各種結構破壞診斷方法中，常採用有效的振動訊號來進行，但這些方法中甚少考慮材料變異因素，而這些變異可來自許多實務上的原因，也將影響振動訊號之內涵，進而誤導診斷及監測結果。本研究提出一種破壞診斷統計模型，並以自由樑裂縫診斷例子驗證模型成效，其中模態曲率用來偵測裂縫位置，並以蒙地卡羅方法建立合適統計資料庫，再以實測模態頻率比對此統計資料庫，以獲得裂縫深度。測試例子之模態參數取自實際加工之鋼樑以及有限元素模擬分析結果，從各演算例子的結果可證明，本研究提出之結構破壞診斷統計模型，具體說明在結構系統中楊式係數及質量密度同時變異情況下，如何診斷出多裂縫自由樑之裂縫數目及其位置。

It is effective to apply damage-sensitive features from vibration response for structural damage assessment. Fewer damage identification algorithms have been taken into account the material variation. The material variation could be caused by many reasons in engineering practice, these variations may affect the features used for struc-
ture monitoring and lead to an inaccurate damage assessment. In this research the au-thors proposed a model to assess statistical structural damage of free-free beam struc-ture. The modal curvature-base feature was used to identify crack location. The statis-tical damage database was build by applying the Monte Carlo simulation with Latin hypercube sampling. The damage probability distribution among various crack depths
were estimated by mapping measured frequencies to the statistical damage database, and its statistical significance of damage level were examined by t-test. The experi-mental modal analysis data from hardware samples and simulated beams were used to demonstrate the crack assessment procedures. From the assessment results, the authors concluded the proposed algorithm was robust and capable of identifying the multiple
cracks in free-free beam with uniform mass density and stiffness variations.