Damage Identification And Assessment Research In Stiffened Plates Based On Ultrasonic Guided Waves And Deep Learning

The stiffened plate is a kind of structure which is bonded or riveted by stiffened ribs and the bottom plate.Because of its high strength and light weight,it is widely used in high-end equipment,such as the skins of aircraft wings and fuselages.Structural Health Monitoring is an effective method to ensure the safety and reliability of these equipment.The Ultrasonic Guided Waves method is a common method for Non-destructive Testing of plate-like structures.However,the current Ultrasonic Guided Wave damage detection method for stiffened plates can not accurately locate and judge the damage degree.The Deep Learning algorithm based on Convolutional Neural Network is an algorithm that can realize end-to-end complex nonlinear mapping.It has the advantages of simpler structure,deeper layers,and faster calculation speed.Instead of manual selection,it can automatically extract damage features.Based on the propagation characteristics of Ultrasonic Guided Waves in the plate structure and the sensitivity of guided waves to the damage in the structure,this paper,aiming at the common debonding damage in the stiffened plate,combined with the Deep Learning method of Convolutional Neural Network,uses a large number of detection data to identify and evaluate the damage,so as to improve the accuracy and efficiency of damage detection.The main contents of the research are as follows:Firstly,the excitation and propagation characteristics of Ultrasonic Guided Waves in typical T-stiffened plates are numerically simulated.Due to the sensitivity of Guided Wave to surface and internal damage,the effects of stiffeners and debonding damage on guided wave propagation are studied.Secondly,the experimental data and simulation data are compared and analyzed through waveforms.The two have a certain similarity.On this basis,the experimental and simulation data are merged to establish a database for training and testing deep learning algorithms.Finally,based on the ultrasonic guided wave damage database,the deep learning damage algorithm of Convolutional Neural Network constructed in this paper is trained,optimized and tested.The damage features captured by convolution kernel are automatically updated,then highlighted through the pooling layer,and intuitively given through the classification layer to realize damage identification and assessment.Results show that the combination of Guided Waves and Deep Learning can effectively identify the debonding damage in the T-stiffened plate,including the location of the debonding damage and the evaluation of the debonding damage degree.The deep learning algorithm of the Convolutional Neural Network constructed in this paper shows great performance in the identification and evaluation of the debonding damage in the typical T-stiffened plate.The Ultrasonic Guided Wave detection method based on Deep Learning proposed in this paper can effectively detect and evaluate the debonding damage in typical T-shaped stiffened plate.It provides theoretical basis and new methods for Structural Health Monitoring,which has practical engineering application value.