201237233 六、發明說明： 【發明所屬之技術領域】 本發明係與損壞躺方式侧，並且尤其與針對賴物的損壞判 斷方式相關。 【先前技術】 近年來因祕震發生爾，造成人員及職等各方_損失，導 致大眾對於土木結構防震的要求日漸提高。除了更為堅鼠防震 係數更同的建材外’對於賴物土木結構的損壞亦應有事先偵、測、預 警的機制，以降低災害造成的損失。 在預3機制方面’傳統的做法係利用快速傅利葉轉換進行土木分 析，以判斷建築物的結構是否正常。但是，在以快速麵轉換所得 的頻譜中，低頻處的能量密度會被稀釋，而在高頻處的能量密度會被 放大’進而_输軸咖糊。再者，峡細轉換的基 底是内定的，#峨與錄絲正辦，錄散絲個分量而無法確 切表現出訊號特性。 倾刀析係採⑽利葉轉換，將振波分解成無 窮個正弦或/及餘弦函數的組合，再由這些正弦或/及餘弦函數的無窮 級數來計減構振動的自然解與反應頻譜，由其結絲判斷此振波 的特性。“，财法_餘紐（1職如賴（咖麵ry)的 結構振紐象，且其計算所得鮮从鱗·_自朗率。實際 上’於檢測賴物_過財所得之振動訊號往往既非線性亦非駐 201237233 態。如此一來，上述振波分析結果極可能無法反應出結構系統的真實 行為。貫際上當建物結構受到強迫振動（f〇rced vibrati〇n)時，結構 振動並非不隨時間改變的常數’而是會隨時間不斷改變的即時頻率 (instantaneous frequency) ° 【發明内容】 為解決上述問題’本發明提出用以—種用以判斷橋樑狀態的方法 及裝置。根縣發明之觸方法和判斷裝置録驗由經驗模態分解 法（empirical m〇de decomposition, EMD)所產生的内部模態函數 (intrinsic mQde function，IMF)做為峨依據。經驗模態分解法係 利用資料變化的内部時間尺度做為能量的直接解析法，將原來的訊號 資料展開成多_部模態函數。由於該等函數可為非線性(随 或非駐態(_tatiQnan〇，轉析法對練峨的躲有較佳的解 析’可以相當程度的避Μ合理的情形產生，得顺能真實反應橋標 狀態的分析結果。 根據本發明之-具體實施例為-種橋樑狀態判斷方法包含下列 步驟：⑷取得與-橋_社—織⑹針對職動訊號進行 -經驗模S分解程序，以制至少-内部模態函數，並由該至少一内 部模態函數中選出至少-目標内部模態函數；以及⑹根據該目標内部 模態函數判斷該橋樑之一狀態。 根據本發明之另-具體實施例為-橋樑狀態判斷裝置，其中包含 一收集模組…經驗藝分賴組及-酬模組。雜細組係用以 201237233 取得與_橋_關之-振動訊號。該經驗織分賴組制以針對該 振動訊號進行-經贼態分解程序，以得到至少—内部模態函數，並 由該至少-内部模態函數中選出至少—目標内部模態函數。該判斷模 組則制錄據該目標内部模態函數判斷該橋樑是否存在—損壞狀 況。 根據本發明之靖找和觸裝£可被設計為完全自動化運作， 達到即時反應的效果。監測橋樑毀損程度並树評估橋樑毀損狀態， 可在橋樑外部未有明顯裂痕或損㈣能事先掌握危魏息，對於用路 人或工程單位皆有極大的好處。此外，根據本發明之判斷方法和判斷 裳置毋須配合侵人式_壞性監測儀器，因此在量測時不會損傷橋襟 結構。 ' 關於本發明的優點與精神可以藉由以下發卿述及所關式得到 進一步的瞭解。 【實施方式】 。月參閱圖’圖-為根據本發明之—具體實施例中的橋襟狀態判 斷方法流㈣。步驟S12首先被執行，以取得與—橋樑侧之—振動 魏。舉例而言’―個或多個振動侧器可以被安裝在受測橋襟的構 棱上’收集對應於橋墩之振動的訊號。實務上，橋墩可能同時發生多 個方向的觸，上述_料被設置為_大致平行於橋樑下方之河 川水流方向的振動’但不以此為限。若將根據本發明之判斷方法應用 於其他種類建物時，可根據與機相關的振動方式決定如何安裝 201237233 偵測器。 接著，步驟S14為針對步驟S12所得之該振動訊號進行一經驗模 悲分解(empirical mode decomposition，EMD)程序，以得到至少—個 内部模態函數(1血丨_ — function，iMF)。實務上，經驗模態 分解法係糊龍變化_料狀度做為能量的直接解析法，將原 來的訊號龍賴部模態函數4明摘說，織模態分解 是根據原始峨的極值包絡線與均值包絡線，經過多次的篩選過程， 來獲得内部鄕錄。由於辑賴程是以極值包絡線與均值包絡線 去除原始訊號中各時間尺度的能量差異，強迫其形成零均值包絡線與 局部對稱’故所得之各内部模態函數中仍包涵各_尺度的振動模 態。換言之’原始振動訊號中所包涵的各種時間尺度振動模態可能因 經驗模態分解的訊號分解而分散於各個内部模態函數中。 於實際應用中，上述經驗模態分解程序可採用一自動執行模式、 -單間斷性準則（single intermitteney emedQn)模式或—雙間斷 性準則（double intermittency critedQn)模式。此處所謂自動執行 模式係指於經驗模態分解程序中未採用任何間斷性準則的方式。 相對地，單間斷性準則係指在於經驗模態分解程序中採用間斷性 準則，並且爾先蚊-傭繼ηι，絲兩極朗所含括的最大點 數限制。藉由此限制值ηι的制定，於間斷性準則執行過程中只有兩極 值間點數小於ηι之訊號被包括並分解成内部模態函數，兩極值間點數 大於m的訊號將被加以保留。相較於自動執行模式單間斷性準則模201237233 VI. Description of the Invention: [Technical Field to Which the Invention Is Ascribed] The present invention relates to the side of the damage lying, and in particular to the manner of determining damage to the object. [Prior Art] In recent years, due to the occurrence of tremors, the loss of personnel and grades has caused the public to increase the requirements for earthquake resistance of civil structures. In addition to the building materials with more shock-proof coefficient of the rat, there should be a mechanism for detecting, measuring and pre-warning the damage to the soil structure to reduce the damage caused by the disaster. In the pre-3 mechanism, the traditional practice is to use the fast Fourier transform for civil analysis to determine whether the structure of the building is normal. However, in the spectrum obtained by fast surface conversion, the energy density at the low frequency is diluted, and the energy density at the high frequency is amplified. Furthermore, the base of the fine-grain conversion is internal, and the #峨 and the recorded silk are being processed, and the weight of the silk is recorded, and the signal characteristics cannot be confirmed. The knives are decomposed into a combination of infinite sine or / and cosine functions, and the natural solution and response spectrum of the constitutive vibration are calculated from the infinite series of sine or / and cosine functions. The characteristics of this vibration wave are judged by the knot. ", the financial law _ Yu New (1 job such as Lai (Japanese noodles ry) structure of the neon image, and its calculations are fresh from the scales _ self-language rate. In fact, 'detecting the _ _ _ _ _ _ _ _ It is often neither nonlinear nor resident in 201237233. As a result, the above-mentioned vibration analysis results may not reflect the true behavior of the structural system. When the structural structure is subjected to forced vibration (f〇rced vibrati〇n), structural vibration It is not a constant that