An Improved Hybrid Highway Traffic Flow Prediction Model Based on Machine Learning

Abstract

For intelligent transportation systems (ITSs), reliable and accurate real-time traffic flow prediction is an important step and a necessary prerequisite for alleviating traffic congestion and improving highway operation efficiency. In this paper, we propose an improved hybrid predicting model including two steps: decomposition and prediction to predict highway traffic flow. First, we adopted the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) method to adaptively decompose the original nonlinear, nonstationary, and complex highway traffic flow data. Then, we used the improved weighted permutation entropy (IWPE) to obtain new reconstructed components. In the prediction step, we used the gray wolf optimizer (GWO) algorithm to optimize the least-squares support vector machine (LSSVM) prediction model established for each reconstruction component and integrate the prediction results of each subsequence to obtain the final prediction result. We experimentally validated the effectiveness of the proposed approach. The research results reveal that the proposed model is useful for predicting traffic flow and its changing trends and also allowing transportation officials to make more effective traffic decisions.

Survey of Decomposition-Reconstruction-Based Hybrid Approaches for Short-Term Traffic State Forecasting

“…First, the datasets used by various applications are different. Generally, they are aggregated at 15 min granularity for short-term traffic flow, speed, and travel time forecasting of highways or city roads [ 17 , 19 , 24 , 27 , 28 , 37 , 38 , 40 , 41 , 42 , 43 , 44 , 70 , 71 , 81 , 82 ]. In applications of bus speed or travel time prediction, they are aggregated at 5–30 min granularity [ 47 , 48 , 55 , 62 , 83 , 101 , 108 ].…”

“…In view of the modal aliasing problem, several extended versions of EMD have been proposed, including ensemble empirical mode decomposition (EEMD) [ 68 , 69 ], complete ensemble empirical mode decomposition (CEEMD), and complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) [ 70 , 71 ]. These extended versions alter the distribution of the extreme points (places where a function takes on an extreme value) by adding noise to the original signal.…”

“…Additionally, reports of short-term traffic forecasting based on CEEMDAN can also be found in [ 70 , 71 , 76 ]. Tian et al [ 70 ] slightly improved the CEEMDAN algorithm and utilized permutation entropy (PE) to analyze the random properties of the decomposed IMFs.…”

“…The kernel online sequential extreme learning machine (KOSELM) and ARIMA were then selected according to the different random properties of the IMFs for short-term traffic flow prediction at signalized intersections. Wang et al [ 71 ] adopted the improved weighted permutation entropy (IWPE) to obtain new reconstructed components after decomposing the raw nonlinear and nonstationary highway traffic flow data using CEEMDAN. In the prediction stage, a grey wolf optimizer (GWO) is used to calibrate the least-squares support vector machine (LSSVM) established for each reconstruction component.…”

A New Ensemble Reinforcement Learning Recursive Network for Traffic Volume Forecasting in a Freeway Network

“…The GSA is applied to optimize the parameters of SVR and combined forecasting results to get the final result [19]. Wang et al chose the gray wolf optimizer (GWO) to improve the parameters and combine prediction subseries from the predictor for further increasing the forecasting accuracy [20]. Besides the abovementioned heuristic optimization methods, the reinforcement learning methods are also popular by scholars recently.…”