To improve the prediction accuracy in foundation pit deformation, an improved supply-demand-based optimization(ISDO)algorithm-exponential power product(EPP)foundation pit deformation prediction model was proposed. The optimization ability of the ISDO algorithm was verified through six standard test functions and three application examples, and the results were compared with the basic supply-demand optimization(SDO)algorithm, whale optimization algorithm(WOA), gray wolf optimization(GWO)algorithm, moth swarm algorithm(MSA), and particle swarm optimization(PSO)algorithm. Taking the three foundation pit settlement predictions as examples, the autocorrelation function method and the false nearest neighbor were used to determine the delay time and embedding dimension of each instance and, construct input and output vectors to train and predict each model. The results show that the search capability of the ISDO algorithm is superior to the five other algorithms such as SDO, and it has higher search accuracy, global search capability and robust performance. The absolute value of the average relative errors predicted by the ISDO-EPP model for the three examples were 0. 73%, 3. 36% and 1. 33%, respectively, which were smaller than the ISDO-SVM and ISDO-BP model, indicating that the ISDO algorithm can effectively optimize the parameters of the EPP model and the ISDO-EPP model is feasible and effective for deformation prediction.