Research On Structure Reliability Calculation Method Based On Respone Surface Method

The reliability of mechanical products or systems refers to the possibility of performing specified functions without failure within a certain period of time and under certain conditions.It involves the knowledge of probability theory,mathematical statistics,fuzzy mathematics,etc.Reliability analysis is widely used in aerospace,bridge,construction,construction machinery and other fields.With the development of basic reliability theory,many effective reliability analysis methods have been derived.The response surface method is a simple and efficient reliability analysis method proposed in solving nonlinear problems,especially in solving implicit limit state function problems.Based on a series of reasonable sample points,the response surface method uses an effective iterative strategy and approximates the real implicit limit state function with a polynomial function to ensure that the reliability converges to the real probability value to the greatest extent.With the continuous and in-depth research on reliability at home and abroad,the accuracy and computational efficiency of reliability have been continuously improved.At the same time,response surface method has also played an increasingly important role in fuzzy reliability analysis and optimization design.The characteristics of the weighted linear response surface method are analyzed,and a more reasonable weight form is constructed.At the same time,the uniform test design method is applied to the weighted linear response surface method,and an improved weighted linear response surface method is proposed.The feasibility of the improved linear response surface method is verified by an example.The advantage of partial least square(PLS)method in multivariate regression is analyzed.The quadratic polynomial response surface method is linearized by using the quasi-linear model to solve the structural failure probability.A partial least squares nonlinear response surface method based on Gaussian kernel function is proposed.The effectiveness of the proposed nonlinear response surface method is verified by an example.The kernel partial least square response surface method(RSM)is applied to the fuzzy reliability problem of structures.The force and stress distribution of cylindrical helical spring under load are analyzed,and the multi-objective optimization design of fuzzy reliability is carried out.