Vibration-based Prediction Of Flexural Bending Fatigue Life Of GFRP Laminated Beams

Fiber reinforced polymer(FRP),also known as fiber reinforced composite,is widely used in aerospace,civil engineering,green energy and many key fields due to its lightweight,high strength and excellent fatigue resistance.However,during application,FRP composite structures often endure bending fatigue loads,such as composite blades and bridge decks.The accumulation of fatigue damage significantly affects the bearing capacity and safety of the FRP components,while the fatigue failure of the components exhibits suddenness and destructiveness.Therefore,based on practical engineering applications,this thesis focuses on the glass fiber reinforced composite(GFRP)and uses various methods,such as experimental measurement,theoretical derivation,finite element simulation,and intelligent algorithm construction,to analyze the relationship between the single-point bending fatigue life of the GFRP laminated beam and its structural frequency.Furthermore,based on the frequency changes during the bending fatigue process,this thesis predicts the single-point bending fatigue life of the GFRP beam.The main research content and conclusions of this project are as follows:(1)Experimental research on the relationship between the structural frequency and the single-point bending fatigue life of the GFRP laminated beam.Different single-point bending fatigue and modal tests were conducted on the GFRP laminated beam under different bending loads to investigate the laws governing its failure mode,failure process,bending fatigue life,and frequency reduction during single-point bending fatigue.The single-point bending fatigue failure of the GFRP laminated beam mainly occurred at the clamping position of the specimen and manifested mainly as a brittle failure of a 45-degree sawtooth shape at the end port.The relationship between the single-point bending fatigue life and frequency mainly showed a sharp decrease followed by a gradual decrease as the bending fatigue life increases.Meanwhile,the decreasing trend of high-order frequency is more stable and gentle compared to that of loworder frequency.The single-point bending fatigue life and the discreteness of test results of the laminated beam are both decreased with the increase of bending loads.(2)Simulation analysis of single-point bending fatigue and modal model of the GFRP laminated beam.A single-point bending fatigue model was built with ABAQUS to simulate three different single-point bending fatigue under different bending loads,mainly to analyze the failure mode and fatigue life of the GFRP laminated beam under single-point bending fatigue load.The results showed that the failure position of the specimen generally occurred at the clamping end,with the failure presenting a 45-degree sawtooth shape.The simulation results were basically consistent with the experimental results of the specimen.The simulation model predicted the single-point bending fatigue life of the GFRP laminated beam under three different bending loads,with results that were well-matched with the experimental results,with a relative error within 9%.(3)Fatigue life prediction methods for evaluating the single-point bending fatigue life of the GFRP laminated beam.The intelligent learning algorithms included backpropagation neural network(BPNN)and support vector machine(SVM)were used to predict the fatigue life of the GFRP beams.The intelligent algorithm has been input various combinations of frequencies to predict single-point bending fatigue life,and the results were compared with those predicted by the bending fatigue theoretical model and the finite element model.The results showed that the average errors between the BPNN and SVM with the bending fatigue test life results were 0.35%and 0.29%,respectively.The average errors between the bending fatigue predicted by theory model and the bending fatigue test results were 0.74%,and the average errors between the finite element model predicted results and the bending fatigue test results were 1.46%.The average relative errors between different single-point bending fatigue life prediction methods and test results were all below 2%,wherein SVM algorithm models had slightly higher prediction accuracy compared with other prediction method when there is a small sample size.