Real-time nonlinear model predictive control for thermal management in a plug-in hybrid electric vehicle
- Autores: Jorge López Sanz
- Directores de la Tesis: Jesús Andres Alvarez Flórez (dir. tes.), Gerhard Lux (codir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2016
- Idioma: español
- Tribunal Calificador de la Tesis: José Luis Romeral Martínez (presid.), Marco Mammetti (secret.), José María Maestre Torreblanca (voc.)
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
Several socioeconomic factors are leading governments to encourage electric powered vehicles. Currently, the bottleneck for electric vehicles mass production lies in the high \Qitage batterytechnology.One ofthe main challenges to ensure batteries safety, comfort, performance and durability requirements is thermal management, since operating at temperatures outside the range specified bythe manufacturer, they age prematurely, lead to dangerous and uncontrolled exothermic reactions and/or be incapable of delivering the electric energydemand to move the vehicle.The tendency in the solutions design for thermal management is to use cooling circuits with more and more sophisticated architectures govemed by an increasing number of electricalactuators like pumps, fans and solenoid valves.The control ofthese systems is complex dueto their nonlinear behavior, the high number ofinputs and outputs and the ne'ed of accomplishing multiple goals, usually contradictory, at the same time. In front of this class of problems, conventional control methods are taken to their limit and new optimization based methods, like model predictive control, capable of exploiting the full potential in this kind of systems, are attracting the attention of the sector.
The presentthesis deals with the design ofa predictive control forthe batteryand powerelectronics cooling circuitin a Plug ln Hybrid Electric Vehicle.The main merit of the proposed solution is that the method validation takes places in a prototype on real-time. which, as it will be seen in the state of the art, is one of the usuallacks in most model predictive control publications in the automotive sector. For reaching this settlement, the development of a suitable model of the system and optimization problem definition togetherwith the use of an efficient and robust numerical tool, have been essential and therefore will be addressed exhaustively in this document. Additionally, the validation by means of simulation as well as the design of repeatable driving conditions for comparing the proposed control with the original one in the vehicle will be shown befare reaching the final validation and discussion.
