Addressing practicalities in multivariate nonlinear regression for mitigating environmental and operational variations

A significant problem associated with the implementation of Vibration-Based Structural Health Monitoring (VSHM) systems originates from detrimental effects caused by Environmental and Operational Variations (EOVs). The EOVs cause observations same structural condition to behave in different manners. As such, this leads issues when defining a robust baseline state as well making discrimination between undamaged damaged more challenging. In order address these challenges, multivariate nonlinear regression is implemented account for EOVs. Damage Sensitive Features (DSFs) are extracted acceleration data then regressed based on environmental operational parameters. New features effectively normalised finding difference measured predicted values. This process removes influence an explicit manner, allowing reliable damage detection. While benefit application methodologies has already been demonstrated past, work addresses number practicalities VSHM real systems. analysis investigates selection DSF investigated alongside another into how detection behaves under varying amounts input information. Furthermore, method proposed understand large outliers training data.