Hybrid Support Vector Machine Optimization Model for Prediction of Energy Consumption of Cutter Head Drives in Shield Tunneling

Abstract

The energy consumption of cutter head drives accounts for over half of their total power capacity, and it can reach several thousand kilowatts in shield machines. The analysis of the energy consumption of cutter head drives is thus essential for power planning and control in shield tunneling operations and can help determine shield performance and efficiency. The accurate prediction of energy consumption, which involves complex coupling and nonlinear parameters, has become a challenging task for site managers and tunnel engineers. A hybrid technique that combines least-squares support vector machine (LS-SVM) and particle swarm optimization (PSO) for analyzing energy consumption is proposed in this study. An adaptive Gaussian kernel function–based LS-SVM is used to establish the relationship between energy consumption and identified factors. The parameters of the LS-SVM model can be optimally determined using a nature-inspired intelligent PSO algorithm to improve prediction accuracy. This method is validated in the first Han River Crossing Urban Metro Tunnel Project in China with a complex urban environment. The relative importance of each factor in the PSO-based LS-SVM model is also compared with the results of the sensitivity analysis. Results show that the proposed method can be applied as a feasible and accurate tool for energy consumption audit in urban shield tunneling projects.