Nonlinear Model Identification Method for Crane Damage Detection Without Baseline Data

Abstract

Damage in the structure may lead to significant reduction in local strength. It is of great significance to localize the damage in the early stage for integrity of the structure so as to prevent catastrophic failure. The traditional damage detection techniques is achieved through comparing the current data with the baseline data obtained from the healthy structure. However, the baseline data is not available as the structure is often subjected to operational and environmental variations that may adversely affect the measurement signals. In this paper, model identification method is adopted for crane damage detection. It only requires the data after the structure is damaged. An improved nonlinear model is proposed based on the traditional models. It is denoted as RBF-BL (RBF Network based State Dependent Bilinear) model. Different damage cases are considered in the girder of the crane. The girder is subjected to transient shock excitation. The nonlinear model is constructed based on the vibration response signals and parameter estimation is performed. The anomalous region is initially determined through the model characteristic parameter. Beamforming algorithm is then performed for precise position of the crack. Numerical simulation and experimental validation are implemented to prove the effectiveness of the proposed model for crane damage detection without baseline data.