Real-time self-learning optimization of diesel engine calibration

Compression ignition engine technologies have been advanced in the past decade to provide superior fuel economy and high performance. These technologies offer increased opportunities for optimizing engine calibration. Current engine calibration methods rely on deriving static tabular relationships between a set of steady-state operating points and the corresponding values of the controllable variables. While the engine is running, these values are being interpolated for each engine operating point to coordinate optimal performance criteria, e.g., fuel economy, emissions, and acceleration. These methods, however, are not efficient in capturing transient engine operation designated by common driving habits, e.g., stop-and-go driving, rapid acceleration, and braking. An alternative approach was developed recently, which makes the engine an autonomous intelligent system, namely, one capable of learning its optimal calibration for both steady-state and transient operating points in real time. Through this approach, while the engine is running the vehicle, it progressively perceives the driver's driving style and eventually learns to operate in a manner that optimizes specified performance criteria. The major challenge to this approach is problem dimensionality when more than one controllable variable is considered. In this paper, we address this problem by proposing a decentralized learning control scheme. The scheme is evaluated through simulation of a diesel engine model which learns the values of injection timing and variable geometry turbocharging vane position that optimize fuel economy and pollutant emissions over a segment of the FTP-75 driving cycle. [DOI: 10.1115/1.3019331]