A Probabilistic Model-Based Framework for Damage Detection in Beam-Like Structures Considering Temperature Effects

To improve the assessment outcomes of structural health monitoring systems, it has been acknowledged that the damage diagnostics process is associated with a considerable amount of uncertainties from operational and environmental sources such as temperature. This study presents a probabilistic framework for structural health monitoring aimed at identifying common defects such as cracks and corrosion in beam-like structures under varying temperature conditions. The framework utilizes optimization methods integrated with a probabilistic approach, comprising two levels of damage identification: first, determining the optimal probability mass function for the damage location ratio, followed by the optimal conditional for the damage depth ratio. A simply supported beam divided into four segments was modeled to evaluate the effectiveness of the proposed algorithm across various damage scenarios using frequency response signals. Structural damage was simulated by manipulating the length and depth ratios of specific beam segments. The modeling procedure for defects in the simply supported beam was validated through comparison with experimental measurements, achieving a maximum difference of less than 1%. The proposed algorithm consistently yielded improved damage detail outcomes in the second level, characterized by higher probabilities and lower uncertainties for both damage location and depth ratios. Identifying cracks and corrosion with a 0.1 depth ratio at low temperatures proved challenging, but the algorithm effectively addressed such scenarios, achieving identification probabilities exceeding 90%. The analysis outcomes demonstrate the advantages of the proposed framework for structural health monitoring that accounts for temperature effects, thereby improving the fidelity of damage assessment in real-world conditions.