does not change with time, but an instantaneous frequency that changes with time. [Invention] In order to solve the above problem, the present invention proposes a method and apparatus for judging the state of a bridge. The method of inductive touch and the judgment device of the county invention are based on the internal modal function (IMF) generated by the empirical mode decomposition (EMD). The empirical mode decomposition method is used as the basis. Using the internal time scale of data change as a direct analytical method of energy, the original signal data is expanded into multiple modal functions. Since these functions can be nonlinear (with or without The resident state (_tatiQnan〇, the analysis method has a better analysis of the practice of hiding) can be reasonably avoided to avoid a reasonable situation, and the results of the analysis of the state of the true response bridge can be obtained. According to the present invention - the specific implementation For example, the method for judging the state of the bridge includes the following steps: (4) obtaining the - bridge_social-week (6) performing an empirical mode S decomposition procedure for the at least one internal mode function, and by the at least one internal mode Selecting at least a target internal modal function from the state function; and (6) determining a state of the bridge according to the target internal modal function. According to another embodiment of the present invention, the bridge state determining device includes a collecting module ...experience art is divided into groups and reward modules. The sub-group is used for 201237233 to obtain the _ bridge_off-vibration signal. The experience is based on the vibration signal--the thief state decomposition program. Obtaining at least an internal modal function, and selecting at least the target internal modal function from the at least-internal modal function. The determining module determines whether the bridge is determined according to the target internal modal function In the case of damage - the look and touch of the present invention can be designed to be fully automated and achieve immediate response. Monitoring the damage of the bridge and assessing the damage state of the bridge, there is no visible crack or damage outside the bridge. (4) Being able to grasp the threat of pre-existing in advance has great advantages for the use of passers-by or engineering units. In addition, according to the judgment method of the present invention and judging that the skirt is not required to cooperate with the invasive _ badness monitoring instrument, it will not be measured during the measurement. Damaged bridge structure. The advantages and spirit of the present invention can be further understood by the following description. [Embodiment] [Month] Referring to the drawings, the drawings are in accordance with the present invention. The bridge state judgment method flow (4). Step S12 is first executed to obtain the vibration of the bridge side. For example, one or more vibrators can be mounted on the ridge of the bridge under test to collect signals corresponding to the vibration of the pier. In practice, the pier may have multiple directions of contact at the same time, and the above-mentioned material is set to _ substantially parallel to the vibration of the river flow direction below the bridge', but not limited thereto. If the judging method according to the present invention is applied to other types of buildings, it is possible to determine how to install the 201237233 detector according to the vibration mode associated with the machine. Next, step S14 performs an empirical mode decomposition (EMD) procedure for the vibration signal obtained in step S12 to obtain at least one internal modal function (1 blood __ function, iMF). In practice, the empirical mode decomposition method is the direct analysis method of energy as the direct analytical method of energy. The original signal is based on the modal function 4, and the modal decomposition is based on the extreme value of the original 峨. The envelope and the mean envelope are subjected to multiple screening processes to obtain internal records. Because the extreme range envelope and the mean envelope remove the