A model-based approach to embedded systems design using heuristic methods: Powertrain applications

The need for an organized approach to embedded control system design and implementation is emerging in several industries. Companies are being tasked to reduce energy consumption and pollution, while improving performance and remaining at a competitive cost to the customer. This dissertation describes a new process, and custom tools necessary to design and implement an efficient embedded controller. It also presents a heuristic approach to supervising a complete powertrain system from the driver command to the power at the wheels. The supervisory system includes intelligent driver characterization along with optimization search methods used to select multiple set points simultaneously. This dissertation also determines an optimal control scheme for a powertrain system by following a unique design process and developing a supervisory system that replaces traditional transmission shift schedules with a dynamic gear scheduler. In conclusion, a new procedure is shown to test model-based designs throughout the development process.