Error-State Extended Kalman Filter-Based Sensor Fusion for Optimized Drive Train Regulation of an Autonomous PHEV

Introduction

The widespread utilization of alternative hybrid powertrains will be inevitable and many sustainable opportunities develop. The concerns about the environmental impact of fossil fuel based vehicles, economy’s foreign oil dependence, and the climate change has made countries to concentrate upon the improved propulsion technologies. Plug-in hybrid electric vehicles (PHEVs) has high potential to curtail the petroleum consumption and reduce the greenhouse gas (GHG) emissions. The scope of Plug-in Hybrid Electric Vehicles (PHEVs’), is huge. In India, where majority of vehicles are IC Engine based, the transition to Electric Vehicles is not easy even though the Government of India is promoting Electric Mobility (Niti Aayog,2018). The PHEV powertrain has an optimized combination of the conventional IC engine and an alternative traction motor. This extra degree of freedom provided by the traction motor, enables the operation of the IC engine operated in the best specific fuel consumption (sfc) range. . The regulation of the power train requires an efficient and Intelligent Energy Management System (IEMS). This controller performs the optimized power-split between the two power trains. The PHEVs in the market today, utilizes heuristic or ad-hoc power management strategies that may not be optimal under a real driving scenario. These approaches has the sparing utilization of the traction motor to prevent depletion of the traction battery. The efficient powertrain optimization requires the determination of the best power-split ratio in real time scenarios. To perform this it is critical to anticipate the vehicle speed trajectory a priori. This would not be practical without real-time data collection about the surroundings of the ego PHEV. Therefore, employing sensors to capture the proximity information would enable the optimal controller operation by anticipating the upcoming vehicle trajectories accurately and then select the appropriate power source intelligently. For example, if the ego PHEV is approaching an intersection and the sensors informs the ego PHEV that a vehicle is going to cross the road and therefore the ego PHEV has to stop, then the controller can choose to use more electrical power to drive the vehicle approaching the intersection. By doing this the engine operation would be reduced to the best sfc region, thereby less fuel would be consumed, while the traction battery can be recharged through regenerative braking for the duration the ego PHEV decelerates to a stop.

This work introduces the modelling and development of a split parallel PHEV. Section II describes the selection of the sensors and its calibration. Section III has the implementation process explained with focus on the Error State Extended Kalman Filter (ES-EKF) State Estimator modelling and implementation. The Intelligent Energy Management System and its role in regulation of the developed PHEV model is discussed in Section IV. Section V has the Results & discussions also considering the effects of sensor calibration error on the vehicle state estimation.