energy difference of each time scale in the original signal, forcing it to form a zero-mean envelope and local symmetry, the internal modal functions obtained by the modal are still included. Vibration mode. In other words, the various time-scale vibration modes implied in the original vibration signal may be dispersed in the internal modal functions due to the signal decomposition of the empirical mode decomposition. In practical applications, the above empirical modal decomposition procedure may adopt an automatic execution mode, a single intermitteney emedQn mode or a double intermittency crited Qn mode. The term "automatic execution mode" as used herein refers to the manner in which no discontinuity criteria are employed in the empirical mode decomposition procedure. In contrast, the single-intermittent criterion refers to the use of the discontinuity criterion in the empirical mode decomposition procedure, and the maximum number of points covered by the first mosquito-servant followed by ηι, silk bipolar. By the limitation of the value ηι, in the execution of the discontinuity criterion, only the signal with the number of points between the two poles being smaller than ηι is included and decomposed into an internal modal function, and the signal with the number of points greater than m between the two extreme values will be retained. Single discontinuity criterion model
6 S 201237233 摘優點在於可有效將分_所_峨_雜鱗取出來，以 德主訊號，使其⑽物。觀編合(她心㈣及主 振模態之頻率損失的狀況可以被改善。所謂模態混合乃指在單一内部 模態函數中同時包含兩個或兩細上不__尺度；辦間尺度的 疋義為連_之_鋼差值。模態混合_仙部模態函數產 生許多額外的、假像的變化，影響彳__綱分析結果。 另方面，雙間斷性準則模式係指在於經驗模態分解程序中使用 兩個間斷性相限舰ni' n2，保留錄模態鮮振雜間的完整頻 谱。相較於單間斷性準職式，若_斷鮮職式，後續以希 爾伯特-黃轉換(HUbert-Huang㈣―，_產生祖頻譜時不需 採用試誤法’並可聽得知、保留主振觀之_模態函數的位置， 故僅需針對該内部模態函數進行隨頻譜解析，獲得臆頻譜與邊際 頻譜圖進行·。藉此’經驗模態分解及頻譜解析之應用可變得 更正確、完整且有效率。 如圖-所示，步驟S14亦包含由該至少一内部模態函數中選出至 少一個目標内部模態函數。舉例而言，步驟S14可包含計算—零交越 率（zero-crossing rate)的子步驟，並根據該零交越率選出對應於_ 目標頻段之目標内部模態函數。由經驗法則可知，橋樑的振動頻率通 ㊉係洛在卜4赫兹之間。因此在實務上’上述目標頻段可被定為η 赫茲，但不以此為限。 接著，步驟S16為根據該目標内部模態函數判斷該橋樑之狀熊 201237233 例如疋否存在損壞狀況。圖二係用以表示步^ ,. ' 6的一種詳細實施範 於此範例中，步驟S16包含三個子步驟。首先，步驟S福為針 一目標内部模態函數進行-希叫黃轉換，以產生一希爾 一物 康°亥希爾伯特-黃轉換頻譜產生 邊際頻譜。接著，步驟S161c為 .f 4邊際觸靖雜樑是否存 在才貝壞狀況(例如結構的損壞或弱化）。 处旦步驟S1_職生__代麵是振__各頻率的 7累積，累魏量越高越可能㈣尖峰值。於實際制中，量測對 象的可能振_率區間中有可能㈣平峰現象、單峰縣或雙峰現 象。出現峨⑷所示之平峰現象絲於即時_盪_能量分 又均勾’輸量未針於高頻或低頻。出現如圖三⑻所示之單蜂 現象則表示於即時頻率«區間内累積能量集中於某個特定頻率值。 此太峰值所代表的頻率可由酣頻譜圖中即時頻率震Μ内能量集中 咖置大致如。物m祕即時頻率 震堡區間内累積“集巾於關高頻上限與低頻下限。雙尖峰值所代 率亦可由HHT頻譜圖中即時頻率震盈區内能量集中的位置預 先確認。 〜錄行步驟S161C時，可首先確認量測對象的即時頻率 盪品1及峰值位置後，再藉由圖形交叉比對的方式來分析高頻與低 二的此里1化及頻率偏移現象。當高頻產生向低頻偏移或產生累積能 降低”擴散時，代表量測對象結構特性在強迫載重作用下增加非線 201237233 性行為含量。當低頻產生向左偏移或產生累積能量變化 τ彳'表董測 對象結構特性在自然振動的情況下已顯現結構勁度的改 — 夂—般而 言，結構體由初期損傷（目視可見損傷或内部預力損失）發展至勁声弱 化的變化歷程，表現在邊際頻譜上則會出現如圖四（Α)或圖四（β)所繪 示之歷程，由高頻能量擴散或降低轉變為高頻偏移，再轉變為低頻率 能量擴散’而後再出現主鮮偏移。因此，根據上述準脚可藉以圖 形交叉比對方式瞭解受測橋樑的狀況。 於實際制中，受測橋樑在各種不同實驗狀況下所對應的邊際頻 譜都可被預先產生，做為後續參考、比對的依據。舉例而言，實:者 可變化行經橋樑之雜絲_射，域㈣同荷麵量所得的分 析圖形進行比對’觀察高頻與低頻的能量擴散與頻率偏移現象是否: 在。再例如，實驗者亦可藉由量測完整與損傷兩種不同跨段所得的: 析圖形進行比對，觀察高頻與低頻的能量擴散與鮮偏移現象是^ 在。或者，實驗者亦可取相同跨段、相同移動振動源在不同量測時^ 點所得卿進概對，觀察高軸低_能錢散麵率偏移現 象是否存在。 綜上所述，判斷橋樑是否存在損壞狀况時，可根據以下幾個考量 點做決t邊_譜之-高縣量是娜域擴散、—高頻尖峰值偏 移程度一低雜«钟加或擴散，錢—域較否祕。此外， 事則預先量測的邊際頻譜可做為參考邊 娜料。藉由比對後續所測得 的邊際頻f#及參考邊關譜之差異，亦 j斷橋樑是否存在損壞狀況。 201237233 於本發明之其他實施例中，在圖一的步驟S14之後可加入幾種不 同的判斷機制，用以決定是否步驟S14中經驗模態分解程序所得到的 結果是否理想，以判斷是否須調整經驗模態分解程序的參數，並重新 執行步驟S14。以下請參考圖五〜圖十及其相關說明。 於圖五所繪示的實施例中’於步驟S14和步驟S16之間進一步包 &步驟S21A和步驟S21B。步驟S21A為判斷該目標内部模態函數是否 存在模態混合狀況。若判斷結果為否，步驟S16將繼續被執行。相 對地，若判斷結果為是，步驟S21B將被執行，以修改經驗模態分解程 序所採用的參數，接著再回到步驟S14。實務上，目標内部模態函數是 否存在模態混合狀況可利用正交矩陣(orth〇g〇nal _Ηχ)運算來檢 測右其中出現模態混合狀況的區間範圍過大，則步驟S21A的判斷結 果將為是。 τ的經驗杈態分解程序係採用 於貫際應用中，若步驟 行模式進行訊號分解，航含絲模態之_鄕函數的位置無法被 預先得知；配合全部内部模態函數之紐頻譜圖相互比對，可以顧 人、用之内部极態函數分量。另一方面，若步驟犯中的經驗模態分 解心序係_單_性相模式（_值為n，）或雙騎性準則模式 (限制值為[ηι，⑹進行訊號分解，則依下列步驟可預知欲選用之内部 模態函數分她置：⑴應_剛_罐公式設定n，或 m值；（2)記錄η，或ηι值執行次數P值；⑶訊號分解完成後可獲得一 系列内部模__，選取其中編號為州的分量，即為欲選用 201237233 的内部模態函數分量。比對編號分別為州和糾的内部模態函數 分量即可圈選出可能為前述出現模態混合的區間範圍。 於圖六所繪示的實施例中，於步驟S14和步驟則之間進一步包 含步驟S22A和步驟S22B。步驟& η 4* 驟挪為判斷該目標内部模態函數中之 一可判讀歷咖是wn值。___ 將繼續被執行。相對地，若判斷結果為是，步驟S22B將被執行，心 改該經驗模態分解程序所採用的參數，接著再重新執行步驟S14。此可 判讀歷時區間亦可根據上述比對編號分別為州*州的内部模態 函數分量的結果來判斷。 於圖七_示的實施财，圖二的步驟smA後進—步包含步驟 S23A和步驟S23B。步驟S23A Λ妒媸_ μ ^ θ 3Α為根據贈頻譜判斷目標内部模態函數 :否存在模歧合狀況。若其靖結果步驟s咖將繼續被執 二’步獅將被執行’崎改該經驗模 ^序所^用的參數，接著再重新回到步驟S14。實務上，步驟 中產生的騰頻譜可用來與内部模態函數圖形 模態混區間範圍。 以厘，月 於圖八崎示的實施财，圖二的步驟5職後進一步包含步驟 。步驟s。步驟魏為根據紐頻譜判斷目標内部模態函數 之可匈讀歷時區間是否小於一第一門捏值。 被執行。