the seismic demands but also allows the inspections of the anchor bolts in the future, if necessary. |
| 標題 | 擴頭鋼筋單體受拉之力學特徵與試驗裝置開發 |
|---|---|
| 作者 | 林克強、歐宸宇、紀凱甯、莊勝智、廖文義 |
| 關鍵字 | 擴頭鋼筋、擴頭鋼筋單體試件拉力試驗裝置、擴頭部滑動量、伸長率、拉力強度 |
| 摘要 | 本文主要目的在探討台灣常用擴頭鋼筋之單體試件受拉的力學特徵，並開發符合時間效益的試驗裝置。本研究進行四種鋼筋強度等級（SD 420W、SD 490W、SD 550W與SD 690）共41組之擴頭鋼筋單體受拉試驗，擴頭部包括摩擦銲、銲接螺桿、擴頭滾牙與螺紋套筒灌漿等型式。試驗之滑動量結果顯示，試驗採用0.7Py與0.95Py為加載上限值，可獲得相同的合格與否鑑識結果，其中Py為鋼筋最小規定降伏強度。摩擦銲式擴頭鋼筋幾乎無滑動量。其餘滑動量由小至大分別為：限滑擴頭滾牙式、螺桿銲接式、擴頭滾牙式、與螺紋套筒灌漿式擴頭部，其中，非螺紋套筒灌漿式擴頭部之滑動量均不大於0.2 mm。對於螺紋套筒灌漿式擴頭鋼筋，有限制滑動裝置者，只要採用適當限滑裝置且施工適當時，其滑動量多能控制在0.3 mm內；無限制滑動裝置者，其滑動量均大於0.3 mm。本研究之強度試驗結果顯示，所有試體的最大強度均能發展鋼筋規定降伏強度的1.25倍與規定抗拉強度。本研究開發完成之擴頭鋼筋單體試件拉力試驗裝置能迅速並目視架設量測設備，以獲得準確的擴頭部滑動量，且較原試驗裝置可縮短約78%的試驗時間，該試驗裝置測試目標上限為SD 690之#12擴頭鋼筋。 |
| Title | “Mechanical Characteristics and Testing Device Development of Headed Reinforcements for Uniaxial Tension Test” |
|---|---|
| Author | Ker-Chun Lin, Chen-Yu Ou, Kai-Ning Chi, Sheng-Jhih Jhuang, Wen-I Liao |
| Keywords | headed reinforcement, testing device to test single headed reinforcement in tension, slip of head part, elongation, tensile strength |
| Abstract | This paper mainly investigates mechanical characteristics of headed reinforcements in tension popular used in Taiwan and develops a fixture for testing to effectively shorten test time. A total of 41 headed reinforcement specimens that include four types of steel grade, SD 420W、SD 490W、SD 550W and SD 690, and three main types of head part, welded head, threaded head, grouted sleeve head. Test results related to head part slip showed that upper loading bounds of 0.7Py and 0.95Py to assess head part slip can acquire the same identified results those are qualified or not. The slip results for the various head parts indicated that the slip of welded head by friction was very limit. The slips of head part with the other heads from small to large in sequence were threaded head and grouted sleeve head. For the threaded heads of the headed reinforcements, all the slips were smaller than 0.2 mm. For the grouted sleeve heads of them, the slips of head part without a restraining-slip nut were more significant than 0.3 mm, but the slips of head part with a restraining-slip nut enable to be reduced under 0.3 mm as long as using proper fabricated process. Test results also presented that the maximum tensile strengths of each specimen exceeded its minimum specified ultimate strength and 1.25 times its minimum specified yield strength. The testing device developed in this study can install measure instruments quick and visually and get reliable measure results of head part slip. It also significantly reduces 78% of testing time per piece compared with the existed testing fixture. The load capacity of the fixture is for #12 reinforcement of SD 690. |
| 標題 | 應用集中塑性模型於鋼筋混凝土結構非線性動態行為之研究 |
|---|---|
| 作者 | 蕭輔沛、盧煉元、黃炫文、鄭弘 |
| 關鍵字 | 集中塑性模型、RC非線性、振動台實驗、TEASDA、ASCE 41、ETABS、OpenSees、ModIMK |
| 摘要 | 近年來國內所發生之重大地震皆造成中高樓鋼筋混凝土建物倒塌，進而導致重大人員傷亡及財產上嚴重的損失，包含南投集集921大地震、0206高雄美濃大地震及0206花蓮大地震。有鑑於此，準確模擬中高樓鋼筋混凝土建物於地震力作用下之非線性結構反應，對於鋼筋混凝土結構之耐震研究是相當重要的課題。本文擬藉由商用軟體ETABS及美國加州柏克萊大學所開發之開放式有限元素軟體OpenSees分別建立不同塑鉸之七層樓鋼筋混凝土構架模型，以探討不同塑鉸之設定對於非線性動力歷時分析結果的影響。接著再根據七層樓鋼筋混凝土構架之振動台實驗結果，驗證何種塑鉸分析結果較接近真實結構反應，期能提供工程界於塑鉸設定方面之依據。再者，由於ETABS為商用軟體，其應用上較為受限，使用者無法自行加入新開發之材料及元素或與實驗控制系統連接進行實驗，因此本文提出將ETABS模型等效轉換至開放式有限元素軟體OpenSees之方式，期能使研究者藉由ETABS為輔助建立等效之OpenSees模型，以運用OpenSees開放式軟體之特性進行分析。 根據本文研究結果顯示，以ETABS建立由國家地震工程研究中心所開發之TEASDA塑鉸且配合Pivot遲滯模型之設定，其分析結果與七層樓鋼筋混凝土構架之振動台實驗結果最為接近。此外，根據各不同塑鉸分析結果顯示，塑鉸設定之遲滯模型不同對於分析結果影響有限，而塑鉸設定之背骨曲線不同對於分析結果才會有所影響。最後由ETABS塑鉸及OpenSees塑鉸分析比較之結果可知，本文成功驗證以ETABS模型等效轉換為OpenSees模型方式之正確性。 |
| Title | A Study on Nonlinear Dynamic Behavior of Reinforced Concrete Structures using Concentrated Plasticity Model |
|---|---|
| Author | Fu-Pei Hsiao, Lyan-Ywan Lu, Hsuan-Wen Huang, Hung Cheng |
| Keywords | concentrated plasticity model, nonlinear RC, shaking table test, TEASDA, ASCE 41, ETABS, OpenSees, ModIMK |
