PSI - Issue 62

6

Masciotta et al./ Structural Integrity Procedia 00 (2024) 000 – 000

Maria Giovanna Masciotta et al. / Procedia Structural Integrity 62 (2024) 932–939

937

Figure 3. First four numerical mode shapes of the Z24 bridge after the calibration process.

5. Data-driven optimal sensor placement Making use of the same amount of benchmark experimental data from the Z24 bridge, Masciotta et al. (2023) demonstrated that by applying a modal-based multi-objective optimization it is possible to assess the performance of different OSP algorithms and identify, out of many sub-optimal candidate solutions, a sensor configuration optimized over different structural scenarios, thus ensuring a cost-efficient monitoring of the infrastructure in the long run. Following a fully data-driven OSP approach, six known heuristic algorithms were tested: (i) Effective Independence (EfI) (Kammer, 1991); (ii) Eigenvector Component Product (ECP) (Larson et al. 1994); (iii) Mode Shape Summation Plot (MSSP) (DeClerck and Avitabile, 1996); (iv) Average Drive Point Residue (ADPR) (Chung and Moore, 1993); (v) Weighted Average Drive Point Residue (WDPR) (Chung and Moore, 1993); (vi) QR Decomposition (QRD) (Schedlinski and Link, 1996). After reducing the dimensionality of the problem from 108 to 5 DOFs, namely to the least acceptable number of degrees of freedom to ensure the identification of the main experimental vibration modes of the bridge, each algorithm allowed evaluating the contribution of each candidate sensor to the mode identifiability of the system and obtaining a candidate sensor configuration according to a specific metric. It is interesting to note that many measurement points were commonly identified as best by all the six heuristics. Details concerning the algorithm formulations and the performance evaluation process can be found in Masciotta et al. (2023) and are not repeated in the present work. For the purpose of this study, only the two data-driven sensor configurations enabling to retain the highest and most accurate amount of modal information about the bridge over different experimental scenarios are considered: the configuration obtained from the Effective Independence (EfI) method, hereafter referred as OSP 1, and the configuration resulting from the QR Decomposition (QRD) method, henceforth referred as OSP 2 (Figure 4).

OSP 1 (EfI) Ch: 19x-19z-8z-25z-33x

OSP 2 (QRD) Ch: 19z-27z-22x-8z-21z

Figure 4. Optimal sensor configurations for the Z24 bridge estimated through a fully data-driven approach.