相對地，若判斷結果為是，步獅繼 4改趣驗模態分解程序所採用的參數，接著再重新明步驟 11 201237233 則。同樣地，步驟S腦中纽的臆頻譜可_____ 形相互比m清可觸麟_位置。或者，可觸歷時區間位 置亦可利料錄(副；mean s_>e)運算來檢測。 於圖九所IT示的實施例t ’圖二的步驟_Α後進—步包含步驟 S25A〜步驟S25C。步驟取為辨識祖頻譜中之一強迫振動頻率區間6 S 201237233 The advantage is that it can effectively take out the _ _ _ _ _ scales, with the main signal, make (10). The situation of the frequency loss of her mind (four) and the main mode can be improved. The so-called modal mixing refers to the inclusion of two or two fine __ scales in a single internal modal function; The derogatory meaning is the _ _ steel difference. The modal mixing _ sin modal function produces many additional, imaginary changes that affect the 彳 _ _ analysis results. On the other hand, the double discontinuity criterion mode refers to In the empirical mode decomposition procedure, two intermittent phase-limited ships ni' n2 are used, which preserves the complete spectrum of the recorded mode fresh-vibration. Compared with the single-intermittent quasi-position, if the _ breaks the job, the follow-up Hilbert-Huang (four)-, _ generation of ancestral spectrum does not need to use trial and error method 'and can hear and retain the position of the main vibration mode _ modal function, so only need to target the internal mode The state function is performed along with the spectrum analysis to obtain the 臆 spectrum and the marginal spectrum map. The application of the empirical modal decomposition and spectrum analysis can be more accurate, complete and efficient. As shown in the figure, step S14 also includes Selecting at least one target interior from the at least one internal modal function For example, step S14 may include a sub-step of calculating a zero-crossing rate and selecting a target internal modal function corresponding to the _ target band based on the zero-crossing rate. It can be seen that the vibration frequency of the bridge is between ten and four Hz. Therefore, in practice, the above target frequency band can be set to η Hz, but not limited thereto. Next, step S16 is based on the target internal modality. The function judges whether the bridge bear 201237233 has a damaged condition, for example. Figure 2 is a detailed implementation of the step ^,. '6. In this example, step S16 includes three sub-steps. First, the step S is The needle-target internal modal function is performed by a yellow-to-yellow transformation to generate a marginal spectrum of the Hill-Hilbert-Huang conversion spectrum. Then, step S161c is the .f 4 marginal touch-to-beam beam. There is a bad condition (such as damage or weakening of the structure). The step S1_professional __ generation is the accumulation of 7 __ of each frequency, the higher the amount of tidal is, the more likely it is (four) the peak value. In the actual system , measuring the possible vibration of the object In the _ rate interval, there may be (4) flat peak phenomenon, single peak county or double peak phenomenon. The flat peak phenomenon shown in 峨(4) appears in the instant _ _ _ energy score and both hooks 'transmission is not at high frequency or low frequency. The single bee phenomenon shown in Figure 3 (8) indicates that the accumulated energy in the immediate frequency « interval is concentrated at a certain frequency value. The frequency represented by this too peak can be approximated by the energy concentration in the instantaneous frequency spectrum in the spectrum of the spectrum. The m-secret instant frequency in the earthquake-storage interval accumulates “the high-frequency upper limit and the low-frequency lower limit of the towel. The generation rate of the double-point peak can also be confirmed in advance by the position of the energy concentration in the real-time frequency of the HHT spectrogram. When the step S161C is performed, the instantaneous frequency 1 and the peak position of the measurement object can be confirmed first, and then the phenomenon of the high frequency and the low frequency and the frequency offset phenomenon are analyzed by the pattern cross comparison. When the high frequency is shifted to the low frequency or the cumulative energy is reduced, the structural characteristic of the measured object increases the non-linear 201237233 sexual behavior content under the forced load. When the low frequency produces