Research on seismic performance optimization of cable-stayed bridge dampers based on machine learning

Abstract The dampers of cable-stayed bridges play a crucial role in bridge seismic resistance. Traditional research requires a large number of trial calculations of damper parameters, and the application of machine learning methods to optimize the seismic performance of dampers in cable-stayed bridges has a great significance. This article is based on the parameter analysis data of dampers for a single tower cable-stayed bridge. Firstly, the advantages and disadvantages of central composite design and comprehensive experimental method were compared and analyzed. Then, the response surface fitting method was optimized using support vector egression. Finally, the optimal damper parameters were studied using particle swarm optimization algorithm. Analysis shows that there is significant nonlinearity in the structural response under earthquake action. The use of support vector machines and particle swarm optimization algorithms can accurately and efficiently fit and optimize damper parameters. From this, it can be concluded that the machine learning method combining support vector machine and particle swarm optimization has good accuracy and applicability in optimizing the seismic performance of cable-stayed bridge dampers, and can be further extended to other research fields.