| Abstract | In recent years, some strong earthquakes, including the 921 Chi-Chi Earthquake, the 0206 Meinong Earthquake, and the 0206 Hualien Earthquake, have caused the collapse of med-to-high rise reinforced concrete (RC) buildings, and resulted in heavy casualties and severe property losses. Because of this, it is crucial to perform accurate numerical simulation for the nonlinear structural response of med-to-high rise RC buildings. This paper establishes a 7-story RC frame model with different plastic hinges using the commercial software ETABS and the open-source finite element software OpenSees developed by the University of California at Berkeley, USA. The effect of different plastic hinge settings on nonlinear time history analysis of the RC building is explored in this study. Then, based on the shaking table test results of the 7-story RC building, it is judged which plastic hinge analysis results are closer to the experimental structural response. Furthermore, since ETABS is a commercial software, its extensibility is relatively limited. In other words, users cannot add newly developed materials and elements to ETABS, nor connect ETABS with the experimental control system to conduct experiments. Therefore, this paper proposes a method to convert an ETABS model to a OpenSees model. This method will enable researchers to establish the plastic hinges for an equivalent OpenSees model with the help of ETABS, and to take the openness advantage of OpenSees software for analysis. |
| 標題 | 基於FEMA P58法之區域韌性評估：以美國洛杉磯非韌性鋼筋混凝土構架為例 |
|---|---|
| 作者 | 陳鵬宇、Ertugrul Taciroglu |
| 關鍵字 | 機率式地震風險評估、區域評估、FEMA P58、非韌性鋼筋混凝土構架、增量動力分析、韌性指標 |
| 摘要 | 都會地區房屋密集，遭遇地震時產生之性能表現係決定災損之重要依據。然而既有之結構性能評估所採用之非線性靜力法為一定量式分析(deterministic)，一來無法有效地將地表運動以及結構動態反應之不確定性納入，二來其結果以結構反應參數來評估其耐震力，不利於非具工程背景之業主進行決策。另一方面，近來國際上朝向發展韌性城市(resilient city)，對於結構工程師而言，如何量化房屋受震後之韌性以及進行大範圍之區域評估尚不清楚。有鑑於此，本研究提出一區域韌性評估之架構，當中依據機率式地震風險評估法(即FEMA P58)，將增量動力分析與蒙地卡羅模擬結合進行災損評估，並依據評估結果量化成韌性指標(resilient index)。為示範此架構之應用，本研究以美國洛杉磯地區之非韌性(nonductile)鋼筋混凝土構架為例，透過開發之自動化建模程序完成1,452棟房屋之災損及韌性評估，當中完成了95萬筆非線性動力歷時分析。評估結果顯示在最大考量地震下的平均損失比為37.3%，震後近一年的平均韌性指標不到50%，且所有房屋恢復其原有功能需要至少三年時間。本研究藉由此案例說明，介紹FEMA P58法應用於區域評估之流程，後續期望能將此架構本土化應用於災前評估與災後應變，來提升台灣的都市韌性。 |
| Title | Regional evaluation of resilience based on FEMA P58: a case study for nonductile reinforced concrete frames in Los Angeles |
|---|---|
| Author | Peng-Yu Chen, Ertugrul Taciroglu |
| Keywords | Probabilistic seismic assessment, regional evaluation, FEMA P58, nonductile reinforced concrete frames, incremental dynamic analysis, resilient index |
| Abstract | Seismic damage and loss assessment are highly related to buildings’ performance. However, the current nonlinear static analysis for performance evaluation is a deterministic methodology, where the uncertainties and variation of ground motion and numerical modeling cannot be considered. Furthermore, the outcome of engineering parameters is difficult to be utilized by decision-makers who may not have engineering backgrounds. While the world is toward developing resilient city, it is still not clear to structural engineers how to quantify the seismic resilience and evaluate it for large-scale regions. Hence, this research proposes a framework for regional resilience evaluation, which is based on probabilistic seismic assessment (i.e., FEMA P58) to incorporate incremental dynamic analysis and Monte Carlo simulation for damage and loss assessment. Moreover, the outcome of the regional evaluation is used to quantify the resilience index to illustrate the ability of a city for recovering from an earthquake. To demonstrate the application of the proposed framework, 1,452 nonductile reinforced concrete frames in Los Angeles are simulated by developing an automatic modeling program. Around 950,000 nonlinear time history analyses are conducted through a supercomputer, and the outcomes are used for loss estimation and resilience quantification. The results show that the mean loss ratio for nonductile frames under maximum-considered earthquake is 37.3%, and the resilience index indicates that the city needs at least 3 years to recover. While the presented work is a US-based case study, the authors hope the framework can be extended and localized for Taiwan’s development of resilient city. |
親愛的會員您好，
代轉發資訊如附檔，提供大家踴躍參加，謝謝．
會議時間：2023.03.28(二)~03.29(三)
會議地點：國家地震工程研究中心13樓(台北市大安區辛亥路三段200號)
報名費用：免費
專業證書：技師研習積分及公務人員終身學習時數認證
報名網址：https://conf.ncree.org.tw/Index.aspx?n=ITJW20230
敬祝 安好
中華民國結構工程學會 敬啟－03/21/2023
附件：
親愛的會員您好，
代轉發淡江大橋施工實務及技術應用研討會資訊如附檔，提供大家踴躍參加，謝謝．
會議時間：2023.04.12(三) 13:30 – 16:30
會議地點：張榮發基金會國際會議中心10樓1001廳(台北市中正區中山南路11號)
報名費用：免費
專業證書：技師、建築師訓練積分申請
若有任何意見歡迎email至csse@csse.org.tw
敬祝 安好
中華民國結構工程學會 敬啟－03/20/2023
親愛的會員您好 :
代轉發國家地震工程研究中心土耳其及池上地震災損與
防災技術研討會，敬請踴躍參加，檔案如附，謝謝．
台北場
時間：2023.03.21 (二) 09:00 – 12:30
地點：國家地震工程研究中心台北實驗室101會議室
費用：免費
台南場:
時間：2023.03.22 (三) 09:00 – 12:30
地點：國家地震工程研究中心臺南實驗室S101演講廳
費用：免費
中華民國結構工程學會敬啟-3/10/2023
親愛的會員您好:
轉寄以下訊息
