A signal energy change-based damage localization approach for beam structures

•An effective signal processing-based damage detection method is proposed.•The method does not require the finite element model.•The method has a strong anti-noise-interference ability up to 15% noise level.•The method can identify the damage of 5% stiffness reduction.•The method avoids the error in the parameter identification process. Beams are widely used in many mechanical systems and other structures. Accurate damage localization of beam structures still suffers from some limitations. To address these issues, an effective damage localization approach, i.e. the curvature difference probability of logarithm of mean square based approach, is proposed on the basis of a new feature (i.e. the logarithm of mean square) and the curvature method. The feature is computed at every measured node with accelerations before and after damage. Then, curvature differences of the waveforms of logarithm of mean square between the undamaged and damaged status are selected as candidates for the potential damage locations. Lastly, the damage probability of each element is considered to determine the final damaged element configuration. Damage localization of a simply supported beam model with numbers of numerical and experimental damage cases are employed to verify the proposed approach. All results verify that the proposed approach can be used to locate damage well and it exhibits high-noise insusceptibility: it is still effective even if the noise level is up to 15%. Also, the proposed approach does not require a numerical model of the measured structure, all of these lay a good foundation for engineering application.