Study On Fatigue Crack Propagation Law Of 16MnR Steel And Its Pattern Recognition Based On Acoustic Emission

16Mn R steel is the most widely used material in pressure vessel.Considering the high pressure caused by inflammable and explosive or corrosive material the vessel contains and terrible working conditions with complicated load,it is likely to lead to the initiation and propagation of crack,which may results in the failure of 16 Mn R steel.The systematical research on the fatigue crack propagation(FCP)law and its state monitoring is of great significance with the increasing demand of safety and economy in pressure vessel.At present,the FCP rate curve is still measured by the test,which is time consuming and expensive.The finite element analysis(FEA)on FCP of 16 Mn R can provide reference for the accurate simulation.Meanwhile,the parameter analysis is still the main method for the FCP state recognition based on the acoustic emission(AE)signal,while less original data and limited data analyzing capability are treated as obstacles.Waveform analysis has been paid much more attention because of the ability of digging into the information of AE source deeply.However,high noise sensitivity hinders it from extracting the eigenvector precisely and building the recognition model of the FCP state.Firstly,this paper conducted the control experiments on FCP in I mode under constant amplitude loading based on the compact tension specimen of 16 Mn R steel.Meanwhile,R-ratios,thickness and notch sizes are taken as three main considering factors on investigaing the effect on the FCP rate curve.The crack section morphology and the microstructure were observed with a scanning electron microscopy(SEM).Second,Simulation on FCP consisted of both simulation on stress intensity factor(SIF)and simulation on FCP rate.The SIF was computed by displacement interpolation method,J integral and interaction integral method respectively,reaching comprehensive evaluation of the three methods by comparison.The effect of the mesh density and the angle of singular element near the crack tip on the computational efficiency and accuracy of the three methods were discussed.The FCP rate was simulated by the Jiang fatigue damage criterion and extended finite element method(XFEM)respectively.The two methods were compared and analyzed by the respective simulation results.Third,A simple acoustic emission(AE)monitoring system was developed,which was controlled by FPGA,cached in SDRAM,and transferred in high speed through USB2.0.The AD circuit was designed,FPGA sequential logic program and the software system based on USB were developed,which was verified by the Modelsim software.Meanwhile,the hardware of the AE monitoring system was debugged and the error was optimized.Finally,The AE signal acquisition in the FCP process was conducted.The best algorithm was determined to denoise the AE signal after studying the LMS adaptive filter,the empirical mode decomposition(EMD)adaptive filter and the wavelet threshold denoising algorithm.The improved wavelet singal–band reconstruction algorithm was proposed to extracte the eigenvectors of the AE signals,the support vector machine(SVM)algorithm was used to classify the different FCP state.A FCP pattern recognition software was developed to improve the computation efficiency and simplify the above process.The Paris formula in the stable FCP stage were obtained through experiments,which laid a foundation for the subsequent simulation verification.The simulation showed that the interaction integral method was appropriate for the SIF simulation,the XFEM was suitable for for the FCG rate simulation and the wavelet threshold denoising algorithm had the best effect for the AE signal.The recognition accuracy of the FCP state based on the AE signal reached to 95.3%,which can be used to forecast and classify the FCP state.The above-mentioned research provided a theoretical basis for the safety performance in many engineering structures.