Fail-safe SPO

Here are the data and codes to reproduce the results in the paper On improved fail-safe sensor distributions for a structural health monitoring system.

• A series of tests were performed on a glider wing in an environmental chamber to provide a data set suitable for this research, which is in the folder ModeShapeData.
• The deployment of 36 candidate sensors is shown in Figure SensorDeployment.jpg.

1ModeSExtraction

Codes in this folder can be used to extract features for the next step. The adopted features include the mode shapes and corresponding labels of structural states.

1. MSExtraction_ModalIdentification.m will provide features for modal identification. Two criteria are considered, including the DFIM and the DFIMADPR.
2. MSExtraction_DamageDetection.m will provide features for damage detection. Two criteria are considered, including the SSC and the SSCADPR.

2DataProcessing

1. The codes of the classical SPO, the fail-safe SPO and the fail-safe with redundancy SPO corresponding to four optimisation objectives, including the DFIM, DFIM-ADPR, SSC and SSCADPR, are in folders FIM, FIMADPR, SSC and SSCADPR seperately.
   • Two search algorithms were applied, including a deterministic algorithm–exhaustive search (ES) and a stochastic algorithm–genetic algorithm (GA), to obtain the optimal sensor layouts.
   • ExhaustiveS refers to ES and SPOGA refers to GA.
   • The file with Improved in its name should be run last in the corresponding subfolder to provide the improved fail-safe SPO results.
2. FIM-FIMADPR-Assessment and SSC-SSCADPR-Assessment contain performance evaluation codes. For the specific related questions, please refer to Section 5.3 of the paper.
3. OptimalSensorDistribution contains codes that facilitate comparisons of optimal sensor distributions. The corresponding part in the paper is Section 5.4.