Design of building construction safety prediction model based on optimized BP neural network algorithm

• Published in: Soft Computing Vol. 24; no. 11; pp. 7839 - 7850
• Main Authors: Shen, Tao, Nagai, Yukari, Gao, Chan
• Format: Journal Article
• Language: English; Japanese
• Published: Berlin/Heidelberg Springer Science and Business Media LLC 01.06.2020; Springer Berlin Heidelberg; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Back propagation networks; Casualties; Computational Intelligence; Construction accidents & safety; Construction industry; Control; Emergency communications systems; Empirical analysis; Engineering; Focus; GNP; Gross National Product; Industrialized nations; Macroeconomics; Management of crises; Management theory; Mathematical Logic and Foundations; Mechatronics; Neural networks; Occupational safety; Prediction models; Robotics; Safety management; Security management; United Kingdom > UK; United States > US; China; Japan; BP network; Genetic algorithm; Safety prediction; Risk factors
• ISSN: 1432-7643; 1433-7479
• DOI: 10.1007/s00500-019-03917-4

In order to solve the safety problem of the construction industry, the construction safety prediction model based on the optimized BP neural network algorithm is designed in this study. First, the characteristics of the construction industry were analyzed. As a labor-intensive industry, the construction industry is characterized by numerous factors such as large investment, long construction period and complicated construction environment. Due to the increasingly serious security problem, widespread concern over such problem has been aroused in society. Second, the problem of building construction safety management was summarized, six influencing factors were explored and a building construction safety prediction model based on rough set-genetic-BP neural network was established. Finally, the model was validated by a combination of multiparty consultation, empirical analysis and model comparison. The results showed that the model accurately predicted the risk factors during the construction process and effectively reduced casualties. Therefore, the model is feasible, effective and accurate.