During the process of dam construction and compaction, understanding the permeability characteristics of dam materials is crucial for effectively controlling the construction quality of earth-rock dams, and it holds significant importance for controlling dam quality and ensuring the safe and stable operation of the dam. Traditional permeability coefficient prediction models often exhibit drawbacks such as slow convergence speed and susceptibility to local minima. To overcome these limitations, this study focused on earth-rock dam materials as the research subject. Experimental results obtained from excavation inspections conducted after the compaction construction of different zones in a water conservancy hub project in Xinjiang were systematically collected and organized, including data on dam material grading characteristics and permeability coefficients. A historical database of permeability coefficient analysis models for earth-rock dam materials was established. Additionally, to improve the accuracy of the permeability coefficient evaluation model for earth-rock dam materials, this study employed Elman neural networks. Addressing issues such as the slow convergence speed and susceptibility to local minima inherent in neural networks, an adaptive simulated annealing particle swarm optimization algorithm was introduced to optimize the Elman neural network. Finally, the applicability and effectiveness of the proposed model were validated using real engineering data.