In order to achieve accurate prediction of the ultimate elevation bearing capacity of the composite foundation of transmission towers and overcome the problems of large error and slow calculation of the theoretical or traditional empirical formulas,an improved pelican intelligent algorithm(IPOA)is proposed to optimize the bearing capacity prediction model of the BP neural network.Firstly,the pelican optimization algorithm(POA)is optimized using SPM chaotic mapping,Levy flight,and a positive cosine optimization strategy that incorporates nonlinear inertial weight factor o.Then,the optimized IPOA is used to find the optimization of the weight and threshold parameters of the BP neural network,and the IPOA-BP prediction model is obtained;finally,a dataset is constructed based on validated simulation experiments and the IPOA-BP prediction model is trained and tested.The results show that compare with the POA-BP prediction model,the square root error of IPOA-BP decreases by 65.75%,the absolute average error decreases by 65.79%,and the average relative error decreases by 65.60%,it can be seen that IPOA-BP neural network can achieve a more accurate prediction of the composite foundation's resistance to elevation bearing capacity,which provides a new method for the prediction of the bearing capacity of this type of foundation.