Prediction On The Pitting Corrosion Resistance Of Duplex Stainless Steel And Its Welding Process Optimization Via SVR

As the social development, the natural mineral resources become less and less, and it is more and more difficult to be exploited. And human beings' requirements on materials' performance become higher and higher. As a kind of commonly and widely employed material, stainless steel(SS) are extensively used in various fields, because of their excellent stainless resistance, corrosion resistance, oxidation resistance, low temperature performance, superplasticity, bactericidal performance and memory function and so on. Although many standard stainless steels had been developed, they were designed based on an empirical formula, and the formula can't grasp how to obtain the highest corrosion resistance of stainless steel very well. Hence, it's important to establish a stable and accurate model for predicting corrosion resistance of the optimal material composition ratio. This thesis aims at the modeling and optimization on the corrosion resistance of non-standard duplex stainless steel, weaker phase pitting potential of non-standard duplex stainless steel, and stainless steel welding process optimization. The research process is as follows:(1)collect related experimental data and modeling,(2)Test the accuracy of the established model,(3)Analysis the related rules revealed by the constructed model.In this thesis, the particle swarm optimization(PSO) method combined with support vector regression theory, were utilized to establish the models. The main work includes the following contents:(1)Modeling on the pitting potential of non-standard low-Ni-high Mn-N duplex stainless steels.In this study, motivated by the researcher's previous work, they regarded the chromium(Cr), molybdenum(Mo), nitrogen(N) elements as the input parameters, and obtained the pitting resistance equivalent number(PREN) as an intermediate index to indicate the pitting potential value(Ep). In this work, we get the optimal intermediate variable(PREN1.47) according to the PREN definition via PSO and also establish directly the relationship between the composition of non-standard low-Ni-high Mn-N duplex stainless steels and its potential value(Ep) with 3D graphics. We evaluated the SVR model's accuracy with test samples. The results show that the predicted values by the established SVR model coincide well with the experimental results. The mean absolute percentage error(MAPE=0.91%) predicted via the SVR model is far better than those of three other nonlinear regression models(PREN16 = 13.70%, PREN30 = 14.68%, PREN1.47 = 12.45%). At the same time, we conducted sensitivity analysis about the sensitive of the pitting potential on the component elements via the constructed SVR model. And it reveals that chromium element has the most influence on the pitting corrosion resistant of the non-standard low-Ni-high Mn-N duplex stainless steels, the others shows same less element sensitivity(2)Modeling on the weak phase pitting potential of the non-standard low-Ni-high Mn-N duplex stainless steelIn the study, the chromium(Cr), molybdenum(Mo), nitrogen(N), nickel(Ni) elements were considered as the independent parameters, and then obtain the pitting resistance equivalent number as an intermediate variable to represent the point potential value(Ep). What the only difference with our previous work is that we take the smaller PREN index from PREN? and PREN?(weaker phase) as an intermediate indicator for Ep. We further establish a SVR model to disclose the relationship between the component and the pitting potential for the non-standard low-Ni-high Mn-N duplex stainless steel, and then use test samples and independent samples to validate the established SVR model. The results demonstrate that the MAPE of the training, test and independent samples(0.0002%, 2.50%, 32.38%) are far superior to those(12.35%, 9.85%, 65.18%) obtained by the PREN models. Sensitivity analysis results via the SVR model illustrate that the element with the greatest sensitivity to the pitting potential is Cr, followed by N, Ni and Mo elements.(3)Influence of plating metal welding factors on the pitting potential of duplex stainless steelIt was found that not only the corrosion resistant properties of duplex stainless steel associate with the chemical composition and the phase distribution, but also the welding process factors of stainless steel do in the welding spot. In this chapter, we adopt the welding current(I), welding speed(S), the distance in contact with the workpiece(N) and the welding angle(T) as independent variable to forecast the pitting potential in the welding spot via SVR. The results discovered that the calculated results by the established SVR model are almost consistent with the experiments. However, the percentage errors for the test samples reported in the literature all are above 0.5%. The MAPEs for the training and test samples(0.32%, 0.42%) are far better than those(14.49%, 6.36%) reported in the literature. At the same time, from the constructed SVR model, it is found that the welding speed plays an important role among all of the welding process factors.In conclusion, the established SVR models are both robust and accurate, and they can quantitatively reveal the effects of the independent parameters on the dependent variable, and also can be accurately and effectively used to establish the highly complex nonlinear relationship between the pitting corrosion resistance of non-duplex stainless steel and their chemical composition. It would provide an important theoretical guiding significance and has practical application value to save the experimental time, and reduce the experimental cost and test times.