Deng, J., 2020. Research on the risk early warning of construction engineering under the coupling disaster of typhoons and rainstorms in coastal areas based on BP neural network. In: Hu, C. and Cai, M. (eds.), Geo-informatics and Oceanography. Journal of Coastal Research, Special Issue No. 105, pp. 151–154. Coconut Creek (Florida), ISSN 0749-0208.
The coupling of typhoon and rainstorm is one of the main natural disasters that often occur in coastal areas at present. It often causes damage to buildings, infrastructure, trees, and even casualties. Therefore, this paper takes coastal areas as the research object to carry out the risk early warning research of construction engineering under the coupling effect of typhoon and rainstorm. Based on back propagation (BP) neural network, this paper constructed a risk early warning model for building engineering, and established the early warning index system of risk, which contains 24 influencing factors. In this paper, 13 construction projects affected by Typhoon Meranti (1614) were selected as risk assessment samples, and the training data and test data of the model were obtained through expert scoring method. BP neural network learning algorithm based on error inverse propagation was used to train the model. The trained risk warning model was used to predict the coupled risk level of typhoons and rainstorms for a construction project in coastal areas. Through simulation calculation, the simulation results were consistent with the actual risk of the project suffering from the coupled disaster of typhoons and rainstorms. The results show that the risk early warning model based on BP neural network has reference value in practical application, which can provide decision support for government departments and relevant construction units to implement disaster prevention and reduction, and take effective measures to minimize the disaster losses caused by typhoons and rainstorms coupled.