From optimization-based EM to rule-based EM

The role of the supervisory control system (SCS) of a HEV is to determine the power split between multiple sources in a hybrid powertrain. In the past years, many kinds of SCSs have been proposed, which can be generally classified into two kinds, namely, optimization-based and rule-based control strategies.

The optimization-based strategies determine the power share rules by minimizing an objective function including all concerned factors, such as dynamic programming (DP), equivalent consumption minimization strategy (ECMS), Pontryagin’s minimum principle (PMP), etc. However, due to the high computational burden and the requirement on knowledge of the driving cycles, such SCSs are impractical for real-time applications. But they serve an important role to benchmark the rule-based control strategies.

In comparison, the rule-based control strategies often based on heuristics, e.g. thermostat control strategy (TCS), power follower control strategy (PFCS), electric assist control strategy (EACS), etc, which are easily implementable in real-time applications. Therefore, among commercial HEVs, the rule-based strategies are more prevalent than the optimization-based strategies due to their simplicity and effectiveness.

A good and effective idea to design a well-performed rule-based control strategy is to observe and emulate the operations of the optimization-based control strategies. The control maps produced by the optimization-based strategies actually provides insights on how the SCS operates the powertrain at different load conditions. For instance, for series HEVs, a control map generated by the ECMS is shown as follows (top) (u=P_PS / P_PL). It shows that the optimal selection is often to operate with the secondary source (SS) at lower powers and the primary source (PS) at higher load requirements. Based on these observations, a rule-based control strategy, named, exclusive operation strategy (XOS) is developed (bottom). It can be seen that the primary source will be activated when the power demand is larger than a SOC-dependent threshold or the SOC drops below its lower bound. By implementing the XOS to a series HEV, it shows that the XOS is able to deliver similar performance as the ECMS.


More details can be found in [1]-[2].

[1]  W. Shabbir and S. A. Evangelou, “Exclusive operation strategy for the supervisory control of series hybrid electric vehicles,” IEEE Trans. Control Syst. Technol., vol. 24, no. 6, 2016, pp. 2190-2198.

[2] W. Shabbir, Control Strategies for Series Hybrid Electric Vehicles. PhD thesis, Imperial College London, 2015.