Comparison of Model Predictive Control and Distance Constrained-Adaptive Concurrent Dynamic Programming Algorithms for Extended Range Electric Vehicle Optimal Energy Management

Intelligent energy management of hybrid electric vehicles is feasible with a priori information of route and driving conditions. Model predictive control (MPC) with finite horizon road grade preview has been proposed as a viable predictive energy management approach. We propose that our novel distance constrained-adaptive concurrent dynamic programming (DC-ACDP) approach can provide better energy management than MPC without any road grade information in context of an extended range electric vehicle (EREV). In this article, we have evaluated and compared the MPC and DC-ACDP energy management strategies for a real-world driving scenario. The simulations were conducted for a 160 km drive with road grade variation between +4% and -1%. Results show that the DC-ACDP approach is near-optimal and improves overall energy consumption by a maximum of 4.25%, in comparison to the simple MPC with a finite horizon road grade preview implementation. Additionally, a higher value for energy storage system state of charge (SOC) tracking penalty p(2) results in the net energy consumption for MPC to converge toward that of DC-ACDP. A combination of the MPC and DC-ACDP approach is also evaluated with only 1.25% maximum improvement over simple MPC.

Automated summary

Intelligent energy management for hybrid electric vehicles with a priori route and driving condition information can be achieved through the novel DC-ACDP approach, which outperforms the traditional MPC approach with finite horizon road grade preview, especially for extended range electric vehicles (EREV).