Abstract
In order to effectively solve the problems of low prediction accuracy and calculation efficiency of existing methods for estimating economic loss in a subway station engineering project due to rainstorm flooding, a new intelligent prediction model is developed using the sparrow search algorithm (SSA), the least-squares support vector machine (LSSVM) and the mean impact value (MIV) method. First, in this study, 11 input variables are determined from the disaster loss rate and asset value, and a complete method is provided for acquiring and processing data of all variables. Then, the SSA method, with strong optimization ability, fast convergence and few parameters, is used to optimize the kernel function and the penalty factor parameters of the LSSVM. Finally, the MIV is used to identify the important input variables, so as to reduce the predicted input variables and achieve higher calculation accuracy. In addition, 45 station projects in China were selected for empirical analysis. The empirical results revealed that the linear correlation between the 11 input variables and output variables was weak, which demonstrated the necessity of adopting nonlinear analysis methods such as the LSSVM. Compared with other forecasting methods, such as the multiple regression analysis, the backpropagation neural network (BPNN), the BPNN optimized by the particle swarm optimization, the BPNN optimized by the SSA, the LSSVM, the LSSVM optimized by the genetic algorithm, the PSO-LSSVM and the LSSVM optimized by the Grey Wolf Optimizer, the model proposed in this paper had higher accuracy and stability and was effectively used for forecasting economic loss in subway station engineering projects due to rainstorms.
Highlights
Published: 19 June 2021In recent years, extreme weather events have frequently occurred in the world, and natural disasters, such as waterlogging, have posed a great threat to public safety [1]
Considering many factors that affect lithology prediction, Gu (2020) used the mean impact value (MIV) to screen indicators to reduce the number of input variables of the particle swarm optimization (PSO)-backpropagation neural network (BPNN), and the experiments showed that this method achieved better computational performance than other methods [31]
least-squares support vector machine (LSSVM) is used to build the prediction model, sparrow search algorithm (SSA) is used to find the optimal parameters of LSSVM and MIV is used to find the optimal input variable system of the prediction model
Summary
Extreme weather events have frequently occurred in the world, and natural disasters, such as waterlogging, have posed a great threat to public safety [1]. There are many influencing factors that affect the disaster losses caused by storm waterlogging in the subway station projects, and there is a nonlinear mapping relationship between these influencing factors and disaster losses. The valuation methods based on bill of quantities and index estimation methods are commonly used to calculate disaster losses in an engineering project These two methods only consider the influence of a small number of influencing factors on disaster losses, and they have some shortcomings, such as low calculation accuracy and efficiency [4]. In order to effectively solve the problems of low prediction accuracy and calculation efficiency of existing methods for estimating economic loss in a subway station engineering project due to rainstorm flooding, a new intelligent prediction model is developed using the SSA, the LSSVM and the MIV method.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
