A Hybrid Approach Based on Variational Mode Decomposition for Analyzing and Predicting Urban Travel Speed

摘要: Predicting travel speeds on urban road networks is a challenging subject due to its uncertainty stemming from demand, geometric condition, traffic signals, and other exogenous factors. This appears as nonlinearity, nonstationarity, volatility in data, it also creates spatiotemporal heterogeneity of link speed by interacting with neighbor links. In this study, we propose hybrid model using variational mode decomposition (VMD) investigate mitigate the speeds. The VMD allows data be divided into orthogonal oscillatory sub-signals, called modes. regular components are extracted low-frequency modes, irregular presenting transformed combination which more predictable than original uncertainty. For prediction, decomposes these modes predicted summed represent speed. evaluation results show that, proposed outperforms benchmark models both congested overall conditions. improvement performance increases significantly over specific link-days, generally hard predict. To explain significant variance prediction according each day, correlation analysis between properties conducted. that nondaily pattern explained through easier was predict Based results, discussions interpretation future research presented.