A novel multi-objective optimization method based on enhanced hippopotamus optimization algorithm and Kriging model

Purpose Aiming at the shortcoming of the Kriging model with low efficiency of fitting accuracy, a new multi-objective optimization method for complex structures based on improved Kriging model is proposed. Design/methodology/approach Firstly, the method introduces an enhanced hippopotamus optimization algorithm (EHO) to address inherent limitations within the hippopotamus optimization algorithm (HO). Subsequently, EHO is utilized to search for the optimal correlation coefficients of Kriging. Then, EHO-Kriging is combined with NSGA-III to establish a multi-objective optimization framework with high accuracy and high solution efficiency. Findings The EHO-Kriging model exhibits high fitting accuracy on test function, with coefficients of determination reaching above 0.99, and mean relative error, mean absolute error, mean absolute percentage error and mean squared error are all close to 0. The proposed optimization approach is implemented to a flat car underframe and an EMU bogie frame, demonstrating that this scheme reduces the maximum equivalent stress and mass and exhibits higher precision compared to traditional optimization. Originality/value Utilizing suitable intelligent algorithms to obtain the optimal correlation coefficients for the Kriging model can greatly improve fitting accuracy. The optimization method based on the combination of high-precision surrogate model and multi-objective optimization algorithm can effectively reduce the product performance fluctuation and achieve high solution accuracy, which has strong feasibility and engineering applicability.