The regression prediction model for the strength of cylinder shell against pulse loading based on support vector machine

Based on experimental data of structural response ofsmall samples, support vector machine (SVM) regression method is employed tosimulate the nonlinear functional relationship among the peak value ofdynamic strain in the cylinder shell, its size and external pulse loading.Meanwhile, an improved simplex-simulated annealing hybrid algorithm isdeveloped to accomplish the optimization of SVM parameters. In addition, thecomparative analysis of forecasting capacity between SVM and backpropagation artificial neural network method is conducted. The theoreticalresults verify that SVM with optimal performance parameters has a betterforecasting capacity under the small sampling condition. Finally, themulti-variable functional relationship between the ultimate strength of thelarge-sized cylinder shell and its size against pulse loading is inferredfrom the SVM regression model. This functional relationship can be served asa referable predication model for the strength analysis of this kind ofcylinder shell devices. Theorefore, the above research shows that SVM willhave wide applications in the mechanical structure analysis, such asstrength predication and reliability analysis.