Structural damage detection using active members and neural networks

Abstract

Introduction T detection of damage in structures is a topic which has considerable interest in many fields. Detecting damage in space structures when subjected to the harsh environment of space could allow for the repair of the structure to occur before the damage threatens the mission objectives. Offshore oil platforms constantly have problems with potential member failure in the corrosive sea environment. Buildings and bridges, where structural failure proves catastrophic, would also benefit from a reliable method of detecting and pinpointing structural damage. In the past, many methods for detecting damage in structures have relied on finite element model refinement methods. Hajela and Soeiro determined the damage present in a structure by updating the finite element model to match the static and dynamic characteristics of the damaged structure. Their method was an outgrowth of those presented in Refs. 2 and 3 where undamaged members' section properties changed during the model update process, thus smearing the damage over a wide portion of the structure and making specific damage difficult to locate. Hajela and Soeiro also extended their damage detection techniques to composite structures where a gradient-based optimization scheme was used to update the finite element model. Other methods of detecting damage in structures rely strictly on measured data. Cawley and Adams used only natural frequency data, Pandey et al. used mode shape curvature data, and Swamidas and Chen used strain, displacement, and acceleration data to monitor and detect changes and damages in various structures. These methods require comparing measurements of the structure in the nominal (undamaged) state with those at a later date where some damage is potentially present in the structure. These methods have the drawback that they can only identify that the structure has changed; they cannot identify the location and extent of the damage. Neural networks have the unique ability to be trained to recognize known patterns and classify data based on these patterns. Neural networks have been used with success for structural design tasks and for classification of experimental data such as sonar target classification. With proper training, neural networks should be able to process the dynamic response measurements taken from the structure, classify the data, and provide a tool for determining the location and level of damage present in a structure. This Note presents a structural damage methodology in which only active member transfer function data are used in conjunction with an artificial neural network to detect damage in structures. Specifically, the method relies on training a neural network using active member transfer function pole/zero information to classify damaged structure measurements and to predict the degree of damage in a structure. The method differs from many of the past damage detection algorithms in that no attempt is made to update a finite element model or to match measured data with new finite element analyses of the structure in a damaged state.