Multi-objective sensor placement optimization of helicopter rotor blade based on Feature Selection

This work aims to develop a Structural Health Monitoring methodology that maximizes the acquired modal response and minimizes the number of sensors in a helicopter's main rotor blade. Although this trade-off is a SHM principle, there is no methodology in literature that opposes these objectives for any structure. Firstly, a real AS350 helicopter rotor blade was experimentally tested and a numerical model was elaborated in FEM. An inverse method found the mechanical properties that fit numerical and experimental models. Then, a new methodology is proposed to address the Sensor Placement Optimization problem using the Multi-objective Lichtenberg Algorithm and Feature Selection. The Multi-objective Sensor Selection and Placement Optimization based on the Lichtenberg Algorithm (MOSSPOLA) has as one of the objectives the number of sensors and the other, one of the 7 best-known metrics in SPO: Kinetic Energy, Effective Independence, Average Driving-Point Residue, Eigenvalue Vector Product, Information Entropy, Fisher Information Matrix, and Modal Assurance Criterion. Pareto fronts and sensor configurations were generated and compared. Linear and convex families of Pareto fronts were unprecedentedly identified, showing a correlation between them. Better sensor distributions were associated with higher Hypervolume and the best metrics for each family were applied to damage identification for final comparison. The MOSSPOLA found a sensor configuration for each sensor number and metric, including one with 100% accuracy in identifying delamination considering triaxial modal displacements, minimum number of sensors, and noise for all blade sections.

Automated summary

This work aims to develop a Structural Health Monitoring methodology that maximizes the acquired modal response and minimizes the number of sensors in a helicopter's main rotor blade. A new methodology, called MOSSPOLA, was proposed to address the Sensor Placement Optimization problem using the Multi-objective Lichtenberg Algorithm and Feature Selection. The results showed the correlation between sensor distributions and better metrics, and the MOSSPOLA found a sensor configuration with 100% accuracy in identifying delamination.