Ensemble Learning Approach for Differentiating Simultaneous Temperature and Strain Impacts on Apodized FBG-based Sensor

The Fiber Bragg Grating (FBG) is incorporated with proposed functional profile of apodization to illustrate the significance of average range of side lobes scale moderation in the spectral response by apodization approach for effectively determining the target measurands. The Bragg wavelength alters based on the impact of measurands that might affect the periodicity in addition to the amplitude of modulation index of the FBG. To determine the responses of the specific measurands effectively, it is exceedingly worthwhile to figure out whether the modifications in the Bragg wavelength are attributed through the responses of temperature or strain, as both target measurands influence the Bragg wavelength alteration. The utmost challenging aspect of implementing FBG is the cross-sensitivity issue amongst the temperature along with the strain responses on FBG. Several Machine Learning (ML) approaches are incorporated through Ensemble Learning based models (like-Extra Trees, Random Forest, Bagging, and AdaBoost) to resolve the drawback of cross-sensitivity of FBG sensors through multi-out regression analysis of target measurands. The predictive analysis performance of the Ensemble models with AdaBoost regressor for the differentiation of the temperature together with the strain influences is noted as significantly high as compared with the other models.