a leftward shift or produces a cumulative energy change τ彳' The structural characteristics of the table object have been shown to change the structural stiffness in the case of natural vibrations—in general, the structure develops from the initial damage (visual damage or internal preload loss) to the weakening process of the weakening of the sound. In the marginal spectrum, there will be a process depicted in Figure 4 (Α) or Figure 4 (β), which is transformed from high-frequency energy diffusion or reduction to high-frequency offset, and then to low-frequency energy diffusion. The main fresh offset occurs. Therefore, according to the above-mentioned standard foot, the condition of the tested bridge can be understood by means of graphic cross-matching. In the actual system, the marginal spectrum corresponding to the tested bridge under various experimental conditions can be pre-generated. As a basis for subsequent reference and comparison. For example, the actual: can change the analysis pattern of the bridge's miscellaneous wire, the field (four) and the surface amount. 'To observe whether the energy diffusion and frequency shift phenomenon of high frequency and low frequency are: In. For example, the experimenter can also measure the difference between the complete and the damage by two different sections: analysis of the pattern, observe the high frequency and The low-frequency energy diffusion and the fresh offset phenomenon are ^. Or, the experimenter can also take the same span and the same moving vibration source at different measurement points, and observe the high-axis low _ energy surface rate Whether the offset phenomenon exists. In summary, when judging whether there is damage in the bridge, the following points can be determined according to the following considerations: the spectrum of the high county is the diffusion of the nano domain, the peak of the high frequency peak The degree is low or mixed, the money-domain is more secret. In addition, the marginal spectrum measured in advance can be used as a reference edge. By comparing the measured marginal frequency f# and reference edge spectrum The difference is also that the bridge is damaged. 201237233 In other embodiments of the present invention, several different judging mechanisms may be added after step S14 of FIG. 1 to determine whether the empirical modal decomposition procedure in step S14 is performed. The result obtained Whether it is ideal to determine whether the parameters of the empirical mode decomposition procedure have to be adjusted, and step S14 is re-executed. Please refer to FIG. 5 to FIG. 10 and related descriptions below. In the embodiment illustrated in FIG. 5, 'in step S14 and step Step S21A and step S21B are further included between S16. Step S21A is to determine whether the target internal mode function has a modal mixing condition. If the result of the determination is no, step S16 will continue to be performed. Yes, step S21B will be executed to modify the parameters used by the empirical mode decomposition procedure, and then return to step S14. In practice, whether the target internal mode function has a modal mixing condition can utilize an orthogonal matrix (orth〇g) The 〇nal_Ηχ) operation is to detect that the range of the interval in which the modal mixing condition occurs is too large, and the result of the determination in step S21A is YES. The empirical decomposition program of τ is used in the continuous application. If the signal is decomposed in the step line mode, the position of the _鄕 function of the aeronautical modal mode cannot be known in advance; with the new spectrum of all internal modal functions Mutual comparison, you can take care of people, use the internal polar function components. On the other hand, if the empirical modal decomposition of the step is _ single-phase mode (_value is n,) or the double-riding criterion mode (the limit value is [ηι, (6) for signal decomposition, then The step can predict that the internal modal function to be selected is divided into: (1) _ _ _ can formula set n, or m value; (2) record η, or ηι value execution times P value; (3) after signal decomposition is completed The internal model __ of the series is selected as the component of the state, which