Smart Diagnosis of a Cantilever Beam Using SVM

Damage assessment (DA) in Structural Health Monitoring (SHM) is receiving a lot of attention from academics, research, and industry due to increased city traffic volumes brought on by urban growth. Using this kind of technique, structures like buildings, bridges, marine structures, etc., may keep track of their own structural integrity both during operation and over the period of their existence. Although reliability increases, the cost of system and structure maintenance and inspection decreases. This kind of structural deterioration testing is performed without actually causing any damage to the system. Before going on to the in-situ experimentation for damage assessment utilizing sensors, a laboratory experiment employing a cantilever steel beam is carried out to understand the modal parameters from the vibration data collected. This study analyzes damage in a cantilever beam using data obtained from vibrations and statistical features derived from time series data. Support Vector Machines (SVMs) are used for both classification and regression tasks. This study achieved a 90% accuracy rate for categorizing the data into damaged and undamaged types using the SVM approach. SVM may be affected by the choice of hyperparameters, such as the regularization parameter and kernel type, and can be computationally expensive for large datasets, but it is still a useful and well-liked approach in the machine-learning community.