Development of a Plug-In Hybrid Electric Vehicle Control Strategy Employing Software-In-the-Loop Techniques

In an age of growing complexity with regards to vehicle control systems, verification and validation of control algorithms is a rigorous and time consuming process. With the help of rapid control prototyping techniques, designers and developers have cost effective ways of validating controls under a quicker time frame. These techniques involve developments of plant models that replicate the systems that a control algorithm will interface with. These developments help to reduce costs associated with construction of prototypes. In standard design cycles, iterations were needed on prototypes in order to finalize systems. These iterations could result in code changes, new interfacing, and reconstruction, among other issues. The time and resources required to complete these were far beyond desired. With the help of simulated interfaces, many of these issues can be recognized prior to physical integration. Two such simulation methods are Software-In-the-Loop (SIL) and Hardware-In-the-Loop (HIL). SIL generally focuses on the development of control algorithms. HIL, on the other hand, generally helps to finalize the interfaces between the developed controls and the systems that are being controlled. This paper aims to develop a plant model that will be used in both SIL and HIL testing and to highlight differences between the two testing methods. Modeling the physical plant includes an effort to accurately replicate all signals that are included in the actual vehicle. These testing methods are used to validate a Hybrid Supervisory Controller for a Series-Parallel PHEV as part of the EcoCAR 2 competition. This paper shows how SIL testing should be used as a feasible option to validate controller algorithms. SIL and HIL testing both end up producing similar results, however, differences can be noticed. Differences in solver characteristics as well as the introduction of realistic IO in HIL tests can be the causes of some of these differences. Issues of latency can also affect the accuracy of simulation results as opposed to final in-vehicle integration. Efforts should be made when modeling systems to represent latency issues. Development of this model and testing strategy is necessary to accurately validate and verify a control strategy prior to in-vehicle integration.