Research On Prediction Method And System Of Mechanical Properties Of Materials

The constitutive properties of metal materials will change according to factors such as aging,service load,embrittlement,and irradiation.In engineering,the in-service mechanical properties of structural materials need to be tested regularly in order to evaluate their loading capacity and safety reliability,so as to ensure the safe operation of the equipment.Commonly used non-contact inspection methods can only find defects on the macroscopic surface of the equipment,but cannot quantitatively characterize the current mechanical properties of the equipment materials.Conventional sampling test methods require sampling,which will cause certain damage to the equipment and cannot be detected on site.Because the double-hole method has the characteristics of simple test method and suitable for in-situ detection,it is expected to use this method to make up for the deficiencies of other methods,and has a very broad application prospect.However,it was found through the test that the thickness of the material between the two holes may have an impact on the test curve,so more deeply study is needed to further improve the accuracy of the double hole method to detect the mechanical property of materials.And a set of mechanical properties prediction methods and systems based on double-hole micro-shearing that can be detected in situ and can be applied are explored,which lays the foundation for the popularization and use of double-hole method.Based on the above,the test system was rebuilt in this thesis.Combined with the principle of the double-hole method,the thickness between holes was taken as the influencing factor of the prediction,and the prediction model of mechanical property parameters based on neural network was established to achieve accurate and convenient measurement of mechanical properties.It mainly introduces the construction of the new double-hole device and the construction of the measurement and control system based on LabVIEW.In order to study whether the thickness between holes has an impact on the test curve,multiple groups of single variable tests for three materials,12Cr13,S31608 and 16 Mn,were designed to collect the load and displacement data during the test process.And the corresponding exploration was made on the measurement method of the thickness between holes.With the help of uniaxial tensile method,and through the preprocessing and analysis of the test data,the normalized parameters related to the the thickness between the holes were introduced to predict the yield strength and tensile strength.The correlation between the parameters is given,and the strength can be predicted by the characteristic value of the double-hole test curve and the thickness between the holes.With the help of finite element modeling analysis,the neural network was innovatively introduced into the double-hole method,and the prediction model of mechanical properties parameters of BP and PSO-BP with 46 inputs and 3 outputs was established to achieve more accurate measurement of mechanical properties.The test results show that the slight change of thickness between the holes has a significant measurable effect on the double-hole test curve.And the impact of thickness between the holes on the accuracy can be effectively reduced by using the new test platform and the normalized parameters to predict the strength.Compared with the traditional BP model,the PSO-BP model has a higher prediction accuracy for the mechanical property parameters,and the average relative error of each parameter prediction is less than 1.1%,which lays a foundation for the promotion and application of the double-hole method.