國家地震工程親愛的會員研究中心(國震中心)特別邀請東京工業大學笠井和彥(Kazuhiko Kasai)教授，於2023年3月2日在國震中心進行兩場專題演講，講題為「Full-Scale Shake Table Test and Analysis of Steel Building Collapse Caused by Two-Directional Column Deteriorations」及「Full-Scale Shake-Table Tests and Analyses of 5-story Buildings with/without Dampers」，分享實尺寸鋼構建築振動台試驗之最新研究成果。
Kazuhiko Kasai教授為國際鋼結構、結構控制和地震工程領域之知名研究者，現任東京工業大學研究員，亦為該校之名譽教授。在過去22年間，Kasai教授擔任日本隔震學會（JSSI）中結構控制委員會主席和被動控制委員會副主席，於日本建築物控制技術發展貢獻卓越；Kasai教授曾任建築學會（AIJ）被動控制委員會主席，亦為E- Defense鋼構建築研究中美日項目的日本代表，主導一般結構和被動控制結構的實全尺寸振動試驗。Kasai教授同時也是國際日本科學技術機構合作項目的日本代表，重點研究高層建築物的地震評估和減震技術。
敬邀您與您的研究團隊出席，演講資訊如下:
時間：2023年3月2日(星期四)上午10時00分至上午12時00分
地點： 國家地震工程研究中心台北實驗室101會議廳(台北市辛亥路三段200號)
費用：免費
專業證書：技師換證積點、公務人員終身學習護照
名額：120人，額滿為止。
報名網址: https://conf.ncree.org.tw/index.aspx?n=A11203020
本研討會已申請技師換證積點，以及公務人員終身學習護照等相關證書。議程表請詳研討會報名網站，誠摯歡迎您透過研討會網站報名參與本研討會。
講者與演講資訊: 詳如附件
聯絡人: 游忠翰副研究員(e-mail: chyu@narlabs.org.tw)
附件：
活動時間：112年2月18日（星期六）13時00分
活動地點：國家地震工程中心101會議室（台北市大安區辛亥路三段200號）
活動課程：
1.耐震弱層補強計畫簡介與審查注意事項 (講師：國家地震工程研究中心)
2.耐震弱層補強工法之設計、評估與施工-開口磚牆耐震補強工法 (講師：施忠賢結構技師事務所 施忠賢技師)
3.耐震弱層補強工法之設計、評估與施工-軟補強典型街屋簡易補強工法 (講師：國立成功大學建築學系 杜怡萱教授)
4.耐震弱層補強工法之設計、評估與施工-超高性能纖維混凝土UHPC補強工法 (講師：國立成功大學土木工程學系 洪崇展教授)
*報名網址：https://conf.ncree.org.tw/indexCht.aspx?n=A11202180 (報名人數上限80人)
*此活動可申請建築師、技師及公務人員積分(點)
本期期刊為特刊
人工智慧在結構工程之應用特刊
客座主編：林子剛
「結構工程」季刊第146期(第37卷第4期)
電子檔已經上線。歡迎會員登入後下載全文。
中文網址: 第37卷第4期 (2022)
英文網址: Vol.37 / No.4 (2022)
若有任何意見歡迎 email 至 csse@csse.org.tw。
人工智慧在結構工程之應用特刊
客座主編：林子剛
| 標題 | 「人工智慧在結構工程之應用」引言 |
|---|---|
| 作者 | 林子剛 |
|
近年人工智慧(AI,artificial intelligence)正對各個領域帶來前所未有革命性的影響，土木結構工程亦朝向「智慧化」邁進。結構工程導入人工智慧，運用在大量的資料彙整、分析及尋找規律上，透過AI 能有效且快速地節省人力，可直接對接進入施工過程，讓大量常規項目規劃和設計變得快速，比起以往人工判讀更具備效能與效率。美國國家科學基金會（NSF）開發了一套名為BRAILS（大規模人工智慧建築識別）的工具，可以檢測城市建築在地震、颶風或海嘯中可能面臨的風險，此即是人工智慧結合土木結構工程的案例之一。 結構工程期刊企盼藉由此次特刊的機會，向各界介紹國內人工智慧在結構工程應用之成果。本特刊收錄了「結合神經網路熵與卷積神經網路於結構健康診斷」、「使用機器學習和統計資訊進行異常訊號檢測和分類」、「深度學習於提升地震模擬振動臺加速度性能之研究」、「含巡檢規劃、劣化辨識、損傷量化之進階建物劣化檢測架構」及「基於深度學習之合成配電網絡建置研究」。希冀藉由這些篇章的介紹，能更深入瞭解人工智慧在結構工程之應用發展與趨勢，讓臺灣在土木結構工程專案管理實務中，嘗試建立類似的工具，進而保障建築安全。 國立陽明交通大學土木工程學系 林子剛 教授 謹誌 |
| 標題 | 含巡檢規劃、劣化辨識、損傷量化之進階建物劣化檢測架構 |
|---|---|
| 作者 | 許舜翔、許筠曼、洪昊天、林禹齊、張家銘、陳宗珷、陳俊仲 |
| 關鍵字 | 目視檢測、損傷偵測、損傷量化、深度學習 |
| 摘要 | 目視檢測常用於結構檢測中，以快速評估建物中構件或材料之耐用性，及尋找可能的損傷。然而，傳統目視檢測相當耗費人力及時間，鑒於以上問題，本研究開發一個進階的建物劣化檢測架構，其中功能包括巡檢規劃、即時劣化識別、損傷程度判別及記錄詳細的損傷狀況。於該架構中，首先規劃巡檢草圖，提供檢測時最有效率檢測路徑；接著本研究使用Scaled-YOLOv4，進行損傷物件偵測，該方法於大規模場域中，也能快速檢測損傷，並使用SOLOv2 模型，對混凝土裂縫位置進行像素等級之實例分割，以利更精確的裂縫量化；最後，根據檢測到的劣化嚴重性及面積大小，對構件的損傷程度進行評級。本研究以一所小學之走廊進行實際場域驗證，目的為檢測和量化混凝土構件之表面劣化。由驗證結果可見，藉由本研究之進階建物劣化檢測架構，於劣化辨識、損傷量化和檢測效率上，相較傳統建物目視檢測都獲得顯著改善。 |
| Title | Framework of Advanced Building Inspection withRoute Planning, Defect Detection, and Damage Rating |
|---|---|
| Author | Shun-Hsiang Hsu, Ho-Tin Hung, Yun-Man Hsu, Chia-Ming Chang, Tzung-Wu Chen, Chun-Chung Chen |
| Keywords | visual inspection; damage detection; damage quantification; deep learning |
| Abstract | Visual inspection is commonly adopted for building operation, maintenance, and safety. The durability and defects of components or materials in buildings can be quickly assessed through visual inspection. However, implementations of visual inspection are substantially time-consuming, labor-intensive, and error-prone because useful auxiliary tools that can instantly highlight defects or damage locations from images are not available. Therefore, an advanced building inspection framework is developed and implemented with route planning, realtime and detailed damage recognition, and damage rating in this study. The inspection route sketching is first exploited to provide an efficient plan with significantly reduced disruption. Then, Scaled-YOLOv4 and SOLOv2 models are considered in this study to detect defects even in a large-scale field quickly and acquire pixel-level damage recognition for more precise quantification, respectively. Finally, damage levels of components are rated following the importance and numbers per unit area of the detected defects. This entire framework is also implemented and verified by the hallway of an elementary school to detect and quantify surface damage of concrete components. As seen in the results, the conventional building inspection is significantly improved by the aid of the proposed framework in terms of damage localization, damage quantification, and inspection efficiency. |
| 標題 | 圖神經網路於線性靜力分析之應用 |
|---|---|
| 作者 | 周遠同、黎光曜 、郭柏志、張慰慈、黃尹男、陳俊杉 |
| 關鍵字 | |
| 摘要 | 結構設計是朝向最佳化的迭代過程，目前多依賴結構工程師過往的經驗結合結構分析的結果完成。結構最佳化設計需反覆進行結構分析，如何加速結構分析則是重要的突破關鍵。本研究使用深度學習模型作為線性靜力分析之代理模型，提供快速即時且精準之結構反應預測。因著結構物之空間幾何關係與圖資料結構之間之相似性，本研究將結構物表達成圖，並以圖神經網路模型去學習結構物所受之外力與結構物反應之間的關係。在透過由結構軟體所生成包含隨機樓層數、跨數、樑柱長度，以及隨機大小的側向外施載重之結構分析資料集訓練後，圖神經網路模型不僅顯示了其在預測位移以及力具有良好之表現，模型也具有很好之泛化能力，能夠預測在訓練時從未接觸過的、更高的結構物。在特徵重要性分析中也顯示模型所學習到之特徵具有一定之物理意義。 |
