Quick Look
Adaptation of Compact Electronic Control Unit Signal Conditioning for High-Speed, Spark-Ignited Engine Control, 03-9283
Printer Friendly VersionPrincipal Investigators
Jayant Sarlashkar
Joseph Grogan
Inclusive Dates: 11/05/2001 - 03/05/2002
Background - The ability to predict timing of (future) piston positions during a combustion cycle reliably is a key element of any engine control strategy. The information necessary for such prediction comes from position sensors mounted on the engine and is incremental in nature. High-speed, spark-ignited engines present a particularly challenging case for this task because of the noisy environment they provide. In particular, in laboratory-grade test cells in which high-resolution encoders are commonly used as position sensors, the radiated/conducted spark noise may be isomorphic to the signals from the position sensors and may mislead the engine position tracker.
Approach - The three main considerations in the design of an engine position predictor include: 1) selection of filter algorithm to reduce effects of noise. It was important to quantify the effect of the filter on uncertainty in projected timing; 2) choice of extrapolation scheme to determine when a particular position will be reached; and 3) realization of proposed solutions in hardware.
Accomplishments - Median filter is a classical nonlinear technique used to remove outliers. Given the digital nature of the signals under consideration, this technique was augmented with a forgetting factor to determine whether transitions in the signal (between logic 0 and logic 1) were real. The transitions were characterized by associating with them the respective duration of assertion. The outliers were identified based on a width constraint.
The resolution of the measurement scheme was shown to have a bearing on the order of polynomial used for extrapolation. For example, for a low-resolution sensor, a higher order polynomial does more poorly than a low-order one. The nonlinear filter algorithm was implemented as a state machine in digital logic, while the extrapolation algorithm was realized as a digital filter. The algorithms have since been used successfully on a number of engines, including a high-speed motorcycle.
