The prediction of sustainable development for coastal city cultural scenes emphasizes the potential to preserve and enhance traditional cultural heritage. In the present society, the prediction of the cultural value of coastal cities has a great error rate and lack of accuracy. LSTM algorithm in neural network is a technology that can effectively predict the sustainable development of cultural value. Compared with previous detection algorithms, BP algorithm has higher accuracy. In this paper, LSTM algorithm is used to predict the cultural value of coastal cities, which greatly improves the accuracy. Finally, through the experiment, the accuracy of the prediction system established by LSTM algorithm is very high, and the average can reach 99.6%.