Reliability analysis of mechanical structures based on truncated normal distribution

Abstract Parameters such as the geometric dimensions and strength limits of a structure, if described using a normal distribution, can theoretically take values ranging from negative to infinite, which clearly does not align with actual engineering conditions. Geometric dimensions and material strength limits cannot be negative, and they also have an upper bound, meaning they cannot be infinitely large. The truncated normal distribution, as a distribution form that limits the range of variable values, provides a more accurate and realistic description for random variables with boundary constraints in engineering applications. Therefore, this paper combines the truncated normal distribution with the Saddlepoint approximation (SPA) method to construct a reliability analysis model that integrates the truncated normal distribution and the SPA method.