is the internal modal function component of 201237233. The internal modal function components of the state and the correction are respectively selected to circle the possible modalities. The range of the mixed interval. In the embodiment illustrated in Figure 6, step S22A and step S22B are further included between step S14 and the step. Step & η 4* is moved to determine the target internal modal function A readable registrar is a wn value. ___ will continue to be executed. In contrast, if the result of the determination is YES, step S22B will be executed, the parameters used by the empirical modal decomposition program are modified, and then step S14 is re-executed. This can be used to read the duration interval It can be judged according to the result that the comparison number is the internal modal function component of the state* state respectively. In the implementation of FIG. 7 , the step smA of the second step includes step S23A and step S23B. Step S23A Λ妒媸 _ μ ^ θ 3Α is the internal modal function of the target according to the spectrum of the spectrum: there is a mode of disambiguation. If the result of the step s will continue to be executed, the second step will be executed. The parameters used are then returned to step S14. In practice, the tem spectrum generated in the step can be used to blend the interval range with the internal modal function graphic modality. Step 5 of the second step further includes a step. Step s. Step Wei determines whether the target internal time function of the target internal modal function is less than a first threshold value according to the New Spectrum. The relative result is Yes, Step lion follows the parameters used in the modal decomposition program, and then re-examines step 11 201237233. Similarly, the 臆 spectrum of the step S in the brain can be compared with the m clear _____ Location. Or, The touch time interval position can also be detected by the operation record (sub; mean s_>e) operation. The step t' of the embodiment shown in Fig. 9 shows the step S25A to step S25C. To identify a forced vibration frequency interval in the ancestral spectrum
^一自難動解_。触雜_區_能量較針，頻率值通 =主頻率略低，震M化區間較不規律穩定。自織動頻率區間的 /里車乂弱解震盈變化區間較規律穩定。接著，步驟观為判斷· 項。曰中之鮮知失區間是否出現於強迫振動鮮區間或自然振動頻 率£間。此處所謂鮮損失是經驗顯分解程序可料觸現象。頻 率損失有可能造成頻譜變化的科續，也可會造成頻率降低的誤判。 若頻率損失區間岐在強迫振動辭區間或自然振動鮮區間，可能 導致邊際頻谱圖中高頻或低頻的尖峰值因累積能量的損失而消失，這 種It形的發生將會使得分析結果與實際_碎。因此，若步驟奶B _斷結果為是’步驟S25C將被執行，以修改該經驗模態分解程序所 &用的參數’接著再重新回到步驟S14。 於圖十所綠不的實施例中，圖二的步驟S161A後進-步包含步驟 a A # ^驟S26B。步驟挪為判斷上述冊τ頻譜中之頻率損失區間 2否大於一第二門檻值。若步驟S26A的判斷結果為是，表示後續也可 «出現决縣析結果的情況，因此步驟遞將被執行，以修改該經 驗私態分解程序所_的參數，接著再讀_步驟S14。^ One self-defeating _. The contact _ area _ energy is better than the needle, the frequency value is passed = the main frequency is slightly lower, and the seismic M-shaped interval is less regular and stable. The interval between the self-weaving frequency range and the weaker solution of the celestial solution is more stable. Next, the step view is a judgment item. Whether the fresh-sense interval in the sputum appears between the forced vibration interval or the natural vibration frequency. The so-called fresh loss here is a phenomenon that can be touched by an empirical decomposition program. Frequency loss may cause a series of changes in spectrum, and may also cause misjudgment of frequency reduction. If the frequency loss interval 岐 is in the forced vibration interval or the natural vibration interval, the peak value of the high frequency or low frequency in the marginal spectrum diagram may disappear due to the loss of accumulated energy. The occurrence of this It shape will make the analysis result and the actual situation. _broken. Therefore, if the step milk B_break result is yes, 'step S25C will be executed to modify the parameter used by the empirical mode decomposition program' and then return to step S14. In the embodiment shown in FIG. 10, the step S161A of FIG. 2 further includes the step a A # ^ S26B. The step is moved to determine whether the frequency loss interval 2 in the spectrum of the book τ is greater than a second threshold value. If the result of the decision in the step S26A is YES, it means that the subsequent occurrence of the result of the decision can be made, so that the step is to be executed to modify the parameter of the experienced private decomposition program, and then read _step S14.