| Title | Linear Static Analysis with Graph Neural Networks |
|---|---|
| Author | Yuan-Tung Chou, Kuang-Yao Li, Po-Chih Kuo, Wei-Tze Chang, Yin-Nan Huang , Chuin-Shan Chen |
| Keywords | |
| Abstract | Structural design is an iterative process for optimum selection, which relies on structural analysis results and experience from structural engineers. Since iterative structural analysis is a necessary for getting a good design, accelerating structural analysis is an important task. In this work, we adopt deep learning approaches as a surrogate model for linear static analysis, which provides fast and accurate structural response prediction. Based on the similarity between the structure’s topology and graph data structure, we represent structures as graphs and leverage graph neural networks (GNNs) to learn the relationship between given external forces and corresponding structural responses. The GNN model is trained with random-generated structures, including random story number, span number, beam-column length, and value of external forces. The results show that the GNN model has good performance on displacement and force predictions and excellent generalizability on unseen, taller structures. In addition, it is shown that based on the analysis of feature importance, the GNN model can extract important physical attributes from the input features. |
| 標題 | 使用機器學習和統計資訊進行異常訊號檢測和分類 |
|---|---|
| 作者 | 林天郇、黃謝恭、周肇昱 |
| 關鍵字 | 異常訊號、機器學習、模式識別網路、GoogLeNet |
| 摘要 | 結構健康監測（structural health monitoring，SHM）和結構完整性管理（structural integrity management，SIM）為逐漸興起的技術，為了持續檢測結構狀態並不斷追蹤結構劣化，海量的資料伴隨著不正常的量測數據正不斷的被產生，至此，檢測與分類這些帶來問題的異常訊號大多依靠人工，這不僅費時費力，工作更是乏味。在本研究中，檢測與分類異常訊號的工作將被機器學習的技術取代，並嘗試利用統計學的優勢來提高檢測與分類的正確性，模式識別網路（pattern recognition network）被採納來使用一維的輸入資料，而GoogLeNet 被引入來使用二維的輸入資料，並利用一組現地量測資料進行學習，其結果顯示兩種機器學習的技術均可有效的檢測與分類異常訊號，並且其算力需求與正確性是個權衡下的結果，因此對於兩種技術的使用情境將可視應用的需求而定。 |
| Title | DATA ANOMALIES DETECTION AND CLASSIFICATION USING MACHINE LEARNING AND STATISTIC INFORMATION |
|---|---|
| Author | Tian-Xun Lin, Shieh-Kung Huang, Jau-Yu Chou |
| Keywords | data anomaly, machine learning, pattern recognition network, GoogLeNet |
| Abstract | Structural health monitoring (SHM) and structural integrity management (SIM) are emerging recently. To continuously track the condition and constantly detect early deterioration of the infrastructure, huge amounts of data are produced and abnormal measurement is inevitable. The corrupted data can produce a lot of problems and, generally, they are examined and classified by humans. In this study, the detection and classification are replaced by the techniques of machine learning (ML) and improved by using statistic information. The neural networks based on 1-dimensional and 2-dimensional data are studied via a field dataset collected from a long-span cable-stayed bridge. Therefore, a shallow network, called pattern recognition network, is selected to use 1-dimensional data as an input and a deep network, GoogLeNet, is selected to use 2-dimensional data. The results show that both models can detect and classify the data anomalies and the usage depends on the assigned application and the trade-off between computation and performance. |
| 標題 | 結合神經網路熵與卷積神經網路於結構健康診斷系統應用之研究 |
|---|---|
| 作者 | 林子剛、林易廷、郭凱維 |
| 關鍵字 | 結構健康監測、神經網路熵、卷積神經網路 |
| 摘要 | 本研究以神經網路熵(Neural Network Entropy, NNetEn)為研究基礎，並將此熵分析方法與卷積神經網路(Convolutional Neural Network, CNN)結合，發展出一套具實用性之結構健康監測系統。為了驗證其系統之可行性，將進行七層樓鋼構架的破壞實驗，並建置與鋼構架相同結構特徵之數值模型。首先使用狀態空間法建立數值模型，模擬與鋼構架相同之十六種破壞模式，並將破壞時之各樓層加速度訊號以神經網路熵進行分析，建立熵值資料庫，再將此熵值資料庫用以訓練類神經網路模型。前期研究之熵值是藉由人為觀察後定義其閥值，藉以判斷結構是否破壞，因此為了避免人為因素的誤判及自動化判讀，本研究以可視化之 heatmap 量化熵值變化，並選用適用於影像處理的卷積神經網路分析，透過將熵值轉為圖像資料之方式不僅能夠減少模型中之參數量還能提升其運算速度。神經網路模型在訓練過程中藉由提取熵值中的破壞特徵並學習，在訓練完成後模型即能在識別輸入資料的破壞特徵後定位結構物之破壞區域。最後，透過國家地震工程研究中心之七層樓鋼構架驗證所設計之十六種破壞案例，逐例討論模型的輸出值並以混淆矩陣量化數值模擬和實驗驗證之預測結果。本研究提出的結構健康診斷系統，將新興之熵分析方法結合類神經網路，其最終驗證實驗之測試樣本結果其準確率 93.13%。 |
| Title | Application of Convolution Neural Network and Neural Network Entropy Algorithm for Structural Health Monitoring |
|---|---|
| Author | Tzu-Kang Lin, Yi-Ting Lin, Kai-Wei Kuo |
| Keywords | Structural Health Monitoring, NNetEn, Convolution Neural Network |
