Artificial neural networks, as a powerful tool capable of handling complex nonlinear relationships, offer higher predictive accuracy and flexibility. To improve the efficiency of cost estimation in shipyards and address the issues of lengthy processing time and high human resource consumption associated with traditional methods, artificial neural networks are employed for cost estimation. First, by performing a cost breakdown for shipbuilding projects, the shipbuilding cost items are determined. At the same time, key factors influencing cost estimation are identified. These factors are used as input variables, while the cost items are treated as output variables, to establish a BP neural network model. The transfer function, the number of hidden layer nodes, and the method for determining the thresholds are explained. To avoid getting trapped in local optima, a GA-BP neural network is further selected for optimization, and the optimization principle is described. Original data from a shipbuilding project at a certain shipyard is collected, and normalization and network training are carried out using MATLAB software, resulting in a stable neural network. Error analysis is conducted on the prediction results of the BP and GA-BP neural networks. The BP neural network achieves an accuracy rate of 97%, while the GA-BP neural network achieves 99%, providing new insights for current shipbuilding project cost estimation.