12 S 201237233 圖五〜圖十中的各種判斷機制可以同時存 須έ兒明的是，實務上， '、ρ被.内入同個流程令執行’或者亦可於單一流程中選用其中 的幾種判斷機制。 :根據本發明的其他具體實施例卜步驟別也可以用其他不同 的方式來實現。比方,兄’步驟S16可包含下列子步驟：⑴針對該至少 "耗内指態函數進仃一快速傅利葉轉換以產生一傅利葉頻譜；⑵ ’謂〆傅利魏4之-極值’·以及⑶根據雜值繼雜樑是否存在 知壞狀况。貫務上，事前測試所得之各種已損壞/未損職況的極值可 做為比對的參考基準。又例如，步驟灿亦可包含下列子步驟：⑴針 對該至少-目標内部模態函數進行以產生一·頻譜；以及⑵根 據該酣頻譜判斷橋樑是否存在損壞狀況。也就是說直接比較現有 的頻譜和—參考頻譜也可用來判斷橋襟是否存在損壞狀況。 再者’步驟S16亦可能包含下列子步驟：⑴判斷該至少一目標内 部模態函數之-零交越率與一能量變化；以及⑵根據該零交越顿能 量變化判斷橋樑是飾在損壞歧。此外，步驟Sl6亦可能包含下列 子步驟：(1)判斷該至少一目標内部模態函數的—般零交越率 (ge麗lized獅-c聰ing rate);以及⑵根據—般零交越率判斷 橋樑是否存在損壞狀況。易言之，根據由步驟S14所產生之目標内部 模態函數触所得料種分析絲皆可做為_橋樑是飾在損壞狀 況的依據，不以圖二所示之範例為限。 之—具體實施例中的 請參閱圖十一(A)，圖十一（A)為根據本發明 13 201237233 橋樑狀態判斷裝置方塊圖。橋樑狀態判斷裝置3q包含收集模組犯、經 驗模態分解模組34和觸模組36。_組32係用峰得與一橋標 相關之—振動輯。經驗娜祕馳34伽崎_減訊號進行 一經驗模態分解程序，以得到至少—内部模態函數並由該至少一内 部模態函數中選出至少一目標賴態函數。判斷模組％係用以根據 該目標内部模態函數判斷該橋樑是否存在—損壞狀况。 。。如圖十—⑻所示，判斷模組36可包含轉換單元·、邊際頻譜產 生單元36B和判斷單元縦。轉換單元·係用以針對該至少一目標内 部模態函數進行—希爾伯特_黃轉換，以產生—希爾伯特—黃轉換頻 a邊際頻Da產生早几編係用以根據該希爾伯特—黃轉換頻譜產生— 邊際頻譜。判斷單元縦則係用以根據該邊際頻譜判斷該橋襟是否存 在該損壞狀況。 橋樑狀態_置3G ____考先前關於根據 Μ之橋樑狀態判斷方法的各流程圖及其朗，因此科贅述。如 爾明者’判斷模組36可根據該綱譜之-高頻能量是否降低 擴散、-高敝峰賴絲度、-低雛量是㈣加紐散，或是 主頻率是錢移’來觸_是否存在損職況。此外，如圖十—(( 所示L橋樑狀態觸裝置30可進一步包含警示模組％。若判斷模 36判Μ橋樑存在損壞狀況，警示模組38即可發出—警示訊息通 單位來儘速别往修復，或於橋樑兩端出入口以燈號或聲響發 告訊息。 日、 201237233 如纖，《她之撕__蝴經 齡解法所產生的内部模態函數做為判斷依據。由於該等函數可為 線性或非駐態，此解析法對原來峨的特性有較佳的解析可以2 程度的避免不合理的情形產生，得到更能真實反應橋樑狀態的分析目二 果。根據本發明之觸方法和判_置可被設計為完全自動化運作 達到即時反應的效果。此外，根據本發明之靖方法和判斷裝置母項 配合侵入式的破雜制健’因此柿料不會損傷橋麵構。、 藉由以上健具體實施例之詳述，鱗望能更加清辦述本發明 之特徵與精神，而並非以上述所揭露的較佳具體實施例來對本發明之 料加以限制。相反地，其目的是希望能涵蓋各種改變及具相紐的 安排於本發明所欲申請之專利範圍的範疇内。 【圖式簡單說明】 圖一為根據本發明之-具體實施例中之橋樑狀態判斷方法流程 圖； 圖二揭露根據本發明之橋樑狀態判斷方法中的判斷步驟之詳細實 施範例圖； 圖三(A)〜圖三⑹係用以表示振動頻率區間可能出現的平峰現 象、單峰現象和雙峰現象圖； 圖四(八)和圖凹⑻係用以表示結構體由初期損傷發展至勁度弱化 的邊際頻譜變化歷程圖； 15 201237233 圖五〜圖十係肋表錄據本發明之橋雛態满方法進-步包 含評估經驗模態分解程序結果之步驟的實施範_; 圖十（A)〜®十-(Q為根據本發明之—具體實施例巾之橋襟狀 態判斷裝置方塊圖。 【主要元件符號說明】 S12〜S16 :流程步驟 S21A-S21B :流程步驟 S23A~S23B :流程步驟 S25A~S25C :流程步驟 30 :橋樑狀態判斷裝置 34 :經驗模態分解模組 36A :轉換單元 36C :判斷單元 S161A〜S161C :流程步驟 S22A~S22B :流程步驟 S24A~S24B :流程步驟 S26A~S26B :流程步驟 32 :收集模組 36 :判斷模組 36B :邊際頻譜產生單元 38 :警示模組12 S 201237233 The various judgment mechanisms in Figure 5 to Figure 10 can be stored at the same time. In practice, ', ρ is entered into the same process order' or can be selected in a single process. Kind of judgment mechanism. The steps according to other embodiments of the present invention may also be implemented in other different ways. For example, the brother's step S16 may include the following sub-steps: (1) performing a fast Fourier transform for the at least "intrinsic finger function to generate a Fourier spectrum; (2) 'predicate 〆Fuliwei 4's - extreme value'· and (3) According to the miscellaneous value, the hybrid beam has a known bad condition. In practice, the extreme values of various damaged/unremoved conditions obtained from prior testing can be used as a reference for comparison. For another example, the step can also include the following sub-steps: (1) performing the at least-target internal modal function to generate a spectrum; and (2) determining whether the bridge is damaged according to the 酣 spectrum. That is to say, directly comparing the existing spectrum and the reference spectrum can also be used to determine whether the bridge is damaged. Furthermore, 'Step S16 may also include the following sub-steps: (1) determining the zero-crossover rate and an energy change of the at least one target internal modal function; and (2) judging that the bridge is decorated in damage according to the zero-crossover energy change . In addition, step S16 may also include the following sub-steps: (1) determining a general zero-crossing rate of the at least one target internal modal function (ge ly lion-c ing rate); and (2) according to the general zero-crossing Rate to determine if the bridge is damaged. In other words, according to the target internal modal function generated by the step S14, the analysis of the wire can be used as a basis for the damage of the bridge, which is not limited to the example shown in FIG. For the specific embodiment, please refer to FIG. 11(A), and FIG. 11(A) is a block diagram of the bridge state judging device according to the present invention 13 201237233. The bridge state judging device 3q includes a collection module, a modal decomposition module 