| Abstract | This study combines Neural Network Entropy (NNetEn) and Convolutional Neural Network (CNN) to develop a practical structural health monitoring system. In order to verify the feasibility of the system, the failure experiment of a seven-story steel frame has been carried out with a numerical model of the same structural characteristics as the steel frame. The state space method is used to simulate the sixteen failure modes on the steel frame, where the acceleration signals of each floor at the time of failure are analyzed by neural network entropy. An entropy database is established based on the model to train the neural network model. To avoid the misjudgment and automatic interpretation of human factors, this study uses the visualized heatmap to quantify the change of entropy value, and the convolutional neural network analysis is selected for image processing. By converting the entropy value into image data, not only the number of parameters in the model can be reduced, but its operation speed can be improved. During the training process, the neural network model extracts and learns the damage features in the entropy value. After the training is completed, the model can allocate the damage area of the structure by identifying the damage features of the input data. Finally, through the verification of 16 failure cases simulated on the seven-story steel frame of the National Center for Research on Earthquake Engineering (NCREE), the performance of the proposed SHM system is evaluated by both numerical simulation and experimental verification with confusion matrix. The SHM system proposed in this study combines the emerging entropy analysis method with a neural network. The test results of the final verification have an accuracy rate of 93.13%. |
| 標題 | 深度學習於提升地震模擬振動臺加速度性能之研究 |
|---|---|
| 作者 | 賴魁星、陳沛清 |
| 關鍵字 | 地震模擬振動臺、加速度控制、三參數控制器、深度學習、長短期記憶神經網路 |
| 摘要 | 地震模擬振動臺是地震工程研究重要的實驗設備，已廣泛應用在各種結構系統的耐震性能試驗，諸如鋼筋混凝土結構、鋼結構、隔減震結構以及精密設備等，因此振動臺加速度控制之精確與否在振動臺實驗中特別重要。本研究提出使用深度學習的方式， 使用了長短期記憶(Long Short-TermMemory, LSTM)神經網路，透過大量實驗資料訓練出振動臺的外迴路控制器。本研究使用油壓驅動之大型單軸向振動臺，設計一組鋼構造試體安裝其上並進行大量的振動臺實驗，以得到LSTM 之訓練資料，再將訓練所得到的LSTM 作為前饋控制器，加裝在振動臺既有的三參數控制架構中，藉此補償振動臺加速度控制的動態響應。實驗結果證明LSTM 前饋控制器可有效地降低振動臺的加速度誤差，並減少試體與振動臺的互制現象，使振動臺之加速度性得到顯著的提升。本研究成果顯示應用深度學習於提升振動臺加速度表現的發展潛能，未來可進行更多的相關研究，以改善振動臺實驗之測試品質。 |
| Title | Improved Acceleration Tracking Performance of Seismic Simulators using Supervised Deep Learning |
|---|---|
| Kui-Xing Lai and Pei-Ching Chen | |
| Keywords | Seismic shake table; acceleration control, three-variable control; deep learning, long short-term memory neural network |
| Abstract | Seismic shake table testing has been widely used for various structural systems such as steel structures, reinforced-concrete structures, energy-dissipated and base-isolated buildings, and nonstructural components etc. Therefore, accurate replication of shake table acceleration is particularly important to these tests. In this study, supervised deep learning approach is applied as an alternative for seismic shake table control. The Long Short-Term Memory (LSTM) neural network is built for training the controller to improve acceleration performance of the shake table. A large-scale servo-hydraulic uniaxial shake table is adopted. A steel specimen is designed and fabricated for performing a large number of shake table tests. Then, the shake table testing data are used to train a feedforward controller using LSTM which is implemented close to an existing Three-Variable Control (TVC) loop. The validating experimental results prove that the acceleration tracking performance is improved compared with conventional TVC. The control-structure interaction is also suppressed. The experimental results demonstrate the proposed control scheme reduces the acceleration tracking error effectively compared with conventional TVC control. The research results also show great potential for deep learning application to seismic shake table control in the future. Keywords: Constitutive model, anisotropy, shear-slip and re-contact, mesh-sensitivity, non-proportional loading, concrete, finite element |
| 標題 | 基於深度學習之合成配電網絡建置研究 |
|---|---|
| 作者 | 林岳鴻、林其穎 |
| 關鍵字 | 合成配電網絡、深度學習、物體識別、地理標籤 |
| 摘要 | 現今時代處於能源高消耗的處境且加上地球暖化帶來的災害規模日益嚴重，使人們對於公共資產管理、減少能源消耗、預測災害損失更加重視。本研究提出利用深度學習建置合成電力網路，以便未來使用其擬真模型進行電網可靠度分析。本研究利用街景圖配合物體識別及計算交集的方式定位出電線桿之地理資訊，並利用此地理資訊建立其合成電力網路。研究首先使用Mask R-CNN 與YOLOv4 物體識別模型，進行時長之條件控制訓練，並比較兩種模型的精確度得出適合此研究之方法。接續調整其參數配置，比較不同參數下對於此研究的最佳模型設置。對於區域的研究進行每條街的個別物體識別，並設計兩種排序方式進行電線桿排序，分別為經緯度以及最短路徑，利用兩種排序方式進行基於馬爾科夫隨機場(Markov random field,MRF)和基於方位線(line of bearing, LOB)與基於密度之聚類演算法(Densitybasedspatial clustering of application with noise, DBCSCAN)兩種方式計算交集，接著進行電線桿之補遺，使其更符合真實電線桿最大容許間距。最後，利用距離迭代的方式合併所有街道並刪除重複預測的電線桿點位，得出四種結果。使用四種結果與真實電線桿地理資訊進行精度比對，選擇出最適者並利用最小生成樹(minimum spanning tree, MST)建立電力網路，建立後分析並比較其性質與真實電力網絡的區別。未來可使用此模型進行改善，使其更符合真實電力網路，並使用其合成電力網絡進行電網可靠度分析、公共資產管理、災害分析等分析。 |
