SPECIAL BORDER DOMAIN SAMPLING METHOD FOR STRUCTURAL RELIABILITY ANALYSIS

The border domain sampling method, based on the Monte Carlo method, is proposed to calculate structural reliability indices. The basic principle of this method is that the major random sampling point near the most joint probability density function of the random variables (PDF) is in a safe domain. Only little random sampling point which is far away the PDF may be in a failure domain. This domain far away the PDF is a specially specified border domain. The calculative number of performance functions and indicator functions which use little random sampling point in the special border domain is cut down. This method can be applied to analyzing the structural reliability by finite element method for the implicit function. It requires fewer finite element calculations with this method, achieving satisfactory accuracy. Meanwhile, the Matlab emulation program is complied to verify this method. Through an example, the efficiency of sampling and the precision of simulating calculation of this method are verified.