34, and a touch module 36. The _ group 32 series uses the peaks associated with a bridge-vibration. The experience Namichi 34 gamma_reduction signal performs an empirical mode decomposition procedure to obtain at least an internal modal function and select at least one target latitude function from the at least one internal modal function. The judgment module % is used to judge whether the bridge exists or not according to the target internal modal function. . . As shown in Fig. 10-(8), the judging module 36 may include a converting unit, a marginal spectrum generating unit 36B, and a judging unit 縦. The conversion unit is configured to perform a Hilbert-Yellow transformation on the at least one target internal modal function to generate a Hilbert-yellow conversion frequency a marginal frequency Da to generate an early sequence according to the Greek The erbert-yellow conversion spectrum produces – the marginal spectrum. The determining unit is configured to determine whether the bridge has the damage condition based on the marginal spectrum. Bridge state _ 3G ____ test the previous flow chart according to the method of judging the state of the bridge and its lang, so the section is described. Such as the Erminger's judgment module 36 can be based on the spectrum - whether the high-frequency energy reduces the diffusion, - the high peak, the low amount is (4) plus the new, or the main frequency is the money shift' Touch _ whether there is a loss of duty. In addition, as shown in FIG. 10((), the L-bridge state contact device 30 can further include the warning module %. If the mode 36 is judged to be damaged, the warning module 38 can be sent out - the warning message is transmitted to the unit as fast as possible. Don't fix it, or send a message by light or sound at the entrances and exits of the bridge. 日, 201237233 如纤, "The internal modal function generated by her tearing __ butterfly age solution is used as the basis for judgment. Because of these The function can be linear or non-resident. This analytical method has a better resolution of the original 峨 characteristics. It can avoid the unreasonable situation to a certain extent, and obtain an analysis result that can more truly reflect the state of the bridge. According to the present invention, The touch method and the judgment method can be designed to fully automate the operation to achieve the effect of the immediate reaction. In addition, according to the method of the present invention and the judging device, the parent item cooperates with the intrusive type of the broken system. Therefore, the persimmon material does not damage the bridge deck structure. The features and spirit of the present invention will be more apparent from the detailed description of the embodiments of the present invention, and the present invention is not limited by the preferred embodiments disclosed herein. Rather, it is intended to cover various modifications and alternative arrangements within the scope of the invention as claimed. FIG. 2 is a diagram showing a detailed implementation example of the judging step in the bridge state judging method according to the present invention; FIG. 3(A) to FIG. 3(6) are used to indicate the peak phenomenon that may occur in the vibration frequency interval. , unimodal phenomenon and bimodal phenomenon diagram; Figure 4 (8) and concave (8) are used to indicate the marginal spectrum change history of the structure from initial damage to weakening; 15 201237233 Figure 5 ~ Figure 10 The method of the present invention includes the implementation of the steps of evaluating the results of the empirical mode decomposition procedure. FIG. 10(A)~®T-(Q is in accordance with the present invention - the specific embodiment of the towel Block diagram of the bridge state determination device. [Description of main component symbols] S12~S16: Process steps S21A-S21B: Flow steps S23A to S23B: Flow steps S25A to S25C: Flow step 30: Bridge state determination device 34: Modal Decomposition Module 36A: Conversion Unit 36C: Judgment Units S161A to S161C: Flow Steps S22A to S22B: Flow Steps S24A to S24B: Flow Steps S26A to S26B: Flow Step 32: Collection Module 36: Judgment Module 36B: Margin Spectrum generation unit 38: warning module
16 S16 S