Gradient Heat Treatment On D406A Steel Weld Specimen And Predictiong Of Mechanical Property

In this paper the organizational structures and mechanical properties of originalstate, welded state of D406A steel and weld specimen after traditional heat treatmentwere studied. The impact of structure and mechanical property of weld specimen afterthe gradient heat treatment was comparatively analyzed, which consisted of circlequenching heat treatment under the condition of space gradient and time gradient heattreatment. The artificial neural network model that forecasted the tensile property ofweld specimen was constructed on the base of gradient heat treatment processparameters and results of tensile property test.The microstructures of weld specimen in different states and tensile fracture wereobserved by optical microscope and scanning electron microscope. Phase in weldspecimen under different heat treatments was identified by XRD. The mechanicalproperties were carried out on CMT5305electron universal mechanics testing machineand HV-5Vickers hardness tester. The creating, training and forecasting of artificialneural network were all done by MATLAB software.The results showed that the original state organization of D406A steel was granularpearlite. Cyclic quenching heat treatment under the condition of space gradient couldeffectively refine the organization of weld zone. The weld zone was martensite whichwas coarse and uneven. The base zone was granular pearlite which was uniform. Theheat affected zone was complex. The martensite exited in the weld specimen under bothtraditional heat treatment and the gradient heat treatment. But austenite was observed inthe weld specimen under the gradient heat treatment through XRD test.Tensile strength of weld specimen under traditional heat treatment could reach1730MPa while its elongation was only9.3%. The highest elongation of weld specimenunder the gradient heat treatment was13.67%, whose tensile strength dropped to1546MPa. Many dimples existed in tensile fractures under different heat treatments. Themechanisms of fracture under different heat treatments were all microsporecongregation fracture. Especially, the dimples were deeper and more uniform in tensilefractures under the gradient heat treatment. The Vickers hardness of base zone waslower than that of the weld zone or the heat affected zone under welding state.Compared with the traditional heat treatment, Vickers hardness of weld specimen under the gradient heat treatment was lower. Under different heat treatment the highesthardness value zone was still the base zone while the lowest was the weld zone.Based on the tensile mechanical properties data under the gradient heat treatment,the BP neural network was established. The training results of the BP neural networksabout tensile strength and elongation showed excellent, whose correlations could reach0.999and0.999.Relationship between parameters of the gradient heat treatment whichcame from training data sources and mechanical properties of weld specimen werereceived through network. Predicted values of tensile strength and elongation whichwere obtained by network fluctuated round the test values. The accuracy of theprediction about tensile strength was higher than that of the prediction about elongation.