MACHINE LEARNING APPROACH FOR SENSOR PLACEMENT OPTIMIZATION

This paper presents a novel approach that combines deep learning techniques with traditional sensor placement optimization methods for modal identification of bridge structures. The proposed methodology is developed and validated using experimental data from three bridges instrumented with more than 20 sensors each. A Neural Network architecture is designed to learn optimal sensor configurations based on Modal Assurance Criterion (MAC) values, modal shapes, and natural frequencies. Comparisons with traditional optimization methods demonstrate improvements in both computational efficiency and placement quality. The proposed machine learning framework effectively captures complex relationships between sensor positions and modal identification quality, while maintaining geometrical constraints. This study illustrates the use of machine learning as an optimization tool in sensor placement and provides a concrete background for general applications in practical structural health monitoring studies.