An approach to reliability analysis using the response surface method and monte carlo simulation

摘要

It is improtant in reliability evaluation to take an approach in which the required calculations can be performed efficiently in terms of time and cost. In this study, an approach is proposed whereby reliability analysis is carried out by means of Monte Carlo simulation in which the actual performance function is replaced by a function obtained using the response surface method (RSM). The common approach in the conventional RSM is to use a second-degree polynomial for the response surface function, but in many reliability problems this may not be the best choice. This paper first reviews the approaches and limitations of reliability methods, and then goes on to discuss a method of modelling error when using the response surface method for reliability analysis. It shows the errors obtained for different response functions under different circumstances, and then describes the application of a network-based analysis system to reliability problems.