| Title | Synthetic Power Distribution Network Construction Based on Deep Learning Algorithm |
|---|---|
| Yue-Hung Lin and Chi-Ying Lin | |
| Keywords | synthetic power distribution network, deep learning, object detection, geo-positioning |
| Abstract | Global warming has caused high energy consumption and an increasing scale of disasters, which make people draw more attention to public asset management to reduce energy consumption and predict losses caused by disasters. Based on a deep learning based object detection approach, this study develops a synthetic power distribution network that can serve as an alternative to the real power distribution net-work and be used to analyze its reliability. This research uses the street view images to detect utility poles and conduct geo-positioning to locate utility poles on the map. For object detection, the Mask R-CNN and YOLOv4 are trained with controlled du-ration, and then the accuracy of the two models is compared to determine which method is suitable for this research. Second, the model’s hyperparameters are adjusted and compared to determine the best model setting for the object detection task in this study. Then the selected model is used to perform the object detection task for each street in the research region. Two sorting methods, namely, the latitude and longitude sorting method and the shortest path sorting method, are proposed to sort the poles for pole geopositioning and supplementation. With two sorting methods, pole geopositioning is conducted based on two approaches: The first is the Markov random field (MRF) approach, and the second is the line of bearing (LOB) with density-based spatial clustering of application with noise (DBSCAN). After determining the detected pole location, pole supplementation is conducted to ensure the maximum allowable distance between poles. Third, four sets of results are obtained by merging all streets and removing duplicate poles by means of distance iteration. Finally, four results are compared with the coordinates of real utility poles. The most suitable method for study region is selected to establish the synthetic power distribution network using the minimum spanning tree (MST). In the future, this model can be improved to make it more in line with the real power distribution network, and the synthetic power distribution network can be used for power grid reliability analysis, public asset management, disaster analysis, power demand-supply analysis, etc. |
Special Issue: Application of Artificial Intelligence in Structural Engineering
Guest Editor: Tzu-Kang Lin
| Title | Framework of Advanced Building Inspection withRoute Planning, Defect Detection, and Damage Rating |
|---|---|
| Author | Shun-Hsiang Hsu, Ho-Tin Hung, Yun-Man Hsu, Chia-Ming Chang, Tzung-Wu Chen, Chun-Chung Chen |
| Keywords | visual inspection; damage detection; damage quantification; deep learning |
| Abstract | Visual inspection is commonly adopted for building operation, maintenance, and safety. The durability and defects of components or materials in buildings can be quickly assessed through visual inspection. However, implementations of visual inspection are substantially time-consuming, labor-intensive, and error-prone because useful auxiliary tools that can instantly highlight defects or damage locations from images are not available. Therefore, an advanced building inspection framework is developed and implemented with route planning, realtime and detailed damage recognition, and damage rating in this study. The inspection route sketching is first exploited to provide an efficient plan with significantly reduced disruption. Then, Scaled-YOLOv4 and SOLOv2 models are considered in this study to detect defects even in a large-scale field quickly and acquire pixel-level damage recognition for more precise quantification, respectively. Finally, damage levels of components are rated following the importance and numbers per unit area of the detected defects. This entire framework is also implemented and verified by the hallway of an elementary school to detect and quantify surface damage of concrete components. As seen in the results, the conventional building inspection is significantly improved by the aid of the proposed framework in terms of damage localization, damage quantification, and inspection efficiency. |
| Title | Linear Static Analysis with Graph Neural Networks |
|---|---|
| Author | Yuan-Tung Chou, Kuang-Yao Li, Po-Chih Kuo, Wei-Tze Chang, Yin-Nan Huang , Chuin-Shan Chen |
| Keywords | |
| Abstract | Structural design is an iterative process for optimum selection, which relies on structural analysis results and experience from structural engineers. Since iterative structural analysis is a necessary for getting a good design, accelerating structural analysis is an important task. In this work, we adopt deep learning approaches as a surrogate model for linear static analysis, which provides fast and accurate structural response prediction. Based on the similarity between the structure’s topology and graph data structure, we represent structures as graphs and leverage graph neural networks (GNNs) to learn the relationship between given external forces and corresponding structural responses. The GNN model is trained with random-generated structures, including random story number, span number, beam-column length, and value of external forces. The results show that the GNN model has good performance on displacement and force predictions and excellent generalizability on unseen, taller structures. In addition, it is shown that based on the analysis of feature importance, the GNN model can extract important physical attributes from the input features. |
| Title | DATA ANOMALIES DETECTION AND CLASSIFICATION USING MACHINE LEARNING AND STATISTIC INFORMATION |
|---|---|
| Author | Tian-Xun Lin, Shieh-Kung Huang, Jau-Yu Chou |
| Keywords | data anomaly, machine learning, pattern recognition network, GoogLeNet |
| Abstract | Structural health monitoring (SHM) and structural integrity management (SIM) are emerging recently. To continuously track the condition and constantly detect early deterioration of the infrastructure, huge amounts of data are produced and abnormal measurement is inevitable. The corrupted data can produce a lot of problems and, generally, they are examined and classified by humans. In this study, the detection and classification are replaced by the techniques of machine learning (ML) and improved by using statistic information. The neural networks based on 1-dimensional and 2-dimensional data are studied via a field dataset collected from a long-span cable-stayed bridge. Therefore, a shallow network, called pattern recognition network, is selected to use 1-dimensional data as an input and a deep network, GoogLeNet, is selected to use 2-dimensional data. The results show that both models can detect and classify the data anomalies and the usage depends on the assigned application and the trade-off between computation and performance. |
| Title | Application of Convolution Neural Network and Neural Network Entropy Algorithm for Structural Health Monitoring |
|---|---|
| Author | Tzu-Kang Lin, Yi-Ting Lin, Kai-Wei Kuo |
| Keywords | Structural Health Monitoring, NNetEn, Convolution Neural Network |
| Abstract | This study combines Neural Network Entropy (NNetEn) and Convolutional Neural Network (CNN) to develop a practical structural health monitoring system. In order to verify the feasibility of the system, the failure experiment of a seven-story steel frame has been carried out with a numerical model of the same structural characteristics as the steel frame. The state space method is used to simulate the sixteen failure modes on the steel frame, where the acceleration signals of each floor at the time of failure are analyzed by neural network entropy. An entropy database is established based on the model to train the neural network model. To avoid the misjudgment and automatic interpretation of human factors, this study uses the visualized heatmap to quantify the change of entropy value, and the convolutional neural network analysis is selected for image processing. By converting the entropy value into image data, not only the number of parameters in the model can be reduced, but its operation speed can be improved. During the training process, the neural network model extracts and learns the damage features in the entropy value. After the training is completed, the model can allocate the damage area of the structure by identifying the damage features of the input data. Finally, through the verification of 16 failure cases simulated on the seven-story steel frame of the National Center for Research on Earthquake Engineering (NCREE), the performance of the proposed SHM system is evaluated by both numerical simulation and experimental verification with confusion matrix. The SHM system proposed in this study combines the emerging entropy analysis method with a neural network. The test results of the final verification have an accuracy rate of 93.13%. |
| Title | Improved Acceleration Tracking Performance of Seismic Simulators using Supervised Deep Learning |
|---|---|
| Author | Kui-Xing Lai and Pei-Ching Chen |
| Keywords | Seismic shake table; acceleration control, three-variable control; deep learning, long short-term memory neural network |
| Abstract | Seismic shake table testing has been widely used for various structural systems such as steel structures, reinforced-concrete structures, energy-dissipated and base-isolated buildings, and nonstructural components etc. Therefore, accurate replication of shake table acceleration is particularly important to these tests. In this study, supervised deep learning approach is applied as an alternative for seismic shake table control. The Long Short-Term Memory (LSTM) neural network is built for training the controller to improve acceleration performance of the shake table. A large-scale servo-hydraulic uniaxial shake table is adopted. A steel specimen is designed and fabricated for performing a large number of shake table tests. Then, the shake table testing data are used to train a feedforward controller using LSTM which is implemented close to an existing Three-Variable Control (TVC) loop. The validating experimental results prove that the acceleration tracking performance is improved compared with conventional TVC. The control-structure interaction is also suppressed. The experimental results demonstrate the proposed control scheme reduces the acceleration tracking error effectively compared with conventional TVC control. The research results also show great potential for deep learning application to seismic shake table control in the future. Keywords: Constitutive model, anisotropy, shear-slip and re-contact, mesh-sensitivity, non-proportional loading, concrete, finite element |
| Title | Synthetic Power Distribution Network Construction Based on Deep Learning Algorithm |
|---|---|
| Author | Yue-Hung Lin and Chi-Ying Lin |
| Keywords | synthetic power distribution network, deep learning, object detection, geo-positioning |
| Abstract | Global warming has caused high energy consumption and an increasing scale of disasters, which make people draw more attention to public asset management to reduce energy consumption and predict losses caused by disasters. Based on a deep learning based object detection approach, this study develops a synthetic power distribution network that can serve as an alternative to the real power distribution net-work and be used to analyze its reliability. This research uses the street view images to detect utility poles and conduct geo-positioning to locate utility poles on the map. For object detection, the Mask R-CNN and YOLOv4 are trained with controlled du-ration, and then the accuracy of the two models is compared to determine which method is suitable for this research. Second, the model’s hyperparameters are adjusted and compared to determine the best model setting for the object detection task in this study. Then the selected model is used to perform the object detection task for each street in the research region. Two sorting methods, namely, the latitude and longitude sorting method and the shortest path sorting method, are proposed to sort the poles for pole geopositioning and supplementation. With two sorting methods, pole geopositioning is conducted based on two approaches: The first is the Markov random field (MRF) approach, and the second is the line of bearing (LOB) with density-based spatial clustering of application with noise (DBSCAN). After determining the detected pole location, pole supplementation is conducted to ensure the maximum allowable distance between poles. Third, four sets of results are obtained by merging all streets and removing duplicate poles by means of distance iteration. Finally, four results are compared with the coordinates of real utility poles. The most suitable method for study region is selected to establish the synthetic power distribution network using the minimum spanning tree (MST). In the future, this model can be improved to make it more in line with the real power distribution network, and the synthetic power distribution network can be used for power grid reliability analysis, public asset management, disaster analysis, power demand-supply analysis, etc. |