Dynamic Traffic Prediction-Based Energy Management of Connected Plug-In Hybrid Electric Vehicles with Long Short-Term State of Charge Planning

Vehicle electrification, automation, and connectivity in today's transportation require significant efforts in control design to meet conflicting goals of energy efficiency, traffic safety, as well as comfort. The rapid development of intelligent transportation systems (ITS) and the rapid growth of connectivity technologies enable vehicles to receive more information about traffic conditions, which provides a reliable solution for the energy management of plug-in hybrid electric vehicles (PHEVs). This article proposes a predictive energy management strategy (EMS) for connected PHEV based on real-time dynamic traffic prediction. First, the future traffic information is predicted by establishing a wavelet neural network (WNN). Thus, the global driving condition can be predicted. Then, the particle swarm optimization (PSO) algorithm is used to optimize the parameters of WNN to plan a global battery state-of-charge (SOC) reference. Second, a long short-term memory-based velocity predictor is proposed for the predictive EMS, by planning SOC over a prediction horizon based on the global SOC reference. Finally, the performance of the proposed EMS with WNN and PSO-WNN is verified by the actual traffic data. The results show that it can improve the fuel economy by 17.57% and 28.19%, respectively.

I. Introduction

As the number of vehicles on the road increases, social problems such as traffic congestion, excessive energy consumption, traffic accidents, and high carbon emissions have emerged [1]. One of the promising technologies to improve fuel economy and reduce emissions is plug-in hybrid electric vehicles (PHEVs) [2], [3], [4]. A key challenge in developing a PHEV is the design of an energy management system (EMS) that aims to properly manage the battery SOC throughout the entire drive cycle. In the past few decades, various studies have focused on the design of EMS to maximize energy efficiency, reduce harmful emissions, and prolong battery life [5], [6]. These studies generally suggest that a near-optimal fuel economy can be achieved only if the prior future traffic condition is known [7]. With the development of internet of vehicles (IoV) technology [8], several traffic monitoring projects, such as Caltrans Performance Measurement System (PeMS) [9] and Houston TranStar [10], have been successfully implemented. These systems can provide more environmental information to the energy management system of connected PHEVs [11], [12]. In the connected environment, environment-tal information can be provided to optimize EMSs [13], which is useful in developing model predictive control (MPC) strategies. For example, based on the dynamic traffic informa-tion, the location of the next charging point is obtained to ensure that the battery SOC of the PHEV is above or equal to the lower limit when it reaches the next charging point [14]. Based on both traffic data (current traffic flow data, speed limit, sensor location IDs and positions, traffic light locations and timing, etc.) and vehicle powertrain, we can optimize vehicle route and speed simultaneously to improve fuel economy [15]. The driving speed profile can be better planned based on the dynamic traffic data (e.g., vehicle's speed, current location, and road conditions, etc.) [16].