Abstract: In order to prevent safety accidents of construction elevator, database and SVM algorithm were used to predict its safety risks. Firstly, based on relevant theory and characteristics of construction elevators, preliminary theoretical qualitative analysis on their risk factors was carried out. Then, safety accident cases were statistically analyzed by utilizing accident database management system, risk factors were defined and risk prediction indicators were determined. Finally, SVM algorithm was applied to construct a safety risk prediction model, and its parameters were optimized by grid search method, genetic algorithm and particle swarm optimization algorithm respectively to determine the best prediction model. The results show that safety accident database can be used to establish a risk prediction index system, and risk level can be divided by constructing an SVM prediction model, which is helpful for us to take corresponding prevention and control measures to reduce risk of construction elevators, ensure personnel safety, and reduce property losses.

Key words: database, support vector machine (SVM), construction elevator, safety risk factor, risk prediction