The creep compliance function is usually expressed by a Prony series model when using finite element codes to calculate the creep effects of concrete structures. To address the physically meaningless phenomenon of negative, oscillatory and unstable parameters when using traditional single methods (such as the conjugate gradient method and the LM algorithm) to fit Prony series models to experimental creep function data, this paper proposes a novel hybrid intelligent optimization method for Prony series parameter identification. The method establishes an objective optimization function with penalty terms and combines a simulated annealing-genetic hybrid intelligent algorithm with nonlinear programming methods to constrain parameter identification within physically meaningful ranges. Subsequently, a simple and practical calculation formulation for the Prony series was proposed by parameter analysis. It was shown by numerical examples that the proposed method not only effectively eliminates the drawbacks of conventional approaches, but also gives the identification with relative errors below3%. The Prony expression of creep function can be directly applied to the development of finite element software program in calculating the creep effect of concrete structures.