A Study On Combination Prediction Method Of Skin Fatigue Crack Growth Based On FRANC3D

The word "prediction ",originally from the Ming's Poems Chronicle,refers that people make scientific quantitative or qualitative speculation on the future development trend of things by using certain methods and theories according to the past development changes and laws of things.The prediction of faults is the speculation of the future fault development trend and law of mechanical equipment or parts based on the analysis of its current operating state,so as to provide a basis for the formulation of scientific equipment maintenance strategy.Fault prediction technology is one of the core contents in the field of aircraft structure health monitoring,and to study the prediction method of structural fatigue crack growth is obviously of great economic value and practical significanceThe development status of structural health monitoring technology and prediction technology are introduced in this paper,and the classification of fault prediction methods and their related applications in structural fatigue crack growth prediction are also stated.Taking the crack of the skin as the specific research object,and a finite element model of the initial crack is established by ABAQUS/FRANC3 D co-simulation,the stress intensity factor of the crack leading edge is calculated,and the partition range of the mesh parameters of the crack leading edge is determined.A random fatigue load is compiled by MATLAB to obtain simulation test data of aircraft skin crack propagation under the spectrum load,so as to provide the data for the prediction method research in this paper.On MATLAB software platform,a multivariate gray prediction model based on particle swarm optimization,and a BP neural network model with reverse error adjustment,and a support vector machine regression model were established.Each model was trained to predict the length of crack growth.The accuracy and validity of the model are verified.Furthermore,in order to make up for the shortcomings of the single prediction method,a gray neural network and a gray support vector machine combination model are presented to predict the length of crack growth,and the prediction accuracy of combination model is compared with single prediction model.Finally,to synthesise the advantages of gray neural network combination prediction method and gray support vector machine combination prediction method,the two combination prediction methods are further combined.The grey neural network support vector machine combination prediction model based on residual correction is proposed.The model further reduces the prediction error of grey neural network and grey support vector machine model,broadens the application range of combined prediction model in the field of fatigue life prediction of aircraft structure,